|
12/24 |
2005/8/7-11 [Science/Space, Computer/HW/Memory] UID:39034 Activity:nil |
8/7 The Space Shuttle still uses floppy disks. Would somebody please tell them how unreliable floppies are. Buy them USB flash drives! http://www.nasa.gov/images/content/124685main_s114e7139_hires.jpg \- high density electronics and media may be more suspectible to instability due to thermal noise or radiation. lbl has/had one of the faciltiies for "space certifications" for electronics. \_ Another question is: why is anything on removable media when everything happens within the shuttle and there is nowhere else to go? \_ networks can go down, they have multiple points of failure - what do you do then? \_ I see. \_ NASA is intentionally slow to adopt new technology for use on the orbiters. In addition to testing the living hell out of particular systems themselves, they also want the general technology to have seen widespread use enough to have flushed out any problems. Note that other thread in the motd at the moment about motherboards with bad capacitors, and consider that it's probably a good thing that the orbiters' avionics are ugraded less frequently than the average sodan's computer. Remember that the shuttle program started in the seventies, and that floppies actually represent an *upgrade*. For the poster who asked why anything is on removable media, what else would you suggest? Hard drives don't perform well under a lot of acceleration, and can be damaged easily by vibration (such as during launch and reentry). Many nonvolatile memory technologies have come and gone over the life of the shuttle program without being adopted. When they started, the computers available simply didn't have enough memory and they had to load things in from tapes over the course of a mission. In a system as complex as the shuttle, there's a tremendous ripple effect to changing the flight computers, both to the physical systems and the procedures and training for all personnel involved. Given the risks, there's a lot to be said for not fixing it if that aspect of the shuttle isn't broken. |
2005/7/4-5 [Computer/HW/Memory] UID:38409 Activity:kinda low |
7/4 Ultimate Geek Award? Man recites pi from memory to 83,431 places http://www.msnbc.msn.com/id/8456677 \_ Lots of people have memorized the bible or koran, but I guess this is all the more impressive as it deals with numerical non- repetitive patterns. -John \_ Well you should really compare the information content. I think because of redundancy and predictability natural language has somehting like 5-6 bits per word, and because of randomness there it like 2.3 bits per digit of Pi. \_ According to the article, this guy memorizes the digits while the current record holder calculated the digits from memory. \_ What the fuck is that supposed to mean. They meant "recite", not "calculate" \_ According to the article, this guy = current record holder. \_ "If verified and recognized by the Guinness Book of Records, Haraguchi's feat would beat his own previous best - currently under review - of 54,000 digits. The official current record- holder, also Japanese, calculated pi from memory to 42,195 decimal places in 1995." \_ You're an idiot. Almost as much as the "journalist". |
2005/6/20-23 [Computer/Networking, Computer/HW/Drives, Computer/HW/Memory] UID:38215 Activity:nil |
6/20 I'm having problems transfering pics between my hard drive and the memory card. Soemtimes it works, sometimes it gives I/O error or other error messages. Is the problem likely due to a corrupt memory card, the cable I'm using, or something inside my computer? \_ Though the motd may often seem prescient, you might have better luck debugging this yourself by testing the components separately, and seeing if any one component causes the errors to recur. -dans |
2005/6/8-9 [Computer/HW/Memory, Computer/HW/Drives] UID:38037 Activity:low |
6/8 Did anyone end up buying a dell after the discussion on noise level last month? How did it turn out? (my company is getting some, and I'd prefer them to be quiet) http://csua.com/?entry=37704 \_ Buy a Dell and you support Michael Bolton, Christians, the War in Iraq, more borrowing and Deficit, more oil drilling, more Corporate power (Enron), and the end of diversity. \_ I'm Michael Bolton's biggest fan. I own his entire collection! His cover of "Sitting On the Dock the Bay" was the best ever! \_ I bought a Dell and I support a nice, reliable server with marginally acceptable Eastern European support but a good raid controller. -John |
2005/6/3-4 [Computer/HW/Memory] UID:37957 Activity:low |
6/3 From http://csua.com/?entry=37951 ...Skipping all the mumbo jumbo about legality and what not, if it came from Proliant, I'd guess that they're registered DIMMs. What kind of motherboards are you putting them into? Most consumer boards don't support registered memory. \_ This is probably the most useful reply I've ever gotten. The eBay site doesn't say anything about the RAM besides the fact it came from a Compaq Proliant Server. Now that I'm looking at the RAM, it says "256MB, Sync, 100Mhz, CL3, ECC." Does Sync mean "Registered RAM?" My Motherboard is an ASUS 5PA and accord to the manual it takes 256MB 100Mhz modules. For ECC it says "ECC or Parity Support, 66Mhz only," and I've set the jumper to 66Mhz but still no luck. The manual says nothing about registered RAM. What do you think? Thanks, -pp \_ Actually, one of the "keys" (dimple in the middle of the connector pins) is different for buffered and unbuffered DIMMs. In that sense, it _shouldn't_ even fit correctly if the type is incorrect. \_ Not true. Plenty of memory that 'fits correctly' is incorrect and will not work. \_ Which doesn't address the specific case of whether buffered vs unbuffered are physically different. \_ Just downloaded manual from ASUS and checked. It says: This motherboard uses only Dual Inline Memory Modules (DIMMs). Sockets are available for 3.3Volt (power level) unbuffered Synchronous Dynamic Random Access Memory (SDRAM) of either 8, 16, 32, 64, 128MB, or 256MB to form a memory size between 8MB and 768MB. Unbuffered == Non-registered. To tell if it's registered, see if there is any "register" chips on the DIMM. You should see either 9 or 18 DRAM chips since it's got ECC. If you see other chips that are not capacitors nearer to the contacts, you've got registered DIMMs. If so, it probably won't work on that board. \_ I got 9 chips, and on the bottom are two smaller chips which I assume are registers? Ok, so I bought the wrong type. The eBay site doesn't say anything about ECC/Register, so it's my fault then? Should I resell it or report to eBay on ambiguous/misleading eBay? \_ I don't see what you have to complain about. I mean, it sounds like this guy didn't know what he had, and you didn't know what to ask. Life lesson I guess. |
2005/6/2-4 [Computer/Companies/Ebay, Computer/HW/Memory] UID:37951 Activity:moderate |
6/2 Help. I purchased 2 RAM modules from eBay, and the listing says "These units were pulled from a Compaq Proliant Server being stripped for parts and are in excellent condition." At the same time, it says "Item sold as is." When I got the RAM they have a strong burnt smell, and just as I feared, they don't work on ANY of my motherboards. I contacted the seller and he says sold "AS IS." This guy is dishonest. I feel jipped. Can I actually do something to remedy this, besides giving negative feedbacks? Does dispute resolution actually work on eBay? \- you can set up throw away accounts and mess with his future auctions. --psb \_ I thought about that but I don't have throw-away credit cards required by eBay. Come on, isn't that kind of childish? \- you can try googlebombing him then. if you cancel a CC does ebay find out about it? \_ oh, googlebombing, I'm sure the seller is SCARED!!! OOOO \_ Skipping all the mumbo jumbo about legality and what not, if it came from Proliant, I'd guess that they're registered DIMMs. What kind of motherboards are you putting them into? Most consumer boards don't support registered memory. \- You can resort to credit card dispute resolution service. that has worked for me 3-4 times. dunno how that will work in this case [assuming you paid via paypal]. \_ Isn't that totally the point of AS IS? What you should look for is a "DOA guarantee". "As Is" means "buyer beware". \_ Unless the post said that they were working, it's your fault. I always avoid "AS IS." \_ Excellent condition would imply working. The seller probably had no clue if they worked or not and didn't want to test them out. The seller is just covering his ass, he should refund you your money but he doesn't have to since it was "as is". \- i do agree that if you didnt even bother to ask the seller "did you even try to verify it is working or do you have no idea" the fault is yours. a nice seller out to refund minus shipping maybe but ethical and legal obligations diverge ... \_ I honestly didn't see the as is claim. Looking back, it's embedded in a very small print, "Seller's payment instructions Please see item description Item sold as is." It's a run-on sentence too. Argh, not happy about this. \_ Based on what I remember from contracts (and a quick search on lexis), "as is" effectively cuts off any liability the seller may have had. \- the large print giveth and the small print taketh away. i doubt you have any legal recourse that makes sense to pursue. it's a matter of cost-benefit now ... i.e. is a bad review worth the money to the seller, how much time do you want to spend on this being an asspain etc. \_ Actually OP says that the "as is" clause was in the small print which he didn't really see. This may change things. iirc, a statement that the ram is in "excellent condition" could be viewed as an affirmative representation of the quality. Even in used goods, this can give rise to a limited warranty. If OP had seen or should have seen the "as is" statement then the warranty would be waived, but if OP couldn't have seen or had no reason to see the "as is" statement, then the warranty may not be waived. Contracts wasn't my best subject so I could be a bit off here. I agree that it isn't worth bringing a case in small claims ct over this, but OP might have a valid claim. \- as you know even if he gets a small claims judgement he will never get the sheriff to go collect for him. how much money are we talking about anyway? what is LOCATION of seller etc. \_ My location is CA, his location is WI. Cost is $40. I can spend $30 on small claims+serve, plus $600 plane ticket. I don't mind doing in on the basis of PRINCIPLE. That motherfucker needs to learn a lesson. On the other hand, it may be just as worthwhile to create eBay accounts, win, then give him bad ratings. The second option looks very tempting so far. How do I go about creating disposable eBay accounts? And will they affect my existing accounts? -op seller may have had. If you had asked him if the ram worked and he had said yes, you might be able to rescind the sale. Absent this, you can't really do much other than give him negative feedback and maybe annoy him by calling him up randomly and leaving irate messages on his voice mail. i doubt you have any legal recourse. it's a matter of cost-benefit now ... i.e. is a bad review worth the money to the seller, how much time do you want to spend on this being a pain in the ass etc. \- you may wish to read Martin Amis: The Information \_ The eBay way to handle this is to request a refund, and if denied leave negative feedback to the effect that he knew he was selling burnt-out goods. He will either leave a negative feedback on you, or negoriate to get you to withdraw your feedback in exchange for a partial refund. |
2005/6/2-3 [Computer/HW/Memory, Computer/HW/Drives] UID:37946 Activity:nil |
6/2 The ultimate ram disk: http://tinyurl.com/dcenq (infoworld.com) |
2005/5/7-9 [Computer/HW/Memory] UID:37567 Activity:nil |
5/6 Just curious what's the upper bound on the number of flash drive read/write before the sector fails? Is it 10^10? 10^20? ?? What about floppies, and HD? Just curious. \_ For flash drives it's around 10^4-10^5 IIRC. \_ What about Secure Digital (SD) cards? \_ Same technology as flash drives. 'Secure' refers to DRM support. |
2005/5/5-7 [Computer/HW/Memory] UID:37539 Activity:moderate |
5/5 Is it a good idea to buy refurbished flash drives? \_ Is it a good idea to be with a refurbished girl? \_ No. Virgins are preferred. \_ no moving parts, acceptable. \_ Finite number of writes to a flash device. Of course, the number is in the hundreds of thousands (if memory serves), so you're likely safe buying refurbished, except in very extreme circumstances. \_ How do you refurbish a flash device? -John \_ Probably hatever problems people had when they returned them were really user error. |
2005/5/4-6 [Computer/HW/Laptop, Computer/HW/Memory] UID:37522 Activity:nil |
5/4 Is kingmax memory a reliable brand (for laptop)? \_ yeah, it works fine just do a http://newegg.com search for "kingmax memory" and you'll get pages of 4 or five egg average user ratings. pages of 4- or 5-egg average user ratings. \_ tnx for the response (I just didn't know if ratings at a merchant site should be trusted). Another question: are 2.5v and 2.6v memory compatible? http://crucial.com give me a 2.5v module for my notebook, while kingmax has a module that has the same spec except a 0.1v higher voltage. |
2005/4/27-28 [Computer/HW/Memory] UID:37380 Activity:nil |
4/27 I want to add memoory to my compaq presario R3000z. Various sites let me choose between pc2700 and pc3200 for this model. Will pc3200 be actaully faster for this system? I couldn't find any info from hp that clarifies this. ok tnx. \_ Yes, although if you mix and match I believe that it doesn't matter, I believe that it defaults to the lower speed memory. It also may not matter if the mobo maxes out the speed, it has something to do with the multiplier that affects both your cpu speed and the memory speed. YMMV. \_ I am sure there is a place where one find technical publication on the ram for my laptop but just don't know where to look for. I can't find anything using google. Where else should I look? \_ http://crucial.com appears to suggest that there are two models -- one with 2700 and one with 3200. They have a "system scanner" that might tell you what you have now. |
2005/4/17-18 [Computer/HW/Memory] UID:37226 Activity:nil |
4/17 For the guy who was looking for portable flash memory backup: http://www.nytimes.com/2005/04/14/technology/circuits/14pogue.html \_ Thx! :-) |
12/24 |
2005/3/24-28 [Computer/HW/Memory, Computer/SW/SpamAssassin, Computer/SW/Unix] UID:36849 Activity:nil |
3/24 I have a procmail process on Linux that I would like it to talk HTTP to a servlet (or any URL for that matter). What is the most efficient (the smaller the memory footprint, the better) and the most scalable (we do have heavy email volume) and the most performing way you can think of? TIA. |
2005/3/24 [Computer/HW/Laptop, Computer/HW/Memory] UID:36844 Activity:high |
3/23 I'm thinking about getting this laptop but the 60GB to 80GB upgrade adds about $150, and the 256MB->512MB RAM adds another $150. Is there any reason why I should do this, when I could potentially just buy a 100GB drive for $230, and likewise something similar with RAM? \_ Tell us which notebook you want and maybe we can give advice. E.g., the ThinkPad T42 has only one accessible memory slot, the Fujitsu S7010(D) doesn't have a user-accessible hard drive, and some ultra-light notebooks accept only very thin hard drives. \_ In a T42, that one slot is shipped unpopulated, so you could still slap in an extra gig for ~$200. \_ how the heck do you repair/change a HD if it's not user-accessible? I mean, is it like welded or something? \_ Not user-accessible doesn't mean its not accessible, it just means you'll have to go through (possibly damaging) grief and possibly warranty voiding stickers to get at the hard disk. My Sony VAIO Z505 back in the day was like this. Since I didn't care about the warranty, I managed to crack the fucker open and upgrade the HD. I managed to do it with minimal damaage (i.e. minor scratches on the pretty case). Similarly, opening up my Powerbook isn't happening, and I'm not going to risk my extended warranty to do it. not going to risk my extended warranty to do it. -dans \_ Fujitsu T4010D (yes it's pricy but I REALLY want it) -pp \_ If the 60GB is 7200rpm you might not want to 'upgrade' to a 80GB 5400rpm. \_ Search http://tabletpcbuzz.com for user comments. I found some on google already just by searching "t4010d upgrade". \_ Get the RAM, don't get the drive. 20 GB is not worth $150. If you need space later you can always add a USB drive. You will always use the RAM, though. \_ Did you check http://portableone.com? They used to be the #1 seller of Fujitsu notebooks before http://newegg.com moved in, not sure now. 40GB->80GB is $200, 40GB->60GB 7200rpm is $170. (I have the 7200rpm drive, and it's great, no problems.) 512MB is standard. 3yr warranty is cheap too. Oh look, you can also customize the display to something that's indoor/outdoor and 180 deg viewing. |
2005/3/19-20 [Computer/HW/Memory] UID:36768 Activity:nil |
3/19 I have 2 HD's. When I transfer 25G files from one that is UDMA/5 to another one that is UDMA/6, it takes 15 minutes, or roughly, 27MB/sec. This is still well below the specifications. I'm sure that the 100MB/sec specification is probably based on certain conditions, like data already in cache, sparse-data-compression, or some other ideal conditions that rarely exist in real life. So now I'm wondering, what is a typical SUSTAINED read/write rate for UDMA/5 and UDMA/6? -ok thx \_ The 'Ultra100' spec means the drive's connection with the system can do 100MB/sec. That is not the speed of the drive itself and you will only see it when reading from the drive's cache. 27MB/sec sounds about right to me for copying. Good drives can sustain 40-50 MB when reading, or 30-40 when writing. Obviously there's more over head when doing both at once and your drives are not say 10K RPM. \_ 100 MB/s is the interface speed. ~ 27 MB/s is the maximum sustained transfer rate of your drive(s). Interface speed is primarily limited by electrical signaling. Maximum sustained transfer rate is determined primarily by the data density of a platter (how many MB per square inch) in the drive, how fast the platters are spinning, and finally, by a combination of how fast the heads can seek to different tracks and how fragmented your files are. \_ FYI I my 2 HD's are on different IDE cables and they're both 5400RPM. I wish they have a more standardized benchmarks. |
2005/3/6-8 [Computer/HW/Memory] UID:36550 Activity:nil |
3/6 My laptop memeory use PC2700, but I have a PC3200, other things being equal. Is memory downward compatible or it has to be an exact match? \_ Yes, no. |
2005/3/5-8 [Computer/HW/Memory, Computer/HW/Drives] UID:36543 Activity:low |
3/5 Can a PC boot from external (usb or ieee-1344) hard drives? \_ did you mean 1394? \_ depends on your bios support. Quite a few now can boot from external usb-anything \_ If you're trying to boot XP, look for Bart's PE Builder. For a USB-bootable Linux, have a look at http://www.knoppix.net/forum/viewtopic.php?t=12964 I have another one lying around somewhere, so drop me a mail if you want me to dig for it. -John |
2005/2/25-28 [Computer/HW/Memory] UID:36429 Activity:low |
2/25 To people running VM Ware, how much memory do you have? I have p4-1.7 and 256mb and its slow. How much memory do I need? \_ A lot. You are running 2 operating systems. I'd double your RAM at a minimum and more is better. \_ Two OS's plus VMWare itself. \_ 256 MB of RAM is not that much. If you're using a Windows host, it's barely enough for even that one OS. \_ 1 GB. RAM is SO cheap. Get plenty. \_ My stupid machine at work uses Rambus memory, fuck. \_ I've heard that when you run VMWare, it's even faster to disable your host virtual memory so that you don't get double swapped (where your host AND your virtual machine swaps at the same time). Is this really true and has anyone benchmarked this? \_ I run a 192 mb linux server as guest with windows as host on 256 seems to work fine, though I only use the linux os remotely and haven't tried doing real work on the windows box at the same time --darin \_ I've always wondered if you can run a simple client on a P100 and hook it up on a Win based server |
2005/2/18-20 [Computer/HW/Memory, Computer/SW/Unix] UID:36235 Activity:low |
2/18 I have a 64MB USB thumbdrive and I'd like to put a small version of knoppix on it to use as a rescue medium. I don't want X. I do want all the cool hardware detection that knoppix does so well. Any ideas for something already made for this? \_ http://www.knoppix.net/forum/viewtopic.php?t=12964 YMMV. Let me know if it works. I want to do the same for Auditor on 1GB (http://new.remote-exploit.org/index.php/Auditor_main -John \_ Hi John, I found this, but can't find the USB for mounting the root FS at bootup. Has an initrd though. \_ Hi John, I found this, but it can't find the USB for mounting the root FS after bootup. Has an initrd though. http://www.tux.org/pub/people/kent-robotti/looplinux/rip It is well-documented, but not as cool as knoppix. -brett \_ Nifty--I intend to muck around with this sometime next month (don't have any working unix boxes right now) so if you drop me a mail I'll let you know if I figure out how to do it. It's also tremendously reliant on whether your bios can boot from usb, in what order you load the drivers, etc. You may also want to look at M0n0BSD (http://www.m0n0.ch -- I'll see the author in a few days and can ask him for help. -John |
2005/2/17 [Computer/Companies/Apple, Computer/HW/Memory] UID:36223 Activity:nil |
2/17 I need to buy a computer for photo-processing. Currently, on a 3 GHz P4, it takes Photoshop ~1min. to open up a 5-6MB raw file, and even longer to do any processing. Are any PC-clones comparable to Apple's Power Mac's in terms of photo-processing power? A Dual PowerMac1.8 w/1GB RAM, and 20" screen is $3300. Top-of-line PC's are much cheaper. Suggestions? Advice? TIA. \_ If you buy the G5 w/o the stock ram and just buy 1 GB of RAM from fry's you can easily save $300 or so. Also, if you are price conscious get the Dell 20" wide-screen lcd rather than the Apple, you will save $300-$400 there as well. Couple of other options to save money on a G5: 1. buy from amazon -> no tax 2. find a friend who is still in school and get a student discount 3. find a friend who has a apple employee purchase program (sun, cisco, lockheed, &c. have such programs) 4. get a refurb from http://apple.com (I've bought several w/o any issues) If you want to stay on PC, perhaps you may want try a dual proc board (PS on Mac is optimized for dual proc, probably is on pc as well), perhaps a dual opteron w/ a 9800 or a X800. |
2005/2/17 [Computer/HW/Memory] UID:36209 Activity:nil |
2/17 I have an early 12" powerbook g4, with a 867MHZ cpu and officially max ram of 640M. It was soon replaced by something with higher speed and 1.12G max ram. I saw some memory chip sites that implies I can put a 1G chip in the rev A model I have as well. Can anyone confirm or refute this? tia. |
2005/2/16-17 [Computer/HW/Memory] UID:36193 Activity:low |
2/16 To those that weighed in on the memory problem for me (peterm) yesterday, thank you. We have found purify lacking in the past, and it turns out the memory problem was not a leak but rather overallocation with no deallocation. Our guy (not me) found the problem by watching the memory use as he stepped through with a debugger! He had tried to use valgrind, which found a tiny amount of leaked memory, not the actual problem. I've cut the thread out of the motd and placed it into ~peterm/motdentry.txt \_ Sounds like a leak to me. \_ The data structures were still linked. A "leak" is when the memory remains allocated but isn't linked so you couldn't deallocate it if you wanted to. Not deallocating was sloppy. --peterm \_ Not true. A memory leak can be defined as above poster suggested. In modern parlance with GC this would very much be defined as a leak. problem by watching the memory use as he stepped through with a debugger! He had tried to use valgrind, which found a tiny amount of leaked memory, not the actual problem. I've cut the thread out of the motd and placed it into ~peterm/motdentry.txt \_ I've used valgrind to find leaks like this in the past. It has a way to show you ALL memory allocated not just leaks. It was a bit of a hassle but it worked pretty well. Valgrind is pretty damn cool. -aspo |
2005/2/13-15 [Computer/HW/Memory] UID:36162 Activity:high |
2/13 Buying 1GB PC3200 DDR memory for my G5-- any suggestions? I'm leary of the top 5 returns from pricewatch.... \_ on a G5 make sure you get good memory and not some generic PW crap. I'd check Kingston to find out what sort of timing and stuff they'd put on their branded Mac compatible RAM and then go w/ a generic maker that matches those specs (or just get the Kingston/Crucial/etc at some discount shop). \_ Are you in the bay area? If so, just wait for Fry's to have a sale on 2x512 PC3200 DDR Ram by either Corsair or Kingston. A agree w/ the above poster, get good name brand ram for I agree w/ the above poster, get good name brand ram for your G5. \_ Geil \_ why are new computers so picky about RAM? I had similar RAM issues with my new Athlon MB which only took expensive RAM even though they had the exact same specifications. So much for plug-and-play. It's becoming more like plug-and-hope-to-play. \_ The reason that newer computers are picky is that as cpus get faster and faster you have to keep the damn thing fed w/ data faster and faster. This means that that the RAM must strictly adhere to the cpu's tolerances or things just won't work (if signal coming from ram is off by just a few ns, it didn't use to be a big deal, now, it is well nigh impossible to tell if there was an error, &c.) While the cheap ram might work, they clearly do not test the chips as rigorously as the brand name vendors. \_ Uhm, no. Computers have ALWAYS been picky about memory. We used to have issues on SDRAMs during the PIII days, especially when the first chipsets came out. I still have a board that only takes Samsung SDRAM. You need to go read your Mobo manual and find out what type and make of RAM it supports. Generic cheapo RAM sometimes are pulls and there are timing issues with pulls. Certain Mobos just are very sensitive to timing issues on RAM, so be careful. -williamc \_ No, they haven't. Previously, only slot type and layout. -John \_ How old are you William? I'm 35. Back in my days when Pentium 75Mhz was hot and everyone overclocked it to 100Mhz, we could use whatever RAM we wanted. You're right in respect to your time frame, and I'm right in respect to my time frame. Kids these days... \_ Dude, you are a fossil. \_ Young punk. When I first started w/ computers RAM was ferrite cores. \_ I used to draw on my dad's old punch cards with crayon when I was a kid, does that count? \_ At least you know what a punch card is. My intern saw one and though it was some kind of ballot. \_ gosh I miss the punchcard days when punchcard specifications were simple and complete and they didn't have any compatibility issue. Nowadays everything's bigger, harder to specify, and near impossible for formal compatibility validation. \_ All heil motd, 70's version of newsgroup, still operational as of 2005 -elite guy from the 80s \_ Is there a difference between a single 1GB stick and two 512mbs? \_ Yes, two take up two slots and can run in dual-channel mode. \_ Clever. Is there a difference in performance? \_ There is with single vs. dual channel if the motherboard supports it. Otherwise it should be the same unless the mobo specifies otherwise. \_ FYI, the G5 supports dual channel. \_ you must install symmetric pairs of RAM stick in a G5 |
2005/2/3-5 [Computer/HW/Memory, Computer/HW/Drives] UID:36055 Activity:moderate |
2/3 So, I'm really confused about this notion of computer science metrics where everything's kind of 2^n, but sometimes not. For example, if I have a 4 GIG hard drive, does that mean I have exactly 2^32 bytes of space? Does that translate to 4000000000 bytes, or some number that's close to it? How about megahertz? Say I have a 2.5GHz computer, does it run at exactly 2500000 hertz? Or 2^n for some multiple of n? \_ As a few other people have said, just about everything is ordinary decimal units now. A 4G hard drive is (about) 4,000,000,000 bytes, and a 2.5GHz processor runs at (about) 2,500,000,000 Hertz. Operating systems still tend to report file sizes in binary units, though, so a 4G file is probably 4,294,967,296 bytes. (Sometimes it's configurable: GNU du and df let you specify -h for binary units or -H for decimal.) The only hardware still sold in binary units is memory -- a 1G flash memory card is 1,073,741,824 bytes. Also note that all storage devices are sold by raw capacity, not counting filesystem overhead. \_ I'd actually argue that in computers, all standards and measurements are 10^n with the exception of RAM and addresses. HDDs, bandwidth, frequency, resolution (megapixels) are all base ten. \_ The ISO standard (IIRC) is Gibibytes for 2^30 and Gigabytes for 10^9. Memory is addressed by logic which is friendly to powers of 2. Non-memory doesn't matter. \_ Frequencies (Hertz) are not stated in powers of two obviously. You can usually find the exact frequency in the tech specs. Computer data quantities are normally referred to with binary prefixes (kilo=1024). Hard drive producers use 1000 because they are fuckers. A frequency is not a data quantity. \_ The whole k=1024 thing was a cute hack invented by computer folks, but it becomes imcreasingly strained as you move into mega- and giga- prefixes. At some point you need to admit it's more trouble than it's worth. The computer should be the one worrying how to address it's memory in base-2, rather than the end user wondering how many bytes are in a gig. \_ Memory is sized in 2^n because that's how the chips are laid out. The capacity of a hard drive is determined by the track width and magnetic domain size. This gives you a non-binary capacity. It makes sense to therefore measure it using the SI system. The fact that CS people started calling 1024 and 1048576 kilo- and mega- when those prefixes had been in use for ages says to me that the CS people are the fuckers, not the engineers who are just adhereing to standard terminology. I personally count all file and data sizes in base-10, except when refering to memory usage. Anyway, this debate has been done to death before. \_ so 1M is what really? 1000000 or 10^20? \_ Officially, 10^6 (you mean 2^20, right?). M = 10^6, Mi = 2^20 See: http://physics.nist.gov/cuu/Units/binary.html \_ No, it doesn't make sense when you're talking about computer storage capacity to use a different meaning of GB than everybody normally uses when talking about computer storage. It's not like the drives are storing non-binary data. It's going to have a filesystem and store vanilla kilobytes and megabytes. Nowhere else does Gbyte refer to 1000*1000*1000. In a computer environment, files are loaded into memory and to disk, it's idiotic to change the terminology just because the underlying media is different. CD and DVD storage is referred to in binary. Sorry, you're wrong. \_ FWIW, floppies are counted under a bastardized hybrid system where 1.44 'megabytes' = 1440 KiB, or 1.44k-KiB \_ No. CDs and DVDs are counted base-10, as is bandwidth. The fact that the drives are storing binary data has no bearing on the method you use to count the bytes, which this debate shows, is a matter of dueling conventions, not some underlying fact. I'm not wrong, you're just an ass. \_ The orange book standard says a cd has a capacity of 650*2^20 bytes. \_ A DVD+-R(W) is 4.7*10^9 \_ Hard to argue, since they charge $5,000 for a copy of a DVD format spec ($500 for each additional spec) and require an NDA. Do you actually have access to the four specs you mention? The holder of DVD Forum's specs is http://www.dvdfllc.co.jp |
2005/1/19-20 [Computer/SW, Computer/HW/Memory] UID:35785 Activity:kinda low |
1/19 What's the difference between an instruction cache and a target branch buffer? \_ This is the fucking csua, we talk about politics and girls, who do you think we are? bunch of nerds? \_ Doing your CS152 homework? Oh wait, now it's semester break. Been out of school for too long. \_ I am guessing. an intruction cache stores previous instructions, whereas a target branch buffer fetches the instruction which the branch prediciton logic anticipate. (thus the target branch) \_ An instruction cache also stores next instructions if, say, the cache only reads from memory 16 bytes at a time and the instructions are less than 16 bytes each. \_ a branch target buffer is a place to store metadata about recently used branches, e.g. profiling statistics to inform branch prediction logic. it's analogus to a TLB in that it holds the scratch state for hardware logic meant to accelerate the typical case execution beyond the worst case one. |
2005/1/10-12 [Computer/HW/Memory, Computer/SW/OS/VM] UID:35646 Activity:nil |
1/10 I have VMware 3, how do I get around the 896M RAM limit? \_ I don't think you can. Isn't it a technical limitation? Upgrade to Workstation 4.5. \_ Workstation 5 is available in beta. Release date? |
2004/12/6-7 [Computer/HW/Memory, Computer/HW/CPU] UID:35186 Activity:high |
12/6 I'm at my wit's end with a problem. I'm running WinXP and a lot of newer 3D games make the machine randomly lock hard. There isn't much pattern to it, but some games do it and some don't. This problem is not mentioned under tech support for the games besides the generic "Update your drivers and DirectX" which I've done. Games which crash include Railroad Tycoon 3, Rome:Total War and Sid Meier's Pirates, and in the non-crash group is GTA:Vice City, Doom3, and MS Flight Sim 2004. I don't know what else to try, is it a hardware failure of some type, or maybe AGP apperture? This is a GeForceFX5700 on an Athlon XP. \_ SP2? Running in safe mode works? Maybe bad memory? -John \_ SP2. Can you use D3D in safe mode? Other apps are stable. \_ SP2 fucks up a lot of apps and games. There are lists of known problems at the MS site. I think there's a way under safe mode to select the drivers you load. Just as a terribly stupid idea, try setting up a vmware session and running the game inside that--it might give you some more clues. -John \_ The SP2 issues page seems all about firewall issues, which is not the problem. Safe mode, while a good idea, prevents DirectX from working. \_ I had some similar problems after I installed a new video card in my old computer. Make sure you are not overloading your video bus. Try turning down the AGP rate from 4X to 2X or even 1X and then see if it still happens. I tried a bunch of things before I hit upon this. You can set it in your start-up screen. -ausman \_ Motherboard supports 2x and 4x. Switching to 2x did not fix it. \_ Do you have the VIA chipset? I do, and I have the same symptoms, as documented on <DEAD>viaarena.com<DEAD>. -ax \_ MB is Abit KG7, with AMD 761 northbridge and VIA 686B south. \_ What did you do about it? -ausman \_ I got used to not being able to play the Medal of Honor series. Most other 3D games work fine. -ax \_ Could be something overheating. \_ But Doom3 runs fine for hours. \_ My younger brother had this exact problem after we added another Corsair CL2 DIMM to get him to 1 GB. Entered BIOS settings and turned CAS settings from Auto or CL2 to CL3 - now everything works fine. \_ Some games worked and other didn't? -op \_ I don't think you should put too much concern on the "some game work and other don't". The game that works may just be not exercising the CPU enough or something. I would suggest the following, download "Ultimate BootCD" and try a burn-in test, if your mahcine locks up, it's BIOS or hardware problem. If it passes, then maybe a software problem, reinstall XP from scrach and see if it helps. If it's a software problem, most likely it'll go away. \_ Yes. \_ Can you start to play some games and then they lock up, or will they lock up your computer on startup? \_ In my case they will start, then lock randomly in 1-10 minutes. Screen freezes and sound gets stuck in a loop. -op \_ Now's the time to upgrade to an Athlon 64 3500+ Winchester! -sl0da l0s3r \_ Yeah, exact same symptoms as mine. I am pretty sure you are overloading your graphics subsystem. Have you installed the latest firmware upgrades on both your graphics card and your mother board? Could be a sound card problem too, but it seems unlikely from what you have said. Any kind of hardware or driver changes lately? -ausman \_ Same symptoms for me too. -ax \_ OK, an update for those whove been helping me: I ran memtest86 and one other memory tester. Both ran for a while reporting no errors until the system suddenly locked up. I ran the mersenne prime torture test and had no problems. So something is flakey, but how can I be sure it's the RAM, not the CPU? My motherboard monitor reports in-spec temperatures and voltages both at idle and when running GIMPS. \_ Can you try a stick from another PC? \_ I have 2 different sticks in there now, both in use for at least a year. I'll try selectively pulling one later. For now I'm wondering about a good CPU-only stress test (like one that uses only a tiny bit of memory and no mallocs() ). \_ OK, I tried each stick on its own. One crashes very fast while the other takes a while. So, either I have 2 bad DDR sticks, or something else is fucked. Suggestions? \_ It's your motherboard. Abit KG7s are know to be flakey. Return it and get something not Abit, like a Gigabyte. Abit + P4 + Via also has problems. I only recommend Abit if it's the more expensive standard Intel chipset version. Avoid Via chipset abits, especially if you are using P4. Also, K series is known to be flakey. You can try upgrading the BIOS and see if that helps, but I never got one to work reliable, especially on high load stuff. -williamc \_ I agree with williamc, I think it's my motherboard as well, but am too lazy to change it since my disks are raided. -ax \_ BIOS is current. I'm afraid you might be right. Can you reccomend a stable MB for an 266FSB Athlon? \_ I'v been using a GA-7VT600 gigabyte for some time now with Linux/XP and no problems. It's got a KT600 and it can support up to 400 FSB. BTW, the reason why your MB is flakey is because of the chipset combo. It's a via/amd solution that just doesn't work. Avoid combo solutions like this. Plus, whatever you do avoid the KT266 like the plague (older via chipset). Your solution was due to the KT266 instability (the A version doesn't help matters much, just avoid any MB like this) Via cleaned up its act aver the KT266 debacle and Via+Athlon is pretty stable now. Via+P4 is the crap, especially the P880 series. -williamc \_ Motherboard having hardware issues does not explain why it would lock up in some games and not in others. Do you mean "flakey' in some other fashion? \_ Yes it does, I've seen this before on Linux also. 90% of programs ran fine, even a kernel recompile. But when we began to load the system up with circuit simulations the thing would segfault at random. It took me a week of replacing parts to find out it was the MB initially. What was interesting was that stuff started to fail in succession. First there was the segfault issue, then all of a sudden kernel recompiles failed, then the IDE controller went out for no reason. Chipset problems usually only appear if you stress the system long enough. That's why it passes QA and gets put on the shelf. -williamc \_ Presume that any game can lock the system by sending a corrupt instruction to the video driver or AGP controller. Now if there's a MB/chipset that is randomaly corrupting some writes to main memory, a game which makes more writes has more chances to get stung, and send a corrupted instruction. Just one explanation. \_ Yeah, but shouldn't that mean that he should see Doom3 crashes sometimes as well? Less frequently yes, but none at all? Something else is happening that we are not quite understanding here. \_ RR Tycoon, Pirates, and Rome must have something different about them than GTA, Doom, and flightsim. Maybe he should try underclocking his FSB and see if things get more stable. Anyway the memtest indicates the RAM is the problem area. \_ Memtest doesn't report any problems, it flat-out locks the system. This is with 2 different RAM sticks. -op \_ No, you're wrong. It's the MB, plain and simple. Just because it appears as if it's a memtest problem doesn't mean it's not the MB. Trust me, it's the MB. I have direct experience with Abit K series MBs, I've seen stuff like the IDE controller going wonky, the serial ATA going wonky, the video AGP problem going wonky, etc. The memtest is only one symptom of basically a chipset problem. Hardware is complex, it's not like software where you can pinpoint stuff to one dll and say the bug is there. It's the MB, stop wasting time and just get it replaced. It costs $50. -williamc \_ Thanks for your help. What do you think of this mobo: http://tinyurl.com/4ul8f \_ It looks fine, I've never had an issue with nvidia chipset. I've had some minor problems with MSI, though, but that's on the P4 side. Your athlon should be fine as long as you get a good board, MSI makes decent (but not really outstanding) boards. The nvidia I have running an athlon is an ASUS from last year. -williamc \_ Could also be the power supply? \_ Hardware monitor reports all voltages 3-5% over spec. -op \_ Sounds like a virus. Reformat and start over. \_ Are you sure? Re-read the thread. \_ http://forums.eyo.com.au/arc/t-1142.html Relax your memory timings. Reduce bus speeds. |
2004/11/30-12/1 [Computer/HW/Memory, Academia/Berkeley/CSUA, Computer/HW/Drives] UID:35127 Activity:nil |
11/30 So can we get a hardware summary (purchased and planned) for the new soda machine(s)? It sounds like there's a dual Xeon getting an OS somewhere-- what kind of ram, what kind of storage does it have? How much was it? How much disk does/will it have? Maybe a file in /csua/tmp would be helpful. Thanks. \_ Maybe a CSUA wish list would be helpful. \_ I would like to see this too, but purely out of curiosity. I think that the CSUA admins are probably doing something pretty reasonable. --PeterM \_ Seconded... I'm kinda excited about what we're getting |
2004/11/16 [Computer/HW/Memory] UID:34924 Activity:nil |
11/16 I did some simple performance test of NS7.1 vs. FF1.0 on my 400MHz Celeron NT machine: NS7.1 FF1.0 Start-up time with blank page: 5.0s 4.0s Memory used after start-up: 9.8MB 8.8MB Memory used for http://www.yahoo.com 1.7MB 1.1MB When measuring start-up time, I started and exited the browser several times to make sure most of the disk files are cached in RAM before I took the actual timing. The timing is very consistent. When measuring the memory used for http://www.yahoo.com I loaded the page into the two browser only a couple minutes apart. The same ads and news headlines were displayed in both cases. I exited one browser before I start the other. \_ This is cool... what about IE? \_ I only have IE 5.something on that machine, so I didn't check IE. \_ What about the rendering time for cnn/ebay/etc (IE vs FF)? \_ I only have a modem connection, so I didn't measure that timing. |
2004/11/15 [Computer/HW/Memory] UID:34903 Activity:very high |
11/15 Can someone tell me why 64 bit machines are the future? I program a lot of things and I rarely need 2^32 memory. My guess why people move to 2^64 is because of RAM usage, but how the heck do you use up 2^32? And regarding addressing and page tables, wouldn't the size just blow up with 64 bit machines? \_ It's mostly in high-end engineering practices. Our ASIC CAD tools, for example, use memory like there's no tomorrow. Simulation waveforms and netlists take up a lot of space and many of our tools crap out on a 32-bit platform. \_ You don't write code to use up 2^32 memory. The stupid people in Redmond who wrote your OS do. \_ Umm... high-performance computing has been wanting more than 4GB of memory for a long time now, and that's because it needs that much, not because it's bloatware. \_ It'a conspiracy between M$ and Intel. If M$ doesn't keep on producing bloatware, nobody would be upgrading their hardward again and again, and Intel will be out of business. \_ The consumer doesn't need it right now, but industrial and scientific users do, and they use much the same chips you do. Also, games are using more and more memory, not from pure bloat but from the need to have hundreds of megs of textures cached. A single 3D scene can use half a gig of textures. \_ A lot of things can use more than 4GB memory. 4GB isn't that large anymore. Think of the size of hard drives and media file sizes. There is a need for 64-bit engineering tools and so forth. Sure it's not "needed", right now, by the vast majority of consumers. But it is an enabling technology and you're stupid if you can't see why it's "the future". With multitasking, I consider 1GB my personal minimum on my home machine and I don't do anything special. \_ Well, it really depends on your apps, but I rarely go over 512MB. -!op \_ Everyone is talking about memory, but memory is just one aspect. Being able to execute 64-bit operations natively is a very nice thing if you need to do that sort of thing. Lots of scientific and engineering apps do this and performance is much better with a 64-bit chip. \_ true enough but what about other things? Adding 64 bit integers take up much more ALU space and a bit more time than adding 32 bits. 64 addressing takes up more space for your cache and longer comparator for the TAG and other hardwares. 64 bit doesn't come free you know. \_ It's a net gain over 32 bit. \_ "640K of memory should be enough for anybody" \_ But that clown is still the richest man in the world. What can we say. \_ Well, wasn't he right? For the time it came out, to succeed it didn't need to support more than that. From a short-term business perspective it was fine. Long-term, it was fixable. DOS was crap in a variety of ways, why focus on that. \_ AFAIK, no one supports addressing the entire 64bit memory space directly (the memory to store the page table would be crazy) be crazy). There are other advantages to 64bit addressing for ordinary people. Imagine being able to open several 8 mp photos in ps and then being able to stitch them together without having to wait for stuff to load from disk. |
2004/11/10 [Computer/HW/Memory] UID:34801 Activity:nil |
11/10 A memory stick for japanO`philes http://www.japantoday.com/e/?content=product&id=795 |
2004/10/25-26 [Computer/HW/Memory, Computer/HW/CPU] UID:34328 Activity:kinda low |
10/25 What's the difference between intel's C and E cpu for p4? What about those processor with names like 540, 530? I am helping a friend pick components for a system. Any recommendations? I am thinking of Asus P4C800-E Deluxe with P4-3.0Ghz CPU and 1GB PC3200 memory, all for about $600. Is the newer socket 775 with ddr2 memory that much better? Not a gamer, just need a fast desktop system for general applications. \_ General apps means Word, surfing, and email. This P4 system is already overkill. It doesn't matter which socket, how fast the ram is, etc. Get your friend to buy a mid ranged Dell. \_ I do not recommend Dell. They don't use standard components, if they go bad, you are stuck with Dell supplies. \_ Dell desktops usually come with 3-year warranty and often 4-year warranty is an option. Besides, I have never seen a Dell motherboard or power supply fail. What other non-standard components are there? I have a Dell Dimension and I had no problems replacing video card, ram, disks, and optical drives. (BTW, those were really upgrades, not because the original parts failed). Yes, if you need a PC that allows you to swap a motherboard a few years later, Dell might be bad for that. \_ My original PS from about 1.5 year ago has gone bad (making loud noises). Dell sent me a new PS, but it's still loud. Their case is also non standard, ie, you cannot put the thing into a ultra quiet case (the motherboard connector to the case is non-standard) So you see, if I want quiet and peace of mind when I am working next to the computer, I have to replace the friggin motherboard as well, because the motherboard is not standard. I can't put in a ultra quiet PS that I can get at Fry's either, because Dell SPECIFICALLY changed the PS layout so standard ones won't fit. Their fan system is also non standard, the back fan that sucks air out of the CPU through the tunnel out the PC is also non standard. So now I am stuck with this loud piece of ($*%($% unless I replace the freggin motherboard, it's a good thing they uses standard CPU. ;) \_ If you care about noise, get an Optiplex. They're pretty quiet. \_ You don't need 'fast' for general applications unless that includes stuff like crazy Photoshop work or video editing, or maybe a PVR. If you're going to run WinXP on it I'd spring for 512MB of RAM... \_ Goto Fry's and buy whatever's on sale. It will be fast enough. \_ Has anyone ever actually bought one of Fry's $200 shitboxes? I'd expect it to explode 2 weeks later. \_ Actually, I used on sale parts from Fry's to to cobble together a machine. The motherboard was a piece of crap with a BIOS problem, but once I flashed the BIOS it was a good deal. So, if you don't mind fighting a little with the machine at first, you can come out really well. -jrleek \_ I think you want the C. The E runs hot, I think. I also think you want a 2.8 GHz CPU. 3.0 jumps up in heat a lot. Check http://newegg.com for user opinions on E vs. C. Then again, you could just get an Athlon 64 3200+ (Newcastle) or 3500+, which both run cooler and generally faster. Search the http://newegg.com reviews for any of the above CPUs if you want suggested mobos + memory to buy with them. |
2004/10/22-23 [Computer/HW/Memory, Computer/HW/CPU] UID:34286 Activity:low |
10/21 The Emperor's battle computer is now fully operational. ---psb http://www.tgc.com/hpcwire/hpcwireWWW/04/1022/108601.html \_ The battle between vector and scalar supercomputers continues! \_ I'm still waiting to see a tensor computer. |
2004/10/12 [Computer/HW/Memory, Computer/SW/OS/Windows] UID:34065 Activity:nil |
10/12 Only what I want to talk about matters and I will keep posting it and reposting it ad initium forever. Get used to it. \_ Your system is low on virtual memory. Windows is increasing the size of your Virtual Memory Paging file. |
2004/10/6 [Computer/HW/Memory, Computer/HW/IO, Computer/Theory] UID:33958 Activity:very high |
10/6 If the technology for borgification was available, what would you do? I think it would be pretty cool to have some sort of 'mind's eye' display and an always-on network connection. The other stuff like collective conciousness and major implantation I'm not so hot on. (I thought of this when seeing someone on their cell phone using a headset) \_ Depends what you mean by 'borgification.' I think it's already happening with cell phones, and the Internet, and it will happen more and more. As long as communicating with others is not intrusive (which are the connotations when you hear about the Borg), it seems like a good thing. -- ilyas \_ If a 'direct neural interface' was available, would you get it? \_ Spam and viruses, directly to your brain, yay! \_ Well, presumably it would handle only media and not 'code' (thoughts). You could get spam, but you'd have to decide to check your mail, which would be handled by a remote server made of silicon. \_ Depends on what I would interface with, I guess. I wouldn't mind things like the ability to instantly do arithmetic, or do google searches. Of course those things don't need a neural interface -- only wearable computers. Can you give me an example of a useful functionality which needs a neural interface? -- ilyas \_ I'd like the equivalent of monitor, keyboard, microphone, and speakers wired into my head. You can have any computer you want on the backend. \_ Why do you need it wired in your head? \_ direct neural interface + Windows 2050 = new meaning for 'blue screen of death'! \_ Which includes Buffer Overrun Suite 2050, the most direct way to overwrite your memory. \_ Be the first to beta-test Microsoft Bloodstream 1.0! \_ Apple iBorg. Think Different. No, Really. (TM) \_ Beware The Phone Company! \_ I just want AR (Augmented Reality) sunglasses. They'd give me heads up GPS stuff, overlay building addresses and street names, use facial recognition to bring up info I have tagged to people, etc. I think that would totally kick ass. --dbushong \_ not sure about borgification, but if I had the holodeck to myself, I'd endulge in my wildest dream. ****DROOOOLLLLL***** \_ 7 of 9 + 69? \_ 69.777... ? |
2004/9/17-18 [Computer/HW/Memory, Computer/HW/CPU, Computer/HW/Drives] UID:33587 Activity:low |
9/17 How difficult is it to host a virtualized FreeBSD server? Not commercially, just want to let a friend go nuts with it. Something like what's described at http://www.johncompanies.com/jc_bsd.html but I don't care so much about uptime and reliability. Is this guy just using a jail, or is it more vmWare-like solution? \_ Jail. Very straight-forward. I love the part about their special technology. You just need a unique IP/server. -John |
2004/9/6 [Computer/HW/Memory] UID:33372 Activity:kinda low |
9/6 Why are memory address changers called "Trainers"? What do you train? \_ The player. They let you keep from getting killed in 5 sec so you get a chance to learn the game. |
2004/8/26-27 [Computer/HW/Memory] UID:33165 Activity:moderate |
8/26 Can anybody else here clarify what this boinc quote means? "We fixed the bogus URL bug and all servers are back up. If your cache is empty of all valid workunits and you have any of the bogus workunits, then reset the project." How does one "reset the project"? \_ It sounds like some internal company lingo. Good luck! \_ not that i've ever used boinc, but based on hearing that it allows other projects besides seti@home, i'd assume it means zapping the data files or whatnot for whatever project you happen to be running. |
2004/8/19 [Computer/HW/Memory, Computer/SW/Unix] UID:33018 Activity:nil 77%like:33005 |
8/18 Anyone have any experience getting a boobable .iso onto a USB memory key (yes, it is boobable and has enough space). -John \_ I just did this recently. I haven't found a way to get a .iso directly on, but here's what I did: 1) Format the USB storage 2) use mkbt to get the boob sector from the .iso and then put it on the USB storage. (Get mkbt at: http://www.nu2.nu/mkbt 3) copy files from .iso to memory key I used daemon-tools to mount the .iso to rip the boob sector. Oh, and if you want to use Ghost's boob disk creator, you can use a virtual floppy drive so you don't have to use a physical floppy: http://chitchat.at.infoseek.co.jp/vmware/vfd.html \_ There's not even a readme for this. What exactly does it do? \_ Readme for which? vfd is a virtual floppy. Install it and you've got a virtual floppy drive. mkbt extracts boob sectors and writes them. \_ Oh, and this is where I got most of my help on this: http://www.weethet.nl/english/hardware_boobfromusbstick.php \_ Many thanks, swami. *bows* -John |
2004/8/19 [Computer/HW/Drives, Computer/HW/Memory] UID:33005 Activity:moderate 77%like:33018 |
8/18 Anyone have any experience getting a bootable .iso onto a USB memory key (yes, it is bootable and has enough space). -John \_ I just did this recently. I haven't found a way to get a .iso directly on, but here's what I did: 1) Format the USB storage 2) use mkbt to get the boot sector from the .iso and then put it on the USB storage. (Get mkbt at: http://www.nu2.nu/mkbt 3) copy files from .iso to memory key I used daemon-tools to mount the .iso to rip the boot sector. Oh, and if you want to use Ghost's boot disk creator, you can use a virtual floppy drive so you don't have to use a physical floppy: http://chitchat.at.infoseek.co.jp/vmware/vfd.html \_ There's not even a readme for this. What exactly does it do? \_ Readme for which? vfd is a virtual floppy. Install it and you've got a virtual floppy drive. mkbt extracts boot sectors and writes them. \_ Oh, and this is where I got most of my help on this: http://www.weethet.nl/english/hardware_bootfromusbstick.php \_ Many thanks, swami. *bows* -John |
2004/8/17-18 [Computer/HW/Memory] UID:32968 Activity:moderate |
8/17 How do I clear the URL autocomplete memory in Safari? Clear history does not work and I could not locate the files storing that info. \_ Safari will also autocomplete URLs stored in your bookmarks, not necessarily in your history. \- is there a way to be able to select the url w/ a single or double click, as opposed to the default RETAHDED triple click? \_ cmd-L |
2004/8/12 [Computer/SW/OS/OsX, Computer/HW/Memory] UID:32852 Activity:moderate |
8/11 Mac guy, did you end up getting the memory? Compusa and Fry's currently have rebates on SODIMM memory \_ I am the mac guy, jackass. Mac's don't work with cable modem. And where's my shirt. |
2004/8/9 [Computer/HW/Memory] UID:32782 Activity:moderate |
8/9 Does Mac memory ever get any cool rebates or specials like you'd find on slickdeals / dealnews, or are we forced to just take whatever price they show on macmall? My friend's ordering an ibook, and wants to know if she should add the pricy 512M now or buy it aftermarket and have me install it- the best I saw was $134 on MacMall vs $180 at http://apple.com. \_ http://www.dealmac.com but most Macs use standard PC memory anyway. \_ For example, the macmall memory I was looking at was PC2100 266Mhz SODIMM. The last deal I saw on slickdeals was for PC3100 400Mhz DIMM. Compatible? \_ you should always make sure to buy and have a professional install the most expensive parts for your mac. this is a work of art, not some ugly beige box! \_ $150 w/ student rebate at any apple store or online. \_ $150 w/ student discount at any apple store or online. make sure your friend gets the ipod rebate and teh free hp printer as well. btw, memory is REALLY easy to upgrade on an ibook. \_ I think the $180 was after the discount. \_ You're right, I was looking at memory for my iMac, not iBook. |
2004/8/9 [Computer/HW/Memory] UID:32775 Activity:insanely high |
8/9 Does the CSUA and/or the UCB nameservers cache results for far longer than the average nameserver? \_ To try to diffuse this, AFAIK the ucb nameservers use isc bind, and are RFC compliant. They cache results according to the SOA of the domain in question. \_ dig \_ I asked a "yes" or "no" question. Do you know the answer? \_ The correct answer is 'dig'. Yes, I do know. I used dig. \_ Wow, you are a delight. Why do you go out of your way to be unhelpful? You assume that somebody asking about name server cache does not know about the existance of dig? Why not use your time to be helful? \_ Because running dig to actually check the answer would have been less typing than asking on the motd. \_ again you assume incorrectly. I did run dig. \_ In that case, could you please post what you found, like "dig http://foo.com from soda shows a TTL of three days, but dig http://foo.com @bar.com shows only a half-hour; why are they different?" \_ man dig \_ Uh, of course I already did that. I'll buy a bind book cause your answers aren't. By the way, your "answers" still doesn't answer the original question, which requires experience and knowledge of the "average" nameserver. reading a man page would not help with that. \_ your question doesn't say you don't have knowledge about the "average" nameserver. you should ask something like "how long does the average name server cache results, and does soda cache longer that that?" if you don't know how to pose a question, stop bitching when people didn't give the answer you wanted. |
2004/8/5 [Computer/SW/OS/Linux, Computer/HW/Memory] UID:32702 Activity:high |
8/4 Lunar landing computer: 74kb: http://www.abc.net.au/science/moon/computer.htm \_ Does anyone else think the switches in the picture look somewhat rude? \_ Those switches serve dual purposes. Why else do you think they are installed at mouth level? |
2004/7/8-9 [Computer/HW/Memory, Computer/SW/OS/OsX] UID:32185 Activity:very high |
7/8 I have a big C++ program that uses tons of memory. What tools would for some reason I read that as mercury _/ be useful to figure out where that memory is going? What I'm imagining is something that works like Quantify, except instead of time for function & time for function + descendants, I want memory for structure & memory for structure + descendants. Context would also be useful (i.e. the call stack when the structure was first created). Any thoughts? Currently platform-agnostic. \_ turns out it's for MSVC++ 6 ... thoughts? \_ YUO = TEH SCREWED! \_ gprof (gcc) will tell you which constructors are being called frequently, which helped me track down mem usage. If you're willing to try it on KDE/Linux, check out valgrind-- I've heard it's really good for this sort of thing. (Both are free). \_ Does Purify do this? \_ Don't think so. It keeps track of leaks etc. but I don't think it does usage. \_ valgrind? \_ Linux only, and though cool, doesn't match too well with what I want. \_ If you can use Mac OS X there are a number of good tools. man malloc(3) for starters. Also look at MallocDebug.app and the other tools under /Developer/Applications/Performance Tools/ --twohey |
2004/7/8 [Computer/HW/Memory, Computer/SW] UID:32183 Activity:very high |
7/8 I looked in the archives but can't find all those computer-related euphemisms for taking a crap... anyone have the list the motd came up with? \_ http://csua.com/?entry=30029 You're welcome. \_ where's the archive? \_ The train has left the station. \_ The log has been dumped. \_ Not geekish, but "dropping a deuce" has recently become popular. \_ Going number 10 \_ I turn the toilet flusher strength up to 11. \_ You don't get it. This one goes to 1011. \_ beaming down an away team- \_ Nuking the motd. \_ flushing my cache |
2004/6/16 [Computer/HW/Memory, Politics/Foreign, Computer/SW] UID:30829 Activity:nil |
6/16 To those of you more familar with the software world than international trade in theory and practice, some of the arguments made by the anti-aumping people are self-serving comments analogous to these ... "people are being raped by paying less ... when will they realize they would be better off with out more expensive product" ... McBride calls these arguments tantamount to a death sentence for a multibillion-dollar software industry that has helped propel the United States to economic and technological leadership in the digital era. In March, he sent a letter to every member of Congress warning that Linux threatens the country's economic well-being and even its national security. "Each Open Source installation displaces or pre-empts a sale of proprietary, licensable and copyright-protected software," he said in his letter. "This means fewer jobs, less software revenue and reduced incentives for software companies to innovate." [http://csua.org/u/7s9] Also to the fellow who raised the example of memory chips, find me a case. I am not saying there has never been such a case [I dont know] but I would bet nobody has ever won a dumping case. --psb |
2004/6/11 [Computer/HW/Memory] UID:30755 Activity:high |
6/11 What's the deal with RAM ratings these days? I tried putting in PC133 in a newer machine and it wouldn't boot. If I buy the PC3200 they're advertising on slickdeals, will that work even if my computer is rated at 2400? TIA. \_ Up till around PC150-166 the number meant the MHz of the bus. When DDR came around it stopped making sense. PC2100 is a 266MHz DDR RAM. PC2400 is 300MHz. Most of the time you can use faster-rated RAM, but there are no guarantees and mixing speeds is more likely to cause a problem. Also, PC600,PC800,PC1066 are all Rambus RAMs with the number directly telling the MHz. \_ DONT TEL ME U PUT A SDR DIMM INTO A DDR D1MM SLOT!!111 |
2004/6/5-6 [Computer/HW/Memory, Computer] UID:30629 Activity:high |
6/5 What's wrong with me? I went to /. and clicked on the ThinkGeek link, and almost everything on that page I wanted to buy. /- that's not what she said! \_ You've been measured and found wanting. You're a geek and they've got your number. It all looks like completely useless crap to me. \_ oh come on, you know you want a binary Clock. And yeah, the swiss memory USB stick looks like crap but as soon as they come out with a leatherman one, i'm there. -jk \_ ok you got me, i was drooling over the binary clock. \_ I must be a nerd's nerd, because the way their binary clock displays the time seems really stupid to me. I mean you read time as "twelve fifty-seven" not "one two five seven", so why'd they make 4 banks of displays for the digits? why not two with more bits? \_ back in my day, we had one bit binary clocks that rolled over once a year, and we liked it. \_ also, I prefer reading binaries from left to right. \_ agreed. the binary clock is dumb. |
2004/6/4 [Computer/HW/Memory, Computer/HW/Drives] UID:30608 Activity:nil |
6/4 Computers might lead to autism: http://arstechnica.com/news/posts/1086361542.html http://www.smh.com.au/articles/2004/06/04/1086203610297.html |
2004/5/12 [Computer/HW/Memory, Computer/HW/Drives] UID:30180 Activity:nil |
5/12 I have 2GB of ram and a server that has 20 20MB processes. I have over a gig free but many of those processes are put on disk. Is there a way to tell Solaris to be less aggressive about putting certain processes on disk? Memory: 2.0G real, 1.5G free, 266M swap in use, 2.3G swap free 28820 root 1 50 0 12.5M 1416K sleep 11:07 0 1.60% dansguardian 28821 root 1 59 0 12.5M 1464K sleep 3:54 0 0.54% dansguardian 28822 root 1 59 0 13.5M 2216K sleep 3:43 0 0.52% dansguardian The processes used to be 50MB but I reconfigured the server to use smaller list files. BTW, this is a content filtering proxy that does a pass through to squid. |
2004/5/1 [Computer/HW/Memory, Computer/SW/OS/Windows] UID:13516 Activity:low |
4/30 My windows XP box has 512 meg memory and is only using about 300 of that, but it is still read/writing to pagefile.sys like mad. Why? \_ presumably it is caching your file accesses, and evicting your "unused" application code from ram. i have 1GB on my winxp machine, so i just turn off swap altogether. kerneltrap had an ok discussion on this topic recently. it's about linux specifically, but the general concept applies to most os's these days: http://kerneltrap.org/node/view/3000 \_ good info, tnx. Too bad i can't set swapiness on XP the way they do on linux, but can you tell me how you turned it off? \_ it's under control panel | system | advanced settings or something. |
2004/4/28 [Computer/HW/Memory, Computer/SW/Unix] UID:13432 Activity:very high |
4/28 Just got a core dump from pine - something about memory full. I would investigate further, but gots to run. \_ prolly isolated incident \_ I just got a memory full error trying to log in. tcsh failed. \_ How to you check how much free/used memory in FreeBSD? |
2004/4/20 [Computer/HW/Memory, Computer/SW/OS/Windows] UID:13290 Activity:high |
4/20 Has anybody ever been able to profile a labwindows program? I'm leaking memory and Windows handles, and consuming 99% CPU. |
2004/4/19-20 [Computer/HW/Memory] UID:13267 Activity:nil |
4/19 Anyone know a place in the bay area that will buy used RAM modules? I've got a couple of 128 and 256 SODIMMS I don't need anymore. \_ "The Used Computer Store" will give you a shiny nickel for them. You might want to try eBay. \_ Nah, I don't want to go through the trouble of ebay. There used to be places where you could sell slightly used RAM. \_ UCS: scummy scammers who only survive due to freshman/student churn. \_ Is your 128MB chip PC66? |
2004/4/19 [Computer/HW/Memory] UID:13266 Activity:nil |
4/18 Anyone add in memory to a PC lately? Do DDR memory sticks have to be symmetric on modern mbs? I seem to recall that I've added in memory in the past that was non-symmetric and it worked. Or is this one of those "it depends on the mb" questions? \_ By symmetric, do you mean "matched in capacity" or are you refering to physical symmetry. \_Matched capacity and speed of RAM(Access time). \_ I built a Pentium 4 2.8 GHz system a couple months ago using an Asus motherboard, and the manual said something about a matched DDR pair giving you two channels -- you could use just one DIMM, but then it would be one channel. Go download the manual, the mobo was P4800 or something. \_ You just have to set up the PC to work at the slowest common denominator for timings. The capacity wouldn't normally have to be the same but they probably do for dual-channel operation. \_ Well, what I did was I got this cheapie PC to run simulations on. It's an Athlon 2400 or something. It had two sticks of 128 in it. I got a stick of 512 DDR and replaced one of the sticks, the mb only had two slots for RAM. When I booted it only detected 128. If I throw out the other stick I get 512. I guess the PC is too crappy for assymetric, but it seems weird because it runs fine with one stick of RAM on it. The two original sticks of 128 do not match, btw. |
2004/2/4-5 [Computer/HW/Memory] UID:12099 Activity:nil |
2/4 Flash memory? not Compaq Flash, is it? \_ Compact Flash. Yeesh. http://www.cnn.com/2004/TECH/space/02/04/rover.halted.ap/index.html \_ Looks like a software problem, not a hardware one. --dim \_ did some engineer with a busted Nomad MUVO come up with the reformat idea |
2004/1/30-31 [Computer/HW/Memory] UID:12028 Activity:nil |
1/29 Does DDR-SDRAM run at twice the FSB? like does 333MHz DDR-SDRAM require a FSB of 166 or 333 MHz? I can't find a lucid explanation. \_ Yes. |
2004/1/16 [Health, Computer/HW/Memory] UID:11805 Activity:nil |
1/15 ecstasy and pot users read: http://www.innovations-report.com/html/reports/medicine_health/report-24750.html \_ So you heard about what happened to the last study that supposedly proved Ecstacy was bad for you, right? Completely debunked and shown to be be based not just on biased research but on completely bogus research. That's not even mentioning how bad science reporting typically is, or how its usual tendency towards simplification is amplified whenever illegal drugs are involved. This study, even from the simplistic description in the article, sounds fatally flawed. Fact is, no substance in excess is good for you. The two substances most likely to kill you, alchohol and cigarettes, are legal and taxed. The drug war is hypocrisy of the worst sort and has become a self-perpetuating industry powered by the prison lobby and other assorted interests. \_ The op isn't saying they should be banned. If you don't care whether there are negative effects--inside your own head-- from drug use, feel free to continue sticking your head in the sand. \_ For what its worth: a med student friend who has done e a few times will not do it again after learning about the physiology of its affects on the brain. \_ op here: I made no statement about anything. Why are you so incredibly defensive? If you're willing to take the risk, you are welcome to. No one will miss you later. The info is there so those not willing to take such risks may be informed of the possibilities. Everyone who smokes or drinks is 100% aware of the risks. Why would you try to hide similar information about the substances you're addicted to (psychologically if not physically)? When you first share drugs with a virgin do you really tell them the drugs are perfectly safe? I hope not. At best it is unlikely and unproven. At worst you are setting them up for a seriously fucked future in the name of pushing your own political agenda and addictions. No one *needs* 'recreational' drugs. It is a choice that should be made with all available information on the table. --op \_ this article is very poorly written. Anyone reading this really should just ignore the reporter's distinction of cannabis--> short term memory impairment and X --> long term memory impairment. Without clarifying his definitions of STM and LTM, it's pointless. It also fails to clarify possible issues with (anonymous?) internet survey methodology, and further fails to even mention cofactors, drug-interaction effects on the results, and other such issues which _need_ to be addressed. I have no clue if the researchers addressed these issues. If not, the research is shitty. If so, the reporter did a shitty job. -nivra |
2004/1/2 [Computer/HW/Memory, Computer/HW/CPU] UID:11644 Activity:high |
1/1 I got a IBM R40e recently with windows Xp. I'm wondering what I should run (if anything) to stress the computer out. I heard that if a laptop is going to break it's going to break in the first few months of use. Should I worry about this? I'd much rather ship this back in 30 days to the seller rather than deal with IBM warranty. Thanks. \_ setiathome \_ bah. recent games are still generally the things that work a computer the most. \_ In terms of actual stress testing, not really. \_ yeah goof point. seti only stresses the CPU. \_ Video editing and reencoding. \_ I had a Dell Inspiron that had an overheating GeForce GPU -- only found out after playing Flashpoint on it for a day. It also took two or three exchanges with Dell before they figured it out. \_ M$ Fight Simulator in demo mode with highest randering options. That's how I stressed a PC I bought a few years ago. Turned out that it hung after 15min or so and it was a CPU problem. \_ Prime 95 has a good stress mode, not for the video card though. It was able to detect faulty CPU operation when overclocked, even though stuff would appear to run ok. http://mersenne.org. You have to run the specific stress test. Some people run two instances. \_ is it just me or it seems like modern computers are less fault tolerant than ever? \_ It's just you. Personal computers are much more fault tolerant now than they used to be. Apparently you've never run an AppleII or a C64. People's expectations change and they forget about the recent past. It would have been ludicrous during the dawn of the PC era to think that a C64 could have uptimes of months. You were lucky if it didn't bugger out after a couple hours. You can't compare today's multipurpose PC to mainframes of the past because that's like comparing apples with oranges. The fact is that PCs have essentially replaced mainframes because they have become increasingly more fault tolerant to the point that they can be reasonably used in an enterprise market. That doesn't mean that they are as fault tolerant as a mainframe, but the fact is that you can keep a Linux/BSD x86 box running for months shows the vast improvement over time of the OS and the hardware archtiecture of PCs. \_ By "fault tolerant" was the poster asking about how often something goes wrong, or how likely the machine keeps going after something has gone wrong? \_ Not all modern computers are designed to be fault tolerant. If you have a plain Pentium IV and non-ECC RAM, you aren't getting the greatest reliability but probably enough for most people. With a Xeon and ECC RAM, you get better data integrity in both the CPU and from the RAM. Enterprise storage equipment has more. I guess data integrity isn't the same thing as the real fault tolerant stuff on servers like RAID and other redundancy, which is what you would need to tolerate disks, psus, cpus etc. going bad. |
2003/12/17 [Computer/HW/Memory, Computer/SW/OS/Linux, Computer/SW/OS/OsX] UID:11496 Activity:moderate |
12/17 Does anyone know why XP seems to run certain programs that consume lots of RAM at full speed right after a reboot, but significantly slower after a lot of uptime? And don't just tell me "obUseLinux," I would if I could, but Cubase doesn't run under Linux. I'll also add that there doesn't seem to be any difference in page file usage between "slow runs" and "fast runs." \_ Sounds like your app is just allocating more over time (mem leak?) or else having to swap stuff back in. \_ No visible memory leak that I can discern, but then the tools that M$ gives you to evaluate performance are pretty rudimentary and I wouldn't be at all surprised if stuff is slipping through the cracks. \_ memory fragmentation might be an issue too \_ If it doesn't run on Linux, it isn't worth running. Get a new app! \_ *laugh* So can you suggest a MIDI/audio sequencer that does low latency audio and runs all the latest and greatest virtual instrument plugins in Linux? Oh wait, you mean there's no such product? Oh darn. Oh well. Guess I'll just have to give up on making music then, because LINUX IS THE STANDARD! \_ obGetAMacRunningOSX. \_ Nah. Why drop 3 grand on a Mac when I could spending that money on new plugins and gear, and even a cheap new PC to run stuff on? \_ You don't need a $3k Mac to do MIDI stuff. Even a $1k 1GHz G4 can handle that sort of thing. \_ Run netstat -an to check if your PC's been hijacked. |
2003/12/11 [Computer/HW/Memory] UID:11412 Activity:nil |
12/9 \_ Thanks for the great feedback! Another DDR-related question - how does DDR MAX 2 compare to Konamix? Worth upgrading to a PS2 for? (got the PS1 just for DDR, thinking of upgrading for MAX and Karaoke Revolution) \_ around 70 songs in MAX2, freeze arrows, 60fps for the arrow movement, "light" mode (1-3 feet). songlists: http://www.ddrfreak.com/versions/listver.php -oj |
2003/12/10-11 [Recreation/Computer/Games, Computer/HW/Memory] UID:11384 Activity:nil |
12/9 Anyone have any experience with the DDR RedOctane pad? I've been happy with my Ignition 2.0 pads, but wanted to get a more affordable for my brother's kids... Thx! http://redoctane.com/ddr.html \_ i got Ignition 2.0's as well, but there are very similar BNS pads for $50. see the listing and review links here: http://www.buynshop.com/productinfo/44/VG-DDR-ULTDX \_ Are the ignition pads really worth the price? I can buy pads at EB Games for 1/3. \_ the ignition pads are *GREAT* - solid, responsive, can't complain at all. After a year of hard dancing, the foam is still as thick/soft as new. Well worth it if you're serious, but at a higher price. -op \_ My friend bought these, $39 for a foam pad: http://shop.store.yahoo.com/ddrgame/deigpad.html He's had it for a few months and it lasted way better than the $20 pads. If you look around you can get the cheap plastic ones for about $10-15, here is one: http://talkall-wireless.site.yahoo.net/playddrdanpa.html -oj \_ Every once in a great while, the motd comes through with a better collection of info than I've been able to find through much googling. I hate that. \_ Thanks for the great feedback! Another DDR-related question - how does DDR MAX 2 compare to Konamix? Worth upgrading to a PS2 for? (got the PS1 just for DDR, thinking of upgrading for MAX and Karaoke Revolution) \_ around 70 songs in MAX2, freeze arrows, 60fps for the arrow movement, "light" mode (1-3 feet). songlists: http://www.ddrfreak.com/versions/listver.php -oj |
2003/11/22-24 [Computer/HW/Memory] UID:11195 Activity:nil |
11/22 Linux RAM question, why the slight descrapancy between the available (non AGP) phsical RAM and the available (non kernel) RAM reported by the kernel messages? (these from three different computers) 1024MB-0MB(pci) = 1024MB : 1048576k - (dmesg: 1048560k) = 16k 512MB-16MB(agp) = 496MB : 507904k - (dmesg: 507840k) = 64k 96MB-0MB(pci) = 96MB : 98304k - (dmesg: 98112k) = 192k \_ Guessing: bios caching? \_ Is that the bios copying itself into RAM? \_ Just a guess, but probably. Check the bios settings for video and other caching. It's likely that if you disable all of that you'll see the ram numbers match or at least get closer. 192k is a lot on that last machine but could still all be bios caching if it's all turned on. Please report back if this is the case. I'm curious. \_ The last machine is a ThinkPad 600. I'll check into the bios when I get a chance. I'll post something. Thanks. -op |
2003/11/22 [Computer/HW/Memory] UID:11189 Activity:nil |
11/21 SAS users - I'm in SAS for Windows 8.2 I've maxed my virtual memory but it won't compute a logistic regression because it is out of memory. What the fuck? It's only about 180,000 observations in the data set. Email fab if you know an easy fix for this. |
2003/11/8 [Computer/HW/Memory, Computer/HW/Display] UID:29626 Activity:nil 74%like:10990 |
11/7 more Linux memory question 1024 MB: I ran memtest86m and it sees all 1GB, so I think that means it's not a bios issue. (I don't have an AGP card) /proc/cmdline says auto BOOT_IMAGE=Linux ro root=301 mem=1024M but meminfo (free, top etc) still say MemTotal: 901392 kB what should I try next? |
2003/11/5-6 [Computer/HW/Memory] UID:10946 Activity:nil |
11/4 How can my FSB run at 133MHz but my DDR-SDRAM run at 333MHz? My BIOS settings appear to allow this. \_ Short answer, FSB and "memory speed" are independant. The CPU doesn't "talk" directly to the memory, they both "talk" through the North Bridge. Google for North Bridge. \_ They are both valid settings, but it you use them together there might be trouble. My guess is the RAM would operate at double the FSB speed. |
2003/10/26 [Computer/HW/Memory] UID:10797 Activity:nil |
10/25 With DDR-SDRAM, what's the benefits/drawbacks of registered vs. un-registered? \_ price vs reliability \_ is registered fully supported? \_ what do you mean by fully supported? \_ If my motherboard uses DDR will registered "just work"? |
2003/9/22-24 [Computer/HW/Memory] UID:10280 Activity:nil |
9/22 Any reason why an app compiled with -pg (gprof) would try to reference mem location 0xffffffff while the same code w/o the profile switch works fine? Both compiled cleanly with -Wall & -pedantic. \_ Long shot, but I've got a machine with some bad memory at a high location. If I run multiple memory hogging apps, the last one will always crash with ugly memory errors like that. One day I'll get off my ass and replace the bad ram. \_ could you try this: http://www.memtest86.com I'm curious if if works. \_ I already did. It didn't. I know the ram is because because crashes only started happening after I swapped some old for new. |
2003/9/17 [Computer/HW/Memory] UID:10226 Activity:high |
9/16 I got a used (free) computer, but I've realized that it locks up. I've swapped out the RAM, Mobo, power supply and disk. It still crashes. (isn't really free, huh). The last time it was a seg fault in the /etc/init.d/umountfs while shutting down. The errors seems to be random, but possibly occur slightly more in the AM wee-hours. So the last thing to swap out is the CPU. It's an Athlon ~1GHz. Are these symptoms typical for a bad CPU? \_ Someone probably tried an overclock and failed. Get what you pay for. Sounds like you bought a new computer in pieces and wasted a lot of time. \_ I used to think all memory are the same. As long as you get the right one with the right cycle (e.g. 2700/3200/etc) then it will work. WRONG. I bought my GigaByte motherboard with a generic RAM and nothing worked. Tried everything. Changed CPU, changed grafx card, even changed MB and nothing worked. Went back to Frys, and had them try out the IDENTICAL RAM but this time using Kingston and other more expensive RAM, and voila, it worked. Did some research on the WEB and found out that 1) some MBs are very RAM sensitive and 2) despite what I experienced in the past 10 years, NOT ALL RAM ARE THE SAME!!!! \_ His mb is older. That's not as big a deal on the pc100 or what ever is in there. His cpu is an over-overclock victim. \_ The original mobo is a Spacewalker AK12 wit PC133 ram, but for the replacement, I bought an Asus A7V8X-X and some high-quality PC2700 DDR SDRAM from central computer. -op |
2003/9/16 [Health, Computer/HW/Memory, Politics] UID:10206 Activity:low |
9/16 You gotta love how these studies are trumpeted all over the news, but then when they are discredited and retracted its barely mentioned: http://csua.org/u/4c6 \_ And pot turns you into a werewolf \_ Oh man, no *wonder* I've been waking up naked covered in blood. \_ "We're scientists, not politicians... but we're not chemists." Great article, thank you. \_ There was no original study trumpeted anywhere but I've seen variations on this article for a week. Shut yer whining trap and go get stoned then flip my burger. \_ You missed all the hooplah about ecstasy and memory damage and the rest? Wow. \_ It's all the crack. It does bad stuff to memory. I think. \_ There was no hooplah unless you're a big ecstasy fanboy. The rest of us saw an article or two and moved on because it wasn't a bfd. \_ Sorry, wrong. This study was used to promote a very serious law and criminalized raves. It was a very big deal for a lot of people, and was covered nationwide. However, I do believe people are making a big deal of the retraction as well. \_ "for a lot of people". translation: the niche crowd raving on ecstasy. the other 99.999% didn't care the first time or anymore about the retraction. this is just the ravers being self centered and overly inner focussed. there are a shitload of laws passed everyday that have a negative impact on my life and you don't see me bitching about being oppressed everytime they pass or repeal such a law. its a party drug. it has zero beneficial medical uses when taken while dancing all night to bad music with a water bottle in hand. \_ It's also a useful psychiatric tool that's been classified with crack and heroin as having "zero medical benefits" because of a flawed study. It's important to everyone because it shows how willing government-backed scientist will be to loudly and falsely proclaim that drugs are bad. By the way, I've NEVER tried E or been to a rave. I am, however, angered by my country's screwy and self-defeating drug policy. |
2003/9/15 [Computer/HW/Laptop, Computer/HW/Memory] UID:10194 Activity:nil 53%like:10203 |
9/14 Digital camera users: how many memory cards do you have? What is their capacity? Do you have any device besides a laptop that you use to transfer memory card data? \- maybe 5 cf/microdisk. 128 to 1gb. usb1 digital wallet. why? --psb \_ my college have this. I think it's a godsend: http://www.image-tank.com/englisch/e_itg2.html Only thing i bitch about this is that it doesn't use standard AA battery. |
2003/9/3-4 [Computer/HW/Memory] UID:10057 Activity:nil |
9/3 Why does (red hat) linux suggest/require swap size to be between 1 to 2 times physical ram? With big physical ram size on modern day machine, how much swap size is really needed in linux? \_ Sun altered that schema as ram went up - 2x up to 128m, 1.5 to 256, 1x to 512, then .5 to 16G, and .35x thereafter. \_ Disk is cheap. But the fact is, with my 2G of RAM (or 512M) I don't really need swap at all. Still, I would give yourself 1-2x physical RAM just so that if you have something going nuts it'll start the machine swapping before processess start dying. That way you have some chance to kill things yourself rather than have the OS kill them for you. --PeterM \_ I heard that SunOS4 had similar requirement. It doesn't run well if you have less than that amount of swap space even if the total memory usage is less than the RAM size. \_ The idea (at least for FreeBSD) is to have enough swap space to crashdump your entire RAM and still have swap space left over. So the range is MEM+1 - 2*MEM. |
2003/9/2 [Computer/HW/Memory] UID:10039 Activity:nil |
9/1 I'm using RedHat 8 on a machine with 512 MB RAM and I only have 35 MB of RAM free according to top even with nothing running but the OS. I added up the memory for the 10 biggest processes and only got 58 MB. Does RH 8 really use around 400 MB of RAM??? \_ The rest of memory is probably used to cache filesystem access. Take a look at the output of "free" to see how much is used for cache and such. \_ Yes, this is one of those funky things about Linux. You can't trust standard tools to tell you how much memory is in use or available for other programs. You can use other things to sort of calculate backwards such as the other person says or you can look at swap in use but that's just an alternative, not better. You might consider writing a script that will parse your top output, calculate, and print the numbers you're looking for. The memory used for file cache, etc, isn't permanently lost to the system, btw. Memory will be released from cache as you load more programs. |
2003/8/29-30 [Computer/SW/Mail, Computer/HW/Memory] UID:10012 Activity:nil |
8/29 Pine core dumped on me, saying memory full. /proc and /kern are full. \_ Do we care? \_ pine eats memory by making two copies of your mail spool in memory. Get rid of your extra old mail and it will stop filling up memory. \_ Better yet, get rid of pine. \_ get rid of both! \_ best answer so far. |
2003/8/27-28 [Computer/SW, Computer/HW/Memory] UID:29485 Activity:nil |
8/27 Is there a good website that does SecureDigital memory comparison? |
2003/8/27-28 [Computer/HW/Memory, Computer/SW/Security] UID:29482 Activity:moderate |
8/27 I just gave a security presentation to a bunch of MBA students working on a market strategy for http://www.giwano.com Aside from a mildly unfortunate name, they have a cute idea, but it seems slightly gimmicky to me. While I can think of nice roles for some kind of "secure" storage like these, what's the almighty motd's opinion? -John \_ sounds like hogwash. Either the user can't get data between the two systems, or it's vulnerable to attack. -tom \_ what do you have in mind re: unfortunate name? gitano? guano? \_ Puerile, but yes. And as the PC (that's what it is) runs XP, it is vulnerable to attack--the idea is to use the flash memory between the two PC units to manually move sensitive data back and forth. It's got a built-in KVM switch to let you work on both units, so you could connect the internal unit to a 'sensitive' network and share it with PCs there. Or something. I think the idea has some merit, but they're going about it all weird. -John \_ Isn't this just reinventing sneaker net? --dim \_ I tried to figure out exactly what they're doing but wasn't willing to invest that much time doing so. Can you explain it in a few short sentences? Generally, people with important data seem happy with their current level of security. If they weren't then you'd see products from the major vendors (EMC, Hitachi, IBM, Netapp, etc) to address the issue. You don't but I wish them well anyway. |
2003/8/6-7 [Computer/SW/Languages/C_Cplusplus, Computer/HW/Memory] UID:29256 Activity:low |
8/6 In C++, if I need to keep an STL vector around (because it's a field in a class I still need) but would like to minimize its memory usage, how do I do it? resize(0), clear(), reserve(0), or something else?? Thanks. \_ none of those will deallocate already-allocated memory. use the "swap trick": assuming your existing field is f, do vector<foo>(f).swap(f). Or since you want just an empty vector apparently, vector<foo>().swap(f) will suffice. -jl |
2003/8/5-6 [Computer/HW/Memory] UID:29243 Activity:low |
8/5 I have a motherboard that supports up to AMD Athlon 3200 with PC3200 RAM. Say I put in Athlon 2500, which RAM should I get? I could get PC2100, 2700, or 3200, but if I use 3200 will it be utilized at all? \_ There's some doubt as to whether any of them take advantage of PC3200. Go to tom's hardware guide for benchmarks. \_ Tom Pabst is an egotistical dick. Try Anandtech instead. |
2003/7/25-26 [Computer/HW/Memory, Computer/HW/Drives] UID:29137 Activity:moderate |
7/25 On a related note, what's a fast CompactFlash brand? Or are they all about the same speed? Thanks. \_ The new Lexars have pretty fast write times. \_ Vikings. \_ http://www.dpreview.com/articles/mediacompare \_ Thanks! \_ According to the forums on that site, a lot of people love Transcend for their speed and reliability. I have their 1GB model, and it's worked like a charm. |
2003/7/10-11 [Computer/HW/Memory] UID:28991 Activity:kinda low |
7/9 what does " 16x16 compatability" mean in regards to memory. My laptop uses 256MB DDR SO DIMM (266 MHZ) and i want to buy another but i don't see any mention of 16x16 vs 16x8 or anything like that in my specs or the specs of my memory. (only when i go to buy..) \_ It's the density of the chips on the memory board. It isn't important as long as you match the rest of the specs. \_ like cl2.5 vis cl2 ? \_ Only to a degree. I really meant like the correct type of memory in the physical and electrical sense. After that it's all mostly the same. |
2003/6/20-21 [Computer/HW/Memory] UID:28790 Activity:moderate |
6/20 What is the cheapest PDA on the market that has reasonable features? \_ Probably M130. It has a Color screen, 8MB of memory, and an expansion slot, rechargeable battery. http://Palm.com's list price is $200 but I am sure you can buy it for a lot less from other vendors. I'd also consider Palm Zire if I didn't mind that it does not have LCD backlight and only has 2MB of memory. \_ If you're not beholden to palm, check out the Dell Axim. The 300MHz version is under $200, it's got 32MB RAM and SD as well as CF slots. And the screen is gorgeous. -former Palm owner who will never go back \_ But Dell Axim runs this OS from Redmond and, therefore, is evil by definition. Don't sell your soul to the Billie boy. |
2003/6/19 [Computer/HW/Memory, Computer/HW/Drives] UID:28767 Activity:nil |
6/18 I want to buy a cheap reliable server. It doesn't need to be fancy. It doesn't need to be blazingly fast (though it would be nice if it was.) It doesn't need insane amounts of disk or anything. What I want is it to be reliable and for it not to cost too much. Anyone have a good recomendation? Would I be stupid to go with dell or someone like that? \_ try explaining what you want to do with and how much you WANT to spend. you can find homemade stuff for $1k or you can buy something like a dell which comes with onsite warranty service.. etc. \_ it is a machine I want to have up for personal use but stuck somewhere where getting to it is hard, so if the machines goes down it will be a pain in the ass to fix. It won't be hit too hard, I mostly want it as a machine that is up 24/7 that I have full control of. \_ i.e. porn collection in the attic / basement \_ Where will you be putting it? I'm looking at colo options right now. Would be interested to know. --aaron \_ I've seen some good deals from Dell that seem to fit this. (Like a 600SC for like 399 one time with SCSI disks) \_ If you want cheap, ryo is the only way to go. Get a decent decent mb with an athlon 2400+ (or faster) along with about 512 MB DDR RAM and a couple of 80 or 100 GB ATA/100 drives. It will cheaper and faster than a Dell and just about as reliable (provided you run a decent OS on it). \_ for under 500 there are lots of random vendors on the net willing to custom build. choose good motherboard, choose a cpu based on the heat it puts out, not it's speed, choose good fans and put it in well vented place, put in slower, lower heat drives and mirror them, put in N+1 power supplies. it's sad to see the rest of you either remained silent or posted cluelessly. it should go without saying that your unused video will be on the onboard chip and don't put in more memory or anything else more than you'll actually need. my crappy very low use server runs openbsd on a p5-166 with 96 megs of ram and under 10 gigs of old crappy drives and has stayed up 24x7 since 1994 (minus a few moves or OS upgrades). you can do better and safer today with modern low-heat parts. \_ is it me, you really meant p3-166 ? |
2003/4/10-11 [Computer/HW/Memory] UID:28065 Activity:high |
4/9 Any thoughts on why yahoo messenger and MSN messenger take such a large memory footprint? (10MB/15MB, respectively) - that seems unreasonably large? \_ Have you seen all the stuff that comes with YIM? Environments, sounds, webcam, etc. I'm not surprised it's bloatware. Incidentally, ICQLite rewls... 1.3MB footprint. \_ wallall: 63K \_ walking over to the next cubicle and hold an actual conversation: 2 calories \_ what is this "conversation" and how much does it weigh? \_ you need to upgrade to tin can + string technology. \_ this is california. we drive to the next cube in our hummers. \_ No, I get a hummer in the next cube from your mom. \_ No, that's Los Angeles. In Silicon Valley, you write a long, elaborate email to the guy sitting next to you. \_ I though you im or irc'ed the guy sitting next to you. \_ of course. he stinks. you dont want to smell him. |
2003/3/25-26 [Computer/HW/Memory] UID:27837 Activity:high |
3/24 Will an Ultra2's processors and ram work in an Ultra 60? \_ any of the ultrasparc II cpu's from an ultra2 should work fine in an ultra60. all the memory should be fine as well. \_ Depends on the Ultra60, I think. Ultra60 is not descriptive enough. U60 spanned a long period of time with a lot of changes. Better question is: WHY?! Trying to scavenge parts? --dim \_ I managed to pick up an U60 chassis (no proc/ram) for free and I've got some bits from an U2 (proc/ram) along with some PCI hme cards. If the U2 bits work in the U60 I was planning on install OpenBSD on it and using it as I was planning on install OpenBSD on it and use it as my server/router/firewall. \_ you're sick. stop that immediately. \_ The memory should be compatible. I am not so sure about CPUs. What kind of CPUs does your Ultra 2 have? The 300MHz CPU used in both machines has the same part#, so it should be compatible. I am not so sure about the 400MHz CPUs that some Ultra 2s used. The UltraSPARC I CPUs (200MHz or slower) that used to come with older Ultra2s are definitely not compatible with Ultra 60. \_ Only the 300MHz are compatible btw the two. All others are NOT: 400,200,167MHz work for U2; 360,450 work for U60 This is from the Sun FE Handbook which is online as the Sun System Handbook linked from http://sunsolve.sun.com Sun is frigging badass when it comes to this shit. \_ True, but up until a year ago, the FE Handbook costed ~$1000. http://sunsolve.Sun.COM/handbook_pub/Systems/U2/components.html http://sunsolve.Sun.COM/handbook_pub/Systems/U60/components.html \_ Thanks. |
2003/3/18 [Computer/HW/Memory] UID:27734 Activity:nil |
3/18 I am going to buy some RAM (for a powerbook G4 12). Havn't shopped for RAM for a while, I looked around on websites and noticed some module manufacturers such as Kingston offer "ValueRAM" (order by spec) and system specific ones. There is a large price difference between the two, but try as I do, I cannot think of a reason a config. specific module is better than a "Value RAM" rated for the same spec, unless the system manufacturer has some secret spec that they only disclose to Kingston. Are there really differences between them other than price? |
2003/2/20-21 [Computer/HW/Laptop, Computer/HW/Memory, Computer/HW/CPU] UID:27470 Activity:high |
l2/20 http://info.lindows.com/mobilepc/mobilepc.htm Do any of you have one of these? \_ They're not being sold yet you idiot. \_ http://csua.org/u/99b \_ They should really put the Fujitsu S series in the comparison. \_ looks kinda cool, I'd be worried about the display though, and does anyone know how the processor is? |
2003/1/14 [Computer/SW/Languages/Java, Computer/HW/Memory] UID:27090 Activity:nil |
1/13 I've been looking at intel's web site and can't find any sort of white paper on design principles that make the CPU keep up with moore's law. I heard that things like the main components of performance is memory and manufacturing process. The fancy stuff like pipelining hyperthreading etc, doesn't really do much. The amount of onboard cache memory is still the determining factor in performance. And that is tied to the manufacturing process. It has been a while since I took CS152. Is this still true? Most of the advances in microprocessor design is in integrating bigger and bigger cache into the chip? \_ Making transistors smaller also helps. Smaller devices run faster than larger ones since both the resistance and the capacitance depend on the device size. \_ given practical limits on size and cost of a die, more cache implies process improvement. the pa guys used to do offchip sram in a mcm, but i think they've given up on that also (because at the end of the day the interconnect between cpu and sram is slow). (of course i've not looked at general purpose cpu's for many years so i may be wrong). intel's claim of everything being essentially only memory and cache bandwidth limited may be true for the class of cpu's they mostly build (general purpose) and the class of problems they solve (large grained), but certainly is not true for all architectures and all applications. also i guess depends on the definition of a cpu. e.g. just as the bandwidth of a truck full of mag tapes is stunning, so is the computational power of a die full of small (friend of mine makes a 27k gate one) processors. so in general they are wrong, though they are probably correct specifically for the problem they are trying to solve. \_ well, my wrong is probably too strong, because it's always memory and cache bandwidth, but it's certainly not the only thing. \_ Moore's law specifies #transistors/chip. The fabrication process allows features to shrink roughly in line with ML. Lately, issues like scaling (wires scale differently than transistors) and routing (connecting the transistors to each other) are tougher than shrinking transistors. \_ nothing lately about it. i took cs250 15 years ago, and it was obvious to me then that routing was the problem. power is a much more interesting and recent problem. \_ When I worked developing CAD tools @ Intel 1997-2000, those were the emerging problems. Hence, lately. -emarkp \_ back in the day when i banged sea-of-gates chips late 80's and early 90's, it was as simple as throwing the netlist over the wall to the backend guys. 93-ish we started having to worry about floorplanning, and 95 i started doing cot and p&r was a problem. which is not to say that p&r was not a problem earlier, as anyone who pushed polygons by hand will tell you (which is what i referred to when i mentioned c250 above), rather that around that time density and technology made the problem much less tractable. you also have to understand that intel does not have the most normal design flow, and your experience at intel probably does not reflect industry experience in general. |
2003/1/7-8 [Recreation/Computer/Games, Computer/HW/Memory] UID:27018 Activity:high |
1/7 Does saving PS1 game data on PS2 require a PS1 memory card? \_ yes. --jwang@playstation I STFWed and found many conflicting answers. Anyone has first-hand experience on this? \_ Ask Sony. \_ Yes. Because PS1 memory cards use blocks and PS2's don't, I think. -geordan \_ I'll try it tonight and post or email the results. --peterl |
2002/12/30 [Computer/SW/Apps/Media, Computer/HW/Memory, Computer/HW/Drives] UID:26942 Activity:nil |
12/28 wanted, dongle that speaks USB and firewire for the purpose of reading and writing to the following media: CF-1, CF-2, MicroDrive, SmartMedia, SD/MMC, and Memory Stick. \_ Why do men want what they can't have? \_ Because he's too stupid to realize he can't have it. \_ I want world peace. |
2002/12/5-6 [Computer/SW/Unix, Computer/HW/Memory] UID:26726 Activity:high |
12/5 We have a Sun E4000 that was upgraded to Solaris9. At some point "soon" after the upgrade the machine began to semi-crash each day right after midnight. You still had a console prompt, but it loses its nis binding/domainname, loses the default route [cant ping it from another subnet], and all the daemons stop listening. Any ideas? \_ do you have all the latest Sol9 cluster patches installed? \_ did you turn off power management? \_ What would that do? What should I look at? does something magic happen at midnight? \_ This would be my first guess too. Take a look in /etc/rc*.d for S85power or something like it. if it's there, take it out and run /etc/init.d/power stop. It could be trying to drop to a different init level on some odd schedule. kill vold while you're at it. grr... --scotsman \_ What's wrong with vold? \_ Its a pos. \_ It is but it won't kill his computer like this. \_ Yeah, but you might as well turn it off and dtlogin. \_ gasp! you'd disable the world's best GUI?! \_ never had problems with it but the dtlogin should certainly go on a server. \_ You have a bad cron job that runs at 00:00? \_ There are some accounting-related cron jobs, but I cant think of a "normal userland process" that can cause all of this. \_ Maybe it can if it runs as root. I don't know. \_ root's cron jobs run as root and thus can fuck up the entire system the same as root @ the keyboard. \_ Let me put it another way: There was no new cronjob added after the upgrade. Maybe some file isnt being found but something like that cant cause you to lose your default route. I'll try checking the run level via who -r before rebooting it again. \_ We had something like this. It was system accounting runnning the machine out of memory. Sometimes it rebooted, sometimes it hung as described. No memory == can't fork new proceses from inetd == no telnet, ftp, etc. -ax \_ but non-inetd stuff dies, like the domain binding and default route being lost. \_ ooh.. Good one. --scotsman \_ system accounting and /var/adm/messages are for weenies. Real men read the pretty lights and feel the hum. |
2002/11/25-27 [Computer/HW/Memory] UID:26629 Activity:moderate |
11/25 Which of these rechargable batteries have memory effect: NiCd, NiMH, Li-ion? Thanks. --- yuen \_ NiCd has major memory effects \_ what's the physical/chemical basis of this memory effect? \_ obGoogle \_ none. modern batteries do not have memory effects. however, the term has been applied to other phenomena, such as voltage drop, which is found in current nicad batteries |
2002/11/8-9 [Computer/HW/Laptop, Computer/HW/Memory] UID:26475 Activity:kinda low |
11/7 hola, will going from 512mb -> 1gb of memory in my laptop significantly affect battery life? 5% less? -10%? --psb \_ Don't forget that DRAM refresh cycles also consume energy. \_ it will most likely increase battery life if it reduces swapping. if it doesn't, what's the point of upgrading? \_ what if 512 MB is insufficient only 5% of the time? \_ Depends on what you call significant. If you're on the wire most of the time, no. If you want to do serious traveling then 512mb is already too much. \- re: memory energy consumption [and attendant cooling demands] is why i asked. in re: swap/no swap ... gee maybe i need the memory to avoid heavy swapping when plugged in and doing a lot of photoediting ... and i dont want to take the extra memory out before going "unplugged". and the "depends what you call sig" person isnt saying anything at all. does anyone actually have a sense if memory is like 5% of total system power demand or more ... vs. lcd, and cpu. --psb \_ Dude, look up the power consumption needs of your chips and do the math. No one is giving you solid numbers because they will vary dramatically by system. I have a sense you don't even know what you're trying to ask. \_ I have a 600MHz iBook with 640MB of RAM. It runs for 4 hours on battery easily. Decreasing LCD brightness tends to increse the run time by 25% or so. Does anyone have a similar model with smaller amount of memory? How much run time do you get on battery power? \- usually a little under 3hrs. i usually run the screen pretty bright. 800mhz titanium with 512mb. i am guessing the memory wont be a big deal then. --psb \_ Let us know after you shell out a few bucks since you're unwilling to use google and a 4 function calculator. |
2002/10/29-30 [Computer/SW/Languages, Computer/HW/Memory, Computer/Domains] UID:26352 Activity:nil |
10/28 What is the register file made up? Same as L1 (e.g. 6 transistors)? \_ It is made up a few hundred years after the Grimm Brother's stuff. \_ depends on architecture. Usually sram array supporting dual/quad port read/write. |
2002/10/16-17 [Computer/SW/Languages/C_Cplusplus, Computer/HW/Memory] UID:26215 Activity:moderate |
10/16 In C, What's the difference between offset_t and ptrdiff_t? I used to use size_t for variables representing offsets, but now I think size_t might not be right. Thanks. --- yuen \_ I don't know for certain, but here's my thinking: ptrdiff_t is signed and should be sufficient to store the displacement between two items of the same type. size_t (produced by sizeof) is unsigned and should be sufficient to store the size of a type. offset_t (produced by offsetof) is unsigned (I think) and should be sufficient to store the offset of an element within a type. Therefore, aside from the sign differences, the range of ptrdiff_t may be greater than that of size_t which \_ I take this back. It makes no sense. (Not that the rest does.) --jameslin may be greater than that of offset_t. Maybe you should ask on comp.lang.c. --jameslin \_ size_t is usually 32 bits, and is used to represent the size of objects or buffers and the like, off_t is used to represent offsets in files and is often 64bits. Why does this matter though? Outside the interfaces these are used in, there is no special meaning to them. \_ I see. off_t (and offset_t on SunOS5) is for file offsets, not memory offsets. So I should use ptrdiff_t for memory offsets then. Thanks. --- yuen |
2002/9/25-26 [Computer/HW/Memory, Computer/SW] UID:26000 Activity:low |
9/25 How do I tell browsers to cache my external style sheet? I'm using <link href="my.css" ...> to link to the css file. \_ won't most browsers cache it on their own? |
2002/9/13 [Computer/HW/Memory, Computer/SW] UID:25873 Activity:high |
9/12 People familar with Sun workstation: does anyone know what this object is/is for: /tmp/sunpart-large.jpg, /tmp/sunpart.jpg --psb \_ why do ppl post pics in tmp instead of on their web page? \- i have no WEEB page. \_ copy it to ~psb/public_html , then fuck off! \_ sunpart.jpg is a GIF file. Anyways, I think it's a torque wrench for riser boards found in models like the U80. \_ I assume the green thing is packing material or background for contrast. The silver tool does look like a memory riser board torque"r". I haven't used one in years. -ax \_ Looks to be a bit of packing or the handle for a board/enclosure. \_ I can't view the image from here. If you have a part number you can look it up in their online hardware manuals at <DEAD>support.sun.com<DEAD> |
2002/9/8 [Computer/HW/Memory, Computer/HW/CPU] UID:25805 Activity:nil |
9/7 Motherboards have a spec called FSB, what is the significance of that? \_ ObYermomboard \_ front side bus, like anything else on a mobo, the faster the better \_ What does it connect to? Memory? Peripherals? \_ It's the bus between the CPU and memory. For future reference: http://www.whatis.com |
2002/7/14 [Computer/HW/Memory] UID:25355 Activity:nil |
7/14 What's a good/cheap motherboard that supports 168 pinn (PC133) and 184 pinn (DDR) ram? I need a slight upgrade to play Warcraft 3 but don't want to buy new RAM. |
2002/3/21-22 [Computer/HW/Memory] UID:24183 Activity:high |
3/21 I just bought no-name-brand memory at frys. Apart from passing a bios check, how do I more rigorously check the memory? \_ Run QAPlus? I don't know how good it works on today's PCs. \_ boot a *nix, and compile and run ~nweaver/mem{test,test2}.c |
2002/1/12-13 [Computer/SW/OS/Windows, Computer/HW/Memory] UID:23540 Activity:high |
1/11 I just installed 256megs of memory on my Dell desktop, bringing it up to a total of 384 megs. Everything seems fine, except one little thing. When I go to the start menu and click "New Office Document" and then select Blank Document, for example, it doesn't do anything...until I do something like just opening up a folder, or another program, or pretty much anything that requires double- clicking. So the new document does pop up, but only after I open something else. I really don't remember if it did this before, but could this be related to the new memory (DIMM, by the way). And it's Win98. Thanks. \_ have you popped the memory out and verified the behavior? \_ You know about windoze wasting heap trying to cache too much, right? Put [vcache]\nMaxFileCache=65536 or something reasonable in system.ini. Not that this is your problem. As for the actual problem, hmm. Maybe it's that you're running windoze, MS office, or the combination of both. hehe \_ It's not smart enough to shrink the cache whe heap space is \_ It's not smart enough to shrink the cache when heap space is \_ No, that's not the problem. As I understand it, it pre- allocates heap for handling the file cache. If you have more than 512MB of RAM, the amount of heap needed for maintaining that much vcache is almost 128k, which is the fixed amount of heap windoze 95/98/ME has. When that happens, you get a out-of-memory error even though you have more memory than ever. It does not affect W2K/NT. running short? Which versions of Windoze have this problem? \_ Well, kind of.... As I understand it, it pre-allocates heap for handling the file cache. If you have around 512MB of RAM, the amount of heap needed for maintaining that much vcache is almost 128k, which is the fixed amount of heap windoze 95/98/ME has. When that happens, you get a out-of-memory error even though you have more memory than ever. It does not affect W2K/NT. \_ If you want to run windows with anything more than 32 or 64 MB of RAM, you should use win2k or NT, only. \_ dude, you got a dell! \_ Hmm. I didn't consider that, but then, it's still better than compaq. \_ Or as we affectionally call it at work, crapaq |
2002/1/8 [Computer/HW/Memory, Computer/SW/Unix] UID:23495 Activity:nil |
1/7 Would someone please point me to a reference explaining memory usage in unix. Specifically, what is "cache"d memory and "buff"ered mem? |
2001/12/18-19 [Computer/HW/Memory, Computer/SW/OS/Windows] UID:23296 Activity:high |
12/18 I just got a BSOD in my NT and it said it's dumping physical memory. Where can I find the file so that I can delete it after I reboot? Thx. \_ look online! |
2001/12/17-19 [Computer/HW/Memory] UID:23275 Activity:nil 50%like:22238 |
12/17 What's a cheap place to buy 64/128MB memory sticks? \_ http://www.pricewatch.com or http://www.pricescan.com but their interface sucks. \_ Hot Hot Deals http://forums.anandtech.com/categories.cfm?catid=40 |
2001/11/29 [Computer/HW/Memory] UID:23140 Activity:nil |
11/28 Has the computer architecture research been so degraded to a point that we're more interested in finding out how big the L1/L2 cache size should be, instead of finding doing cool things like CISC->RISC-> ???? \_ Has the computer architecture research been so degraded to a point that we're more interested in finding doing cool things like if instruction set X >> instruction set Y instead of finding out how big the L1/L2 cache size should be? \_ Well, they still haven't successfully designed a chip to make my penis grow bigger. So yeah, it's degraded. \_ <DEAD>www.longtitudecapsures.com<DEAD> |
2001/9/20 [Computer/HW/Memory] UID:22552 Activity:nil |
9/20 I recently bought some cheap 512 MB DIMMs for my system and I sometimes I get the message "Invalid number of bits in the column address" during boot up. What does this mean? Is this bad ram? Do I need to get the more expensive ram (I have unregistered)? |
2001/6/21-22 [Computer/HW/Memory, Computer/SW/OS/VM] UID:21582 Activity:moderate |
6/20 I have an OS question. Most programs make use of shared libraries and use a dynamic link loader to relocate branch targets at run-time. The problem is that if you have multiple copies of, say, emacs running under their own user space wouldn't those relocated addresses (either relative or absolute) conflict under each process? For example, if multiple programs were using OpenGL they would create virtual memory entries for libGL.so in their own process space. The location of libGL within this process differs from program to program so if libGL calls libm.so the dynamic link loader will place that library in a different location for each process and the branch targets under one process won't correspond to those of the other. \_ The shared bits are mapped to the same virtual addresses in each address space. \_ Not sure what you're asking. But only the text portion of the shared lib is shared amongst the different processes. All processes that need to use libGL.so have addresses that point to just one copy of the text portion of libGL.so. It's the OS's job to keep the program counter and the VM straight. Why would there be any conflict? \_ That's the problem. If you only have one copy of libGL.so in memory there would be a conflict in the outgoing branches from libGL. For example, if we were calling glVertex within libGl. That function would have a jump an link to another absolute address to the libm math library. The dynamic link loader is responsible for resolving the branch addresses. The problem is that if one user was running Quake while another person was running Doom or something like that then the jump target addresses would be different. Let's just say, for example, that glVertex called the pow function: Quake Doom ----- ---- 0x00000000 main 0x00000000 main 0x00003fff end of quake 0x00001fff end of doom 0x00004000 libGL 0x00002000 libGL 0x00005000 libm 0x00003000 libm 0x00005040 pow() 0x00003040 pow() If we were to have a jump and link to the pow() function then the addresses would be different in both copies of libGL. \_ Also, most, if not all, shared libraries are compiled to be position-independent. |
2001/4/18 [Computer/HW/Memory, Computer/HW/Drives] UID:21013 Activity:very high |
4/17 Now that those rat bastards at VALinux are getting out of the desktop market who is a reasonable alternative as a supplier? IBM? HP? Dell? Someone else? --dim \_ what was good about VALinux? there's penguincomputing. I don't know why you'd care, just install Linux on anything. \_ Just say no. If you must use PC hardware, at least install FreeBSD. \ \_ and deal with freeBSD? fuck that. \_ Deal with what? You're either a troll, ignorant, or simply deeply confused. what's so bad about PC hardware? it's cheap and fast _) hey, this is a good troll. \_ BIOS, crappy memory bandwidth, floppy disks, IDE drives, 32bit PCI, ISA, bad power supplies, awful SMP performance. I don't think that I need to go on. \_ You do realize you can buy a PC with SCSI and PCI64, right? SMP performance is partly a function of the OS. Have you seen what kind of crap Sun is shipping these day? Their low end boxes are IDE! --dim \_ Sun is shipping this bottom-end server crap, the 280R: redundant hot-swap power supplies with independent power cords, hot-plug power subsystems, fibre-channel disk drives, backplane has software-mirrored hot-plug boot drives. 8MB cache. 8GB RAM. lights-out management card. Free solaris. Now hook it up to the T3 external hardware RAID array, GBIC cards, dual-ported 10K RPM FC-AL disk. Max. multimode optical fiber length of 500 meters. Two redundant loop cards per enclosure FC-AL circuitry Two Power/cooling units per enclosure, Integrated backup battery power, Redundant fans, Battery backup for cache destage. Free Veritas. Yeah I agree its shitty product. But your cheap PC will always be its bitch. You can mass your million-man Red Army, but their few precision jet bombers will blow you to bits. So what if their products "suck", I'd like to take a poll: if SUDDENLY TODAY, someone gave you choice of a FREE $1000 PC or a FREE $1000 SunBlade, which one would you take? \_ VA Linux sales were good at calling over and over and over like clock work. I always knew it was the first thursday because that's when VA Linux sales would call. \_ the SunBlade is a piece of shit. The 280R, configured with 8GB of RAM and a T3 RAID array, costs $90K, hardly "bottom-end". -tom \_ You are nothing more than a shill for Sun. -ausman \_ No, that's the top-end E280R configuration. I just priced a $10,000 one on their site (one CPU, one disk, etc..) \_ A $1000 PC is a POS. A Sun Blade is not. I'd take the SunBlade over even a $10K PC. \_ The 280R *is* the bottom-end of the Sun server line. It's a workgroup server. Max two CPUs only. \_ are you stupid or just lying? The bottom-end server is the Netra X1, which costs an order of magnitude less than the 280R, and is a piece of shit. -tom \_ are you comparing PC servers to desktop machines? --jon \_ Ultra 2, 2 300 MHz US2 Procs, 1GB Ram, Creator 3D now that is a Desktop Machine (~ 2 - 3K now). Or an Ultra 60. I prefer real hardware to some cheap broken PC crap. \_ An Ultra 2 is a piece of shit these days. --dim \_ Really? I guess that the 2 GB of RAM it supports along with the 2 300 MHz US2 procs are just no good compared to your OC'ed Celli 933s. I'm so sorry that your IO bandwidth and your memory bus speed SUX, as does your SMP bus. But I'll bet your GeForce2MX make up for that in your "real world" applications. \_ PowerMac G4 w/2x533 G4e's, 1 GB Ram, Ultra 160 drives, GEForce2 (or 3). Now that is a desktop machine. \_ P5-166, 96 megs RAM, 3 EIDE disks: 1.2 GB, 2.0 GB, 4.3 GB, Matrox Mystique (the first one), OpenBSD, open air case, 300 watt PS, 1.44 diskette, 2 serial, 1 par, dual 10/100. Rock solid! Top that, kids! \_ SparcStation 10, 1 SM61 proc, 272 MB Ram, 2 Barracudas, QFE, no framebuffer (serial console only), OpenBSD CURRENT. \_ p5-166? \_ Dell makes rather nice Linux boxen. Don't expect their Linux (software) support to be anywhere as good as VA Linux though. \_ Are you joking? My research group ordered some computers from Dell and they SUCK. We ordered the plane vanilla machines with Linux pre-installed and Dell installed unsupported video and network cards then dragged their feet when we asked them to fix there mistake. I will never again by Dell. \_ VA has support? Surely you jest. \_ VA Linux sales were good at calling over and over and over like clock work. I always knew it was the first thursday because that's when VA Linux sales would call. \_ I said support. |
2001/4/18 [Computer/HW/Memory, Computer/HW/CPU] UID:21010 Activity:very high |
4/17 I'm between jobs and want to build a fast Athlon Thunderbird box. Any suggestions? I'm not sure about which motherboard to get (KT133A?), what the diff is with the T-bird 200 and 266 MHz bus versions, and whether PC133 or DDR memory is what I need. Thanks! \_ What does being between jobs add to your query? Does that mean you want one on the cheap? --dim \_ stop being impertinent. \_ Go to http://www.anandtech.com Lots of info there. They just did a review of KT266 vs. all the other DDR 266 MBs out there. The forums are of great help too. \_ Thanks. The message I took home was: "If anything, you should still stay away from the KT266, the platform is simply not mature enough." So I guess I'm going with a KT133A chipset, probably an Asus. After reading a # of articles, I'm still not sure about whether a 200 or 266 is what I need, and whether the KT133A supports DDR. \_ Um. There are othe choices than those two. Go with an Asus A7M266; it's gotten good reviews, wins all of the benchmarks, and isn't based on that chipset. \_ What is "those two"? Thanks. \_ Currently there are 3 chipsets: AMD 760 Ali Magik something or another Via KT266 \_ KT133A? KT133? Are these older? \_ Yup. \_ The two stated: KT133A (old tech) and the KT266 (unstable) \_ Is there a winning combo for Athlons? For Intels, you get a Pentium 3 866 with PC133 CAS2 memory on a CUSL2-C and whatever video card you want. (since the Pentium 4's blow, currently, and the higher-speed Pentium 3's have heat problems, to say the least) \_ just one word about the KT133A, dont get any orb-type fan with it it'll bash and cut the capacitors situated right next to the CPU \_ I think this fall will be a much better time to buy, fwiw. The past six months have not brought enough advancement to be worthwhile. Laptops will get more choices by then too, with amd and ati getting new chips out. \_ "later" is always a better time. \_ You keep talking about this filling some need but never say what your need is. Is this a game box, a cpu cruncher, a web server, or just a random toy because you're bored? http://www.tomshardware.com is a good place to start for opinion/info/benchmarks. \_ Get the 266 FSB. This means that the cpu bus is running at 133MHz (or DDR if you have DDR RAM), and the mult is much lower. The 133 bus wins all around over a 100mhz bus, imo. Also, it would behoove you to get a system that can eat the $50 256M DIMMS. This memory can only be used by certain chipsets. I have a 1000/266 TBird with $512M of the cheap RAM. It's on an ASUS A7V133. --sowings \_ jee-zuz. That will get you, what, 2500 terabytes of RAM? \_ All of you giving advice don't have a basis for it. You don't know what he wants it for. |
2001/3/14-15 [Computer/HW/Memory] UID:20780 Activity:kinda low |
3/14 Which would be better for home/home-office, a laser printer (10 ppm, 1200x1200) or a high-end inkjet (1.5-15 ppm, up to 1200x600 four color)? Both come with PostScript and Ethernet for about $1K... \_ I like that laser printed pages don't smear ink. If you mostly print b&w, go for the laser, and get a cheapo ink jet for the occasional color print jobs. \_ I have a cheapo inkjet, but it's annoying to use a serial cable from my laptop. Oh, other factor: laser printer has 8 mb, expandable to 40 with 100-pin EDO DIMM ($58 at Chip Merchant); inkjet has 24 mb, expandable to 88 w/144-pin SO-DIMM PC-100 SDRAM ($32). \_ I think color laser gives sharper text and graphics (e.g. charts in presentations), but inkjet gives better quality with photos because it has more shades of color. -- yuen |
2001/2/27 [Computer/HW/Memory, Computer/SW/Languages/C_Cplusplus] UID:20713 Activity:high |
2/26 Hey folks, I'm looking for something like Purify but cheap. What do people do for C programming to guard against memory bashing and leaks, other than write their own malloc and buy expensive software from Rational? \_ wasn't there some kind of program to analyze C code for no no's. I think it was called something like lint. \_ uh, lint is a syntax checker. it does not figure out if your code has memory leaks. \_ memory bashing. memory leaks. \- wasnt there something called electric fence or something like that? --psb \_ Yes. -lefence on some Linux systems, or you can go and get the library. However, efence just puts illegal pages before and after every memory allocation. This means that if you write outside of the bounds of any allocation, it seg faults immediately there, showing you the memory error immediately if you're using a debugger. This is VERY demanding of memory however, and slow. Memory use will 10x, speed will .04x. efence is in NO way as good as purify.... --PeterM |
2001/2/9 [Computer/HW/Memory] UID:20546 Activity:kinda low |
2.8 box can run before running out? \_ Run out of what? If you mean memory, there is no simple answer - it depends on how much RAM & swap you have and how much each process uses. -alan- \_ 2.8 same as pre-2.8 (except ptys are dynamically alloc'd) Examine with: sysdef -i or # adb -k /dev/ksyms /dev/mem parameter-name/D ^D (to exit) or # /etc/crash > od -d parameter-name > var where parameter-name is: maxusers, maxnprocs, or maxnuprc 2.8 can do >30K processes by modifying several params(bet its buggy) Put set pidmax=999999 in /etc/system modify pre-2.8 formulas: max_nprocs = ( 10 + 16 * maxusers ) ==> procs system-wide maxuprc = ( max_nprocs - 5 ) ==> procs per user & 5 reserved for root but can be modified. See http://docs.sun.com for real details -- I'm just guessing. |
2001/1/2-3 [Computer/HW/Memory, Computer/SW/OS/Windows] UID:20220 Activity:kinda low |
1/2 How do I increase the environment memory size for CMD.EXE (not http://command.com) on NT? Thanks. -- yuen \_ What's the difference between the two? \_ I don't know. -- yuen |
2000/11/1 [Computer/HW/Memory, Computer/HW/CPU] UID:19624 Activity:low |
10/31 When one CPU on a dual Pentium II machine executes an instruction that writes to memory, and later the other CPU executes an instruction that reads the same memory, how does the second CPU get the correct value from the first CPU's cache? Thanks. \_ Many cheepo/low end systems use a cache coherent protocol called MESI (modify, exclusive, shared, invalid). This works fine under smaller systems that can do bus snooping. But for much larger systems (like 512 processor systems) they use a more sophisticated directory lookup structure. -jeff \_ For more information about MESI on intel processors, see http://developer.intel.com/design/pentiumii/manuals/24319202.pdf chapter 9, especially section 9.4. \_ Distinction between cache and system memory. Dirty|clean bits. \_ Do you mean the two CPUs share the same caches instead of having separate caches? \_ No, each Pentium has its own cache. \_ They probably run in separate processes, separate address spaces. \_ You're probably a total fucking idiot. \_ I believe that Hennessey and Patterson (not to be confused with Patterson and Hennessey) discusses multiprocessor machines. You you might want to check it out for a detailed explantion. -dans \_ Just an aside, that naming convention is only used in Berkeley. - current grad student somewhere else \_ Uh, that is the author order on the books themselves. Although I think Culler's parallel architectures book does a better job discussing cache coherance. -nweaver \_ Thanks for all the answers. I had a lousy prof for 152 (last name started with R and was very long, don't remember exactly), and it has been 7 years. |
2000/7/27-29 [Computer/SW/Languages/C_Cplusplus, Computer/HW/Memory, Computer/SW/Unix] UID:18795 Activity:very high |
7/27 Where can I find an explanation of what usually causes a core dump, bus error etc. ? \_ core dump = just that. core memory used to me a type of volatile storage which died out quick but the term still lives on (hence \_ As opposed to current storage which dies out quick? Get your story correct or don't try to sound leet. \_ As opposed to current solid-state memory technology which has been used for decades. Get a clue before you flame. The point, which you missed utterly, is that "a type... _/ which died out quick" also describes modern memory, and neither distinguishes between core and solid-state nor accurately distinguishes between core and solid-state memory nor accurately explains why it was called core in the first place. Twink. \_ modern memory technology has not died out slowly or quickly it's still used all over the place. No one has used core memory in decades. It doesn't explain why it was called core, but does distinguish a short-lived, long-dead technology from a long-lived, still-used technology \_ Why are you talking about the technology and not the storage itself? [a place for twinks on the web: http://www.twinkparadise.com ]_/ core dump, out of core, core map, etc...) core dumps are usually a result of an illegal operation and can be enabled and disabled. bus error = i think means misaligned address (obviously illegal). segmentation fault = there are only certain segments a user program can read and write from. these access bits are usually written to the TLB and automatically cause an exception when you access a segment in a way you're not supposed to. \_ Get this book, it kicks ass: Peter van der Linden, Expert C Programming. ISBN 0131774298 \_ you mean you want something more to know about getting a bus error because you are accessing memory which is "not valid" (note the quotes). \_ I was looking for an answer like "This happens when you dont allocate enough space to an array, or you look past the end of and array" etc. \_ accessing null pointers, accessing memory out of bounds (reading past allocated memory), etc. Its good idea to check a pointer's validity before accessing it (like a->foo()). \_ no, you can get core dumps from lots of uncatched exceptoins. i don't have to deref NULL or an address outside my addressspace to get a SIGABRT for example (which coredumps). \_ true but the original poster probably wanted to know common reasons. \_ Yes, Thank You. Now what is a Bus Error? \_ As explained above, a bus error is caused by attempting to read an address/size your memory bus considers illegal, such as a 32-bit word at an odd address. Usually caused by utter garbage in your pointers, due either to not initializing it or overwriting it with other data. |
2000/7/12-14 [Computer/HW/Memory, Computer/SW/OS/Solaris] UID:18649 Activity:kinda low |
7/11 On Solaris is there an xhost FILE where i can put a list of computers that can open X displays. (I need 'nobody' to be able to display to this SUN because of a lame ass java package (jchart) which my company paid for). \_ for h in `cat FILE`; do xhost +$h; done \_ umm, no. this is session specific. I need NOBODY to be able to display there. \_ read the answer again, or provide more details \_ /etc/X0.hosts \_ doesn't seem to work on Solaris 7 (i HUPd Xsun and no change to "xauth -list" \_ "xauth -list" lists your keys, "xhost" lists who is allowed to connect to the server \_ Use Xvfb instead. \_ hmm, o.k. except that requires actual work and learning and stuff. (and how bad is the resource (memory) hit? |
2000/7/7-9 [Computer/HW/Memory] UID:18610 Activity:high |
7/7 LDAP question. It seems oriented mostly for reads and single adds. Does that make it a poor decision to do things like act as a service discovery server for something like Ninja/Jini? What's "normally" used? Small database or standard data structure in memory? \_ I'm not sure I understand the question, but since LDAP is just a format it seems like you could implement your own directory with LDAP and have it support the discovery/lookup multicast protocol... the actual registering of names in that case would be done in LDAP. But since Jini groups must be implemented by using serialized objects it would be a waste of LDAP's functionality. Sun's implementation of the naming directory, "reggie," is based on rmid, which is running on JRMP or IIOP and uses a propreitary naming protocol.... if you're implementing all your stubs yourself, go for it. reggie puts things in memory in a data structure... \_ I don't see what it has to do with singles ads. \_ i am a ninja! bow down to the almighty bunghole! does that answer your question? \_ SLP \_ implement an LDAP server using Oracle \_ Can you implement an LDAP server using ED? \_ Real men use echo. ED is for whimps. \_ Real men use echo. ED is for wimps. Show-offs use a magnet to twidle the bits on the drive. \_ Real Men do not twidle their bits, and not on their hardware. \_ Apologies. My question was really about whether or not LDAP is poorly suited for an environment (specifically service discovery) in which many changes need to be made to te dataset. \_ Yes. Jini will probably make use of Javaspaces which uses serialization and rmi-style registries |
2000/7/5-6 [Computer/HW/Memory, Computer/HW/CPU] UID:18586 Activity:moderate |
7/4 What is the deal with the cheaper (not slot 1) P3 say in a Dell 667. How do they compare to the reg slot 1 type? Are they socket 7? Are they more error prone/slower what? Why did intell go backwards in their design. \_ FC-PGA? \_ It certainly makes low profile form factor machines easier \_ Call Dell and ask. \_ the flip chip (FC) packaging dissapates heat much better than the slotted versions. it's cheaper just cause it's cheaper to make a chip (corn cheaps!) than a chip on a board. Same performance, though (except of course, some P3's have a 133MHz system bus, and some have 100MHz system bus, depending on whether the suffix is EB or E, resp.). \_ Go for the EBs. If it's a non-slot one, it's most likely an EB. The cache is half the size of slot one but it runs full speed and is on-chip. Don't get the E when you can get the EB for pretty much the same price.. an almost 33% increase in memory performance.. \_ Actually, I would look for an "E", not "EB". The EB's are already pretty much clocked as high as they can go and go for the same price as the E's. The E's can be clocked to hell and back...avg clocker can get about 850mhz out of 600mhz (EB ~ 700). Check the overclocking pages on the net for more info. I found http://www.overclockers.com to be a good resource among others. |
2000/7/5-6 [Computer/HW/Memory, Computer/HW/Drives] UID:18585 Activity:high |
7/4 I want an Ultra 5. Should I pay the SunStore price or get a used one? \_ Whichever way you go, check out the processor specs - some come with CPU's with 256k L2 cache, others with 2mb L2 cache. There is a huge difference in performance between the two. \_ LOTS of people i know are getting ultra-5s. I'd be worried that they are gonna run out of them. Buy now before it's too late! \_ Weren't workstations with the UltraSPARC-III coming out or something? I want my cheap, previous-generation Ultra 5. \_ I'd rather wait for the next generation Ultra5s with UltraSparcIII processors. The current model is a lemon, just like it was three \_ You'll be waiting a long time. The US-III is going into servers & higher end machines, too expensive for U5's right now. \_ Don't be too sure about that. years ago. Performance is bogged down by some cheap-ass parts that SUN is using, some of which are obsolete even by PC laptop standards You get only 4Megs of video memory. The disks on the "cheap" low-end models are 5400 RPM IDE disks. In addition to the slow IDE disks they don't have a UDMA/33 or UDMA/66 capable IDE controller. Their IDE controllers only support DMA mode 2 interface (capable of 16.6MB/s data transfers). I'd like to find the engineer who designed \_ Unfortunately "cheap" is still $2.5K. Any idea when the new ones will be out so I can get my discounted Sun? the Ultra5 and 10s and beat him with one of those. \_ you can't beat anyone with an ultra 5, they're cheaply constructed and would fall apart. Beat him with a 3/50 \_ When are the UltraSPARC III's coming out? instead. -tom \_ You can beat 'em once before it breaks and that's the one that counts. \_ So ... new, not used, I guess? \_ Looks like my offer got deleted. I've got a Ultra Enterpise 2 (256 MB Ram, 2 x 200 MHz USII, Creator 3D) that I'm not using and would be willing to sell/trade it to you. ----ranga \_ Does it play Quake 3? How many FPS? \_ Can you beat the U-5 price? Then i'd be interested in upgrading from the U-1 i have at home. |
2000/7/3-4 [Computer/HW/Memory, Computer/SW/Languages/Functional] UID:18580 Activity:insanely high |
7/3 Is it possible to derive a mathematical or logical proof which shows that, given a set of computations to perform, an instruction set consisting of only push/pop instructions (1 register) requires less memory footprint than that of a general purpose registered machine? \_ java troll, go away \_ Wow. Berkeley is graduating idiots like this? \_ hot market + idiot programmers = old news \_ awww, cut the kid some slack -- ignorance != stupidity. just chalk it up to being naive, inexperienced, and enthusiastic. \_Would someone please explain exactly what the poster is asking, and why it's ridiculous? What does it have to do w/ java? \_ cs152+252 would help \_ cs164 talked about it a little \_ More like who cares? Memory footprint ceased being deeply meaningful after BG chewed on those "more 640k" words. RAM is cheap and other kinds of higher density storage are always in the works. Don't bother me kid. \_ just like soda to call someone stupid, then not be able to answer (ooh, I'm too busy to answer, but have plenty of time to carp). ehe. \_ Didn't say you're stupid. Said it isn't important. \_ i dare you to to write a loop using push/pop. \_ (define (loopy numtimes doit result) (if (= numtimes 0) result (loopy (1- numtimes) (cons (doit numtimes) result)))) ITERATION = RECURSION = stacks (with push = cons) \_ NO! With the assembly push/pop instructions. Not LISP _simulating_ push/pop. You don't get "cons" and if/then tests in a push/pop-only instruction set. Read the test question fully before answering. Grade: F. \_ I think the guy who invented Forth wrote a whole book about when stack architectures are better that general purpose RISC and I read it online, but I don't have the URL anymore. -muchandr \_ Found it. I ment this guy really: http://www.cs.cmu.edu/~koopman/stack_computers For example section 7.2.3 deals with rule-based systems being faster. Another good page http://www.ptsc.com/psc1000/mpu.html has an example on how you can do better because of better instruction bandwidth of 8-bit zero-operand instructions -muchandr \_ dude, heavy computer science! The poster isn't as clueless as I thought. |
2000/5/12 [Computer/HW/Memory] UID:18244 Activity:nil |
5/11 I need to allocate more memory than "new" gives me... how do I do this? mmap? Why can't I "new" 100 megs if there is enough VM on the system? \_ Probably have per-process memory limits set too low - see the man pages for limit/ulimit/setrlimit |
2000/5/5-6 [Computer/HW/Memory] UID:18183 Activity:high |
4/35 Back in my Good Ol DOS days, i used to create a ram disk for fast read/write scratch space. How would I do this under linux? -crebbs \_ Aren't files supposed to be cached in RAM? \_ Files are "cached" in RAM. However, they must exist on disk first. What i want to do is create a partition of RAM, say /ramdisk, which i can write files too, work on and which will stay around untill I delete_them/turn_off_the_computer. \_ This is what swapfs is for. I don't know if it exists under linux or not. Yuo can usually mount it as /tmp. \_ Geeze, would you bozos get a clue before you give stupid answers? mkdir /ramdisk mke2fs /dev/ram5 5000 (that's 5 megs) mount /dev/ram5 /ramdisk -tom \_ Why do you feel you need this? If you think it will make accessing files in the ramdisk faster you are almost certainly wrong. I suppose a ramdisk might be faster under weird situations. If you think ramdisks are good just because they were under DOS, don't bother. --Galen |
2000/4/5 [Computer/SW/Languages/C_Cplusplus, Computer/HW/Memory] UID:17919 Activity:high |
4/4 JVM usually takes care of stack memory fragmentation right? (Sun's \_ Wrong. You've confused stack & heap. JVM is mark&sweep with conservative compacting algorithm) What about compiled code? Let's say I have a C/C++ server running that keeps doing new+delete, some big, some small. Eventually, there will be a lot of fragmentation. What happens? \_ This depends on what your server is doing. Depending on your memory allocation regime you may consider using a garbage collection library or implementing your own memory management layer. \_ C++ by default ends up fragging the heap. In general, assuming you don't run forever, this isn't too bad a problem. If the server is doing alloc/free's all the time and runs for a long time you should consider rolling your own memory allocator. - seidl \_ Yeah, it's like this: either you care a LOT about memory performance, or you prefer to have ease of use. If you want ease of use you just alloc away and don't care. If you want performance you write your own allocator that does exactly what you want. In games we use a mixture of these two things based on how frequently a particular allocation is going to happen. -blojo \_ blojo, what game are you talking about? Solitaire? \_ If yer gonna say something dumb at least sign your name so I can laugh at you. -blojo \_ Yeah, you never say anything dumb, John. \_ It helps if you can spell his name --oj \_ When I say dumb things, i sign my name to them, dammit. -blojo |
2000/4/2-3 [Computer/SW/Database, Computer/HW/Memory] UID:17911 Activity:high |
4/1 EE question. Why would the delay through a CAM (content addressed memory) cache, like those used in TLBs, increase with size. Since CAMs are built with each entry connected to a comprator and the selection is done through an N-way multiplexor using straight pass logic gates (no multiplexor select line decoding is involved) then theoretically a 64 entry CAM should be about as fast as a 32 entry CAM right (at least it shouldn't be drastically slower)? \_ how much slower is a 64 entry CAM? if it's on the order of a few ns, then it's the additional stage required to select between the output of the decoders and routing. if it's larger, then your assumption that there's one decoder per entry is wrong. \_ But in this case there is no need to decode the select lines since the select lines are not in an encoded form. They come from the actual comparators themeselves so there are N select lines for an N-entry CAM. There is no time to decode the select lines. \_ i guess if each select line is an OE for a buffer which drives the output pads, then yeah. I guess using tristate buffers instead of muxes on the output is the right way of doing this. so what's the answer? have you looked at a datasheet? \_ I actually don't know the answer. I don't think anyone makes an entire IC as a CAM for them to publish a datasheet for them. But every prof I've had in \_ MCM69C233, for example. you must not have looked really hard, since i found one at the first place i bothered to check. computer architecture related courses have always said that larger TLB's and higher associativity caches are slower. I don't see why. \_ bigger rams are slower. larger area is slower than smaller area. i can get a big tag ram for a direct mapped cache. or i can split the tag ram in 2 and do 2-way associativity. each ram is smaller, but i eat an extra couple of levels of logic to figure out which way hits. is one always faster than the other? doubt it for small cache sizes/ways. probably for extreme cases the n-way cache is always slower. ditto for cams. i can see for small sizes the difference in loading/driving the pass gate wouldn't make much difference if num loads go from 32 to 64. can't do pass gates for large loads. for large cams the extra levels of logic to decode the hit entry will make a difference. |
2000/3/13-14 [Computer/HW/Memory, Computer/HW] UID:17755 Activity:high |
3/12 http://www.networkice.com/Products/BlackICE/default.htm this is just too funny \_ Hax0rz y00n1t3!!!11 D-str0i bl/-\k1c3!!!!!11 \_ I can't remember... is the term from cyberpunk or shadowrun? \_ I've seen it in early works from both but don't know which was first. I'm guessing the cyberpunk stuff. \_ William Gibson you illiterate idiots. \_ Specifically, this was the big bad stuff that attacks our hero, Case, in Neuromancer. \_ WG is hardly the height of mandatory reading for the educated individual. Some of his stuff is amusing, but he's not Shakespear. \_ No, that was fair (I was the one who asked) I've read neuromancer, but the earlier / salient memory was of an rpg. \_ Or Shakespeare either. \_ Sorry. I've read nothing but the best for the last 10 years. Gibson all the way! |
2000/3/6-7 [Computer/SW/Languages/C_Cplusplus, Computer/HW/Memory, Computer/SW/Compilers] UID:17696 Activity:nil |
3/4 What does a Segmentation fault mean? I have a program with and gdb reveals that the program segfaults at while (x == 0) { It seems to segfault on one machine each time but never on another. \_ what's "x"? \_ did you malloc enuf space? \_ IOW, did you allocate memory for x? Is x defined, or just declared as a variable? What type is x? \_ Illegal reference to memory not owned by the process |
2000/3/1-2 [Computer/HW/Memory] UID:17673 Activity:nil |
3/1 What is the soda DNS cache timeout? How long should it take to update the IP of a named system? \_ Doesn't this depend on the information the name server recieved about the ttl of the IN A/PTR record? Also the name server may have a minimum ttl below which it will not accept an update. |
2000/1/28-29 [Computer/HW/Memory, Computer/SW/OS/Solaris] UID:17364 Activity:kinda low |
1/28 On Solaris 2.7 I have a process that dumps core when its memory usage passes 2 GB. I have 4 GB ram and this is the only process that requires significant resources. Is this related to resource limits? According to ulimit my memory utilization is unlimited. \_ probably related to your bit-bucket only handling 32 bits \_ recompile your binary to be 64-bit (-xarch=v9 if using Sun cc) \_ fix your program so that it checks the return value of malloc() rather than blindly assuming there is always more memory. |
2000/1/17-19 [Computer/HW/Memory] UID:17258 Activity:kinda low |
1/17 soda 17: w w: kvm_getprocs: Cannot allocate memory: Cannot allocate memory \_ Not enough Clue, You must mine more clue. \_ Yeah, if I had more Clue, then I would've magically been able to allocate more memory on soda for this. Gosh, thanks! U R D CL00MAST0R!!!11 |
2000/1/13-14 [Computer/HW/Memory] UID:17229 Activity:nil |
1/12 In nslookup, what does "Non-authoritative answer" mean? \_ that the answer came from the cache \_ That the name server you asked isn't the name server for that domain, but asked someone else and believed what it was told. If you nslookup -query=ANY the domain, you'll get the list of authoritative name servers for that domain. |
1999/12/20-23 [Computer/HW/Memory] UID:17072 Activity:moderate |
12/20 What's the best way to figure out the maximum memory a process uses in its lifecycle? \_ call wait3() or wait4() from the parent and check the values in the rusage struct \_ is this because IdiotOS doesn't support getrusage()? -ali \_ man time(1). 'time -l' will tell you. you will probably have to avoid your shell's builtin time. --aaron \_ man asian_chix man asian_chix No manual entry for asian_chix \_ unsetenv MANPATH LocalWords: motd download www samesexmarriage org com btw tmp csua pdf aspo \_ Didn't make a difference. |
1999/9/8 [Computer/HW/Memory] UID:16482 Activity:moderate |
9/7 As a general rule, do mobile celerons outperform mobile pentium 2s? \_ I wouldn't think so, since the general appeal of celerons seems to be their ability to take massive overclocking, and in a laptop you probably won't be able to fit the requisite cooling. \_ mobile celery has 128k L2 on die, mobile p2 has 256k L2 on die, and are pretty much the same else where, which do u think is faster? \_ It's not as simple as comparing cache sizes. The cache on the Celeron runs at a faster speed than the cache on the PII. The chips are very, very similar in benchmarks. Don't buy the PII unless the price difference isn't large (PII is a little bit faster overall). --dim \_ it's so odd but I was talking to a friend about this last night; apparently the mobile P2's do have big on-die cache and are faster, no numbers though -jctwu \_ celery's cache runs at faster speed? nope, the new P2 (dubbed P2E?) has 256K L2 on die running at FULL speed. Unless P2's on die cache is designed differently logic wise, P2 and celery's cache should be running at the same speed. \_ about damn time they made the P2 better than the Celeron, other than by bus-speed locking the Celeron. Oh wait, P2's are phased out for P3's. I guess they still need P2's for the laptops. \_ does the L2 cache of mobile P2's run at CPU speed? \_ yes and no. New generation of mobile P2 has 256k L2 running at cpu speed. Old mobile P2 has 512k L2 running at 1/2 cpu speed. |
1999/8/5-6 [Computer/HW/Memory, Consumer/PDA] UID:16253 Activity:nil |
8/4 Do you have a Palm III or IIIx? How much memory does PalmOS and the built-in apps take up? I'm trying to decide whether 2 megs on a Palm III is good enough to run the map applications I need. Thx. \_ all that is on the roms \_ OS and built in live in ROM, so you get the full 2 or 4 megs. I find as long as I don't put more than one whole book on it or too many games (> 3) 2mb is o.k. The IIIx has a better screen though, so you might want to look at it. Not sure which screen the IIIe has. - seidl |
1999/7/6 [Computer/HW/Memory] UID:16080 Activity:low |
7/5 $tanfraud rules1 http://www.iptq.com/benveniste/Default.htm \_ jeehosephat. If this guy's theories are true, it looks like Yoda was right. The Force is everywhere. Or at least the memory. Then again, he first published 10 years ago. Why has noone made news about this since then, if it were verifiable? \_ His results were NOT verifiable, and in a later issue of NATURE his results were refuted and he was made to look pretty stupid. He complained about the methods used to test his results, but ultimately nobody really believes him. -drex |
1999/5/11-12 [Computer/HW/Memory, Computer/SW/Unix] UID:15790 Activity:nil |
5/10 /tmp/q3test_1_05.exe \_ and /tmp/gamespyinstaller210std.exe to actually find net games \_ Please, please please don't put large files in /tmp. It's a memoryfs mounted filesystem. (translation: fill it up and you fill up memory on soda) Put large files in /csua/tmp |
1999/4/28 [Computer/HW/Memory] UID:15712 Activity:high |
4/27 Is the cache expansion port on different brands of PC motherboards compatible? ie, can i take a cache module from one motherboard and plug it into another to increase cache memory to 512k? \_ Absolutely yes, unless the modules are not compatible. \_ how many computers have you built?? there's no place to add the L2 cache on most MBs. Plus most pentium/I/II (non-xeon) use Intel chipsets that don't bother with >512K cache. only PC MBs i've seen with >512k cache have VIA (ie MVP3) chipsets \_ You, my friend, are definitely irony challenged. |
1999/3/20-22 [Computer/HW/Memory, Computer/HW/Display, Computer/HW/CPU] UID:15617 Activity:insanely high |
3/19 Dell anounces it will ship PCs with RedHat Linux preinstalled: http://www.news.com/News/Item/0,4,34036,00.html \_ Is there going to be a Linux refund day? \_ Why? Isn't Linux free, unlike Windoze? \_ Most of it is free but RedHat throws in a bunch of commercial crap that nobody uses (like a Real media server) so that can charge $40 for it. It also comes w/ a manual. \_ RedHat's Lunitux. And it costs them money to install it. \_ Good luck to Dell to get Linux users to buy Dell computers. \_ What's wrong with Dell's? They only cost twice as much as a computer you build yourself. \_ and they're worth more than twice as much. -tom \_ The only x86 based PC worth paying extra money for performance are IBM's and SGI's. \_ Yeah all that off the shelf hardware magically becomes much more reliable and better performing when installed by a Dell technical genius. \_ No, it becomes WARRANTIED, and YOU DON'T HAVE TO SPEND TIME PUTTING IT TOGETHER AND MAKING IT WORK. -tom \_ My parts are covered. I think I can afford an hour or two to save $1000 or more. Anyway, it's a hobby thing and I get *exactly* the hardware I want. To the person below, who said I was using cheap parts? What's a cheap part anyway? They all come from the same factories. Please name a "cheap part" that will be hard to install. You just need a good case. The rest is generic. \_ Selecting a mother board would be your first problem. Intel, Asus, SuperMicro, Tyan, etc? For example, I tend not to want to use a model that just came out. New models tend to have problems fixed only in later revisions. They are usually only tested with brand name components, and even that is not guanranteed. When there is a problem, you are sometimes left wondering whether the problem is with the PCI bus, a BIOS setting, a motherboard switch setting, your network card, a device driver problem, compatibility problem, etc. When going crazy trying to solve a problem, you may unknowingly create a second problem zapping another component with static. I would gladly let Dell do the dog work and quality control for me, and let them iron out all the problems. \_ I don't see MB selection as a "problem". People, this isn't rocket science. You're just assembling stock parts. This isn't the old bad days of Dos 2.1 (or even 6.2) where nothing worked and it took hours of fucking with config.sys to get something working. The parts Dell uses are the same you're getting. They don't come from a Magic Dell Factory whever everything is always wonderful Factory where everything is always wonderful while your parts are coming from Factory Hell where nothing ever works. It's the same stuff. The only difference is you won't be paying for parts you don't need/want and can put that money towards things you do. \_ Yes, Dell parts are the same you are getting but they do a couple of things. One, they test and throw out all the defective parts. Two, they make sure all their components work together well. Three, they configure and assemble the system for you on a production line. Four, they test the whole system again after assembly. Go to a computer reseller, open a box of say Diamond Multimedia video cards or Intel motherboards and test them. The defective rate can be as high as 10 percent. \_ Ok, granting all these things are true, pleasure. which they're not entirely, even if I got a defective board, so what? I return it. I get a new one. This isn't for work. It's a hobbyist home system. Dell buys in bulk so they have lower price purchase prices but their rates aren't *that* much lower, you *are* paying for them to build it and run their business and you *don't* get exactly the parts you want. If you don't care about what's really under the hood, then you are a Dell customer and I hope the value you feel they're adding is worth the price you're paying. Nothing wrong with that. It just isn't the best you can do for your money. \_ Just make sure you test your motherboard well before assembling your system. I hate taking a motherboard out. I also hate driving one hour to Hi-tech USA (or whatever) to exchange a part, or packing stuff and going to the mail shop. I do care about what is under the hood, and Dell components seem fine to me. Price advantage of resellers over brand name systems has been eroding over the past few years, so I doubt you can save much money. You do get to use the exact parts you want though, and if you need to tinker with your system in the future, you already know it through and through. Enjoy building your system. I love the ones I built, but would not do it again. \_ look, have fun building your box, but you're fooling yourself if you think it makes economic sense. -tom \_ Whatever Tom. \_ Ja, building your own system is a nice learning experience, but after doing it for a few times, you decide that the money saved is not worth the amount of effort you have to put in. You also realize that the cheaper the components you used, the higher the amount of time you have to put in. \_ Agreed. I've been building my own computers since the 486 days. But for my dad, I'm gonna plop down some $$$ for a Dell Celeron. It's just not worth the hassle each time. \_ celeron sux. go with amd-k2. \_ On Friday I saw a celeron 300MHz on an overclocked board (66->100Mhz) perform a little less than twice as fast as an AMD-K2 300Mhz. Celeron used to suck, but no longer and for $60/chip... do the math fuckwit. \_ I built my first computer before the 8086 days. I don't get so easily "confused" that I need Dell to pick all my parts for me and charge me more for the pleasure. For your dad, sure, spend his money. For yourself... don't you care? \_ how much do you care about your time? I can configure, price and order a Dell system in 5 minutes, sitting at my computer. How much time do you spend contacting 8 different vendors for parts? How much time do you spend putting it together? If one of the DIMM slots is bad, how much time do you spend taking the machine apart again, and trying to convince the motherboard manufacturer that it wasn't your DIMMs that were bad? If you're still a student, or some other class of person whose life decisions are more cash-bound than time-bound, that's fine. If you have a job, and particularly if you're a computer professional in a job that pays good money and probably takes more than 40 hours, why the hell would you spend one minute more of your time than is necessary, to try to wring, at best, minor cash savings out of a purchase? Seriously, how much does your free time cost, per hour? -tom \-that's a psb law: "there are two kinds of people. peopel with more $ than time and people with more time than $." \_ It's a hobby, you freak! How much does your netrek or other gaming time cost per hour? \_ I don't try to claim that netrek is cost-effective. Like I said, if you enjoy building systems, have fun. -tom \_ Well, nowhere in my answer did I say I was gonna stop buying for myself. Unless a computer maker magically has all the exact parts I want and I don't feel like spending an entire weekend going to all these small stores finding the cheapest prices and constructing it myself. \_ It isn't about money. I'll spend a few extra bucks _AND_ the time to get _exactly_ the system I want. _Exactly_ what I want. Not from their approved list on a web form. I'm glad the consumer route works for you. \_ Don't some companies, even Dell, allow you to customize the computer you buy from them? \_ On some web form, yes, to a *very* limited extent. But if you want that safe feeling from knowing you have the same computer Dell has built for 10k others before you, you can't get that from a highly customised system. It isn't possible. At that point, you're now paying them to collect and assemble the same random pile of parts you would've built for yourself anyway. You've already done research and everything else short of actually popping it all in case. Might as well just take that last step yourself. And btw, I've seen vendor specific versions of some hardware which require you to use the vendor drivers. Generic drivers don't work. Upgrading sucks too because the cases are fucked also. Caveat emptor and all that no matter which way you go. Dell is not the be-all end-all perfectly safe answer to hardware buying. \_ Rather the contradiction with "off the shelf" parts & "vendor drivers", no? In any case, buying from a vendor is all about convenience and no hassle, especially when you need 20-30 identically set-up computers. \_ No, it isn't a contradiction. They're only tweaking the HW to force you to come to them for future support and upgrades, not to enhance it in any way. Compaq is especially prone to this and has a lousy system of storing and categorising patches to go along with it. Other than making sure you're trapped with them forever, there's nothing better or different about the HW. It's just a scam, not an improvement. As far as 20-30 goes, go ahead. Buy Dell. No one said you should build 20-30 machines from scrap. The idea is stupid and you're creating a weak straw man argument. |
1999/1/28-29 [Computer/HW/Memory, Computer/SW/Languages/Misc] UID:15309 Activity:nil |
1/27 If the name server (DNS/named) caches the domain name<->IP pairs, where is the cache file located? Is the cache persistent in reboots? \_ It caches them in memory. That's one of the reasons why named grows in size the longer it's been running. A reboot (or even kill and restart) will clear the cache. --dim \-bind is moving away from kill -SIG "control" to use ndc [named control program] for more sophisticated control/ communication with a running named. i dont remember off the top of my head if the onld-style signals will continue work. --psb \_ ndc is just a shell script that sends a kill -SIG you dork. \_ If you kill the named process, they will. :-) |
1998/11/28-30 [Computer/HW/Memory] UID:15036 Activity:high |
11/28 Let's talk about virtual memory: \__ if you'r edealing with virtual memory which far exceeds physical memory, you've already lost. \_ if you'r edealing with virtual memory which far exceeds \_ First of all, the guy is talking about number crunching, not image processing. It is likely that he's going to be addressing all of the memory he's crunching with. And second, no one said it was impossible--it just is painfully slow. disk is like 6 orders of magnitude slower than RAM. \_ uhh. "number crunching" applications usually exhibit greater locality than almost any other app, if optimized properly. -nick \_ which will help not at all if it's using more than physical RAM. \_ What part of "locality" don't you understand, twink? Nick knows what he's talking about. \_ Whether you can get away with crunching >> than RAM depends strongly on the precise task. A blocked matrix multiply might perform well, especially if you somehow anticipate the data needed next and stream it into RAM beforehand so you don't have to eat the large latency while it is paged in..... Having a RAID array (expensive) also helps. A non-sparse matrix-vector multiply, however, requires 1 memory reference for every 2 flops (not counting the memory reference for the vector element). memory reference for the vector element or result). Assuming .5 flop per tick on a 400MHz P-II, we'd need floats from the matrix at 100MHz, or 400MB/sec. SDRAM might sustain that, but if the matrix were SDRAM might sustain that, but if the matrix were much larger than memory.... Performance would drop 10x at least. --PeterM \_ I don't think ILP is heavily influenced by how much a process' virtual memory compares to the physical memory. Virtual memory pages are usually on the order of 64kb. Compare that to, say, a Cray vector register file which is a 32x32 64bit matrix. That's 8kb and it takes several clock cycles for a number crunching program to process the data in the 64kb page anyway. But this guy is talking about a Pentium and a K6 to do number crunching. I don't think he's going to benefit from that kind of ILP and even if he did he would still benefit from the spatial and temporal locality of the program. physical memory, you've already lost. (\_ This has pretty much been my experience--PeterM ) \_ First of all, the guy is talking about number crunching, not image processing. It is likely that he's going to be addressing all of the memory he's crunching with. And second, no one said it was impossible--it just is painfully slow. disk is like 6 orders of magnitude slower than RAM. \_ uhh. "number crunching" applications usually exhibit greater locality than almost any other app, if optimized properly. -nick \_ which will help not at all if it's using more than physical RAM. \_ What part of "locality" don't you understand, twink? Nick knows what he's talking about. \_ Whether you can get away with crunching >> than RAM depends strongly on the precise task. A blocked matrix multiply might perform well, especially if you somehow anticipate the data needed next and stream it into RAM beforehand so you don't have to eat the large latency while it is paged in..... Having a RAID array (expensive) also helps. A non-sparse matrix-vector multiply, however, requires 1 memory reference for every 2 flops (not counting the memory reference for the vector element or result). Assuming .5 flop per tick on a 400MHz P-II, we'd need floats from the matrix at 100MHz, or 400MB/sec. SDRAM might sustain that, but if the matrix were much larger than memory.... Performance would drop 10x at least. --PeterM \_ I don't think ILP is heavily influenced by \_ ILP ==> "Instruction Level Parallelism" I don't understand why you mention it here. --PeterM \_ You mentioned vector ops which is a form of instruction level parallelism. how much a process' virtual memory compares to the physical memory. Virtual memory pages are usually on the order of 64kb. Compare that to, say, a Cray vector register file which is a 32x32 64bit matrix. That's 8kb and it takes several clock cycles for a number crunching program to process the data in the 64kb page anyway. But this guy is talking about a Pentium and a K6 to do number crunching. I don't think he's going to benefit from that kind of ILP and even if he did he would still benefit from the spatial and temporal locality of the program. |
1998/11/11 [Computer/HW/Memory] UID:14942 Activity:kinda low |
11/10 Does anyone know if the 100MHz SDRAM is compatible with non-100Mhz bus? The reason I ask is because I need to buy new RAM for my overlocked 83MHz bus w/ P250 CPU and I prefer not to spend money on traditional 66MHz RAM that is practically obsolete. \_ I don't know if they're compatible, but don't see why not. Since RAM standards are going to change before you can upgrade, why do you care? You'll be buying Rambus DRAMs soon enough... \_ http://www.tomshardware.com suggests that Rambus isn't all its cracked up to be. \_ 100 MHz SDRAM will work. I frown upon non-66, non-100 system bus frequencies. :( Yer peripherals are choking. -jctwu |
1998/10/7-10 [Computer/SW/Graphics, Computer/HW/Memory] UID:14749 Activity:high |
10/7 Brand New Acer Aspire 3060 Minitower w/ AMD K6-2 350MHz, 10.2GB HD, 32X CD, 56K modem, 16-bit audio card with 3D stereo sound, on board ATI 3D Rage Pro AGP with 4MB SGRAM Video memory, MPEG video playback. Plus 64MB SDRAM at 60ns , and one year warranty for $950. \_ 4sale I presume. Does that come with a monitor? Who R U? \_ no monitor -- tora67@hotmail.com \_ No RAM? Forget it! \_ hahaha my bad, I thought I wrote in 64MB, obviously someone removed it. Bo hid behind some bushes as thirty women stampeded by him. "Shit!" he said as they were all past. "What the hell am I gonna do? I can't get laid, I can't stay safe, and damn it, my mom wants me! Gag!" He stood up slowly and looked around. "There he is!" somebody cried. "Oh fuck, man!" Bo yelled,and sprinted away. He had to get help. This wasn't what he wanted at all. This sucked. He'd just wanted good sex and to be adored by women. He didn't want to be rampaged by sex crazed sluts. What was he going to do? Mick. That was it. Mick could help him. He would be able to get him out of this predicament. \_ no penis \_ Where's this from? |
1998/10/5-6 [Computer/HW/Memory, Computer/HW/CPU] UID:14734 Activity:insanely high |
10/5 overclocked my Celeron 300A to 450 MHz. ~jctwu/pub/mendocino. \_ YouGo Girl \_ Hey, and I put nitrous in my Yugo! \_ the new Celerons are very well-engineered and -fab'd. My bet is that Intel had to cripple them to 66 MHz bus speed so as not to compete with the Pentium 2. -jctwu \_ good for you. go away \_ hmmm... well, i guess this is a step in the right direction instead of outright deleting a post you don't agree with. \_ good for you. go away \_Intel Celeron sucks \_ good for you. go away \_ good for you. go away \_ good for you. go away \_ yes the L2 cache-less ones. read file for details. \_ Even the zero cache ones easily overclock and are great for gaminng where the loss of cache is usually over come by the higher bus speeds. If you're concerned about poor "Business Benchmarks", well... seriously, just how fast do you want MSWord to wait for you to type something? Nothing wrong with Celerons, old or new. \_ is it possible that they overclock MORE EASILY, due to no cache? (I'm CS, not EECS, and curious) \_ yes, take 150. cache increases probibility of setup time violations. \_ no. \_ yes (non-EE). 1) You don't get any traffic on the system bus between the CPU and L2 cache => easier \_ you never get traffic on the system bus between the CPU and L2 anyway. [ CPU ] | [ L2$ ] [ Mem ] | | ===========system bus============= overclocking. 2) The L2 cache is usu. off-die, and thus not under a heat sink / fan. The cache would get super hot. -jctwu \_ hmm, right. What happened to my Hennessey? Anyways, reason 2 still applies. -jctwu \_ Merging different things together onto a single package (ie proc + mem) usually cuts down on power consumption. Check out U.Wisconson Gallileo and Berkeley iram. \_ Well, no dah. Probably 1/3 of the power goes to i/o drivers. As drivers get smaller to nonexistent, obviously power consumption goes down. \_old news. btw which motherboard are you using? \_ abit BH6. read file. \_ the BH6 rocks. \_ what file? says "file not found" \_ was there; updated (in pub dir, not ~). \_ what's the whole url dodo? \_ it's not a url you fool; "more ~jctwu/pub/mendocino" \_ symbolic link now on <DEAD>.../~jctwu/docs/mendocino.txt<DEAD> \_ So you liked hardware tweaking your celeron? Feel like an Uber-hacker? Now try the dual-cpu-enable tweak. <DEAD>kikumaru.w-w.ne.jp/pc/celeron/index_e.html<DEAD> |
1998/9/28-10/2 [Computer/HW/Memory] UID:14691 Activity:moderate |
9/28 Need to find a 4kx4 RAM chip to fix an arcade game. The simians at AL Lashers were unhelpful. Radio Shanty is right out. What other options do I have? -ERic \Mailorder: digikey, et al. \_ couldn't find anything relevant on digikey's website. Any other ideas? -ERic \_ how about simulating one with a larger chip? might be cheaper. \_ I was thinking about that -- but I'm not enuf of an eecs geek to know whether it'd have a prayer of working. -ERic \_ http://www.jameco.com . Or if it's something obsolete and obscure you can try Halted. \_ Try a military warehouse, they have to keep obselete \_ or Haltek. It's cooler. \_ Try a military whorehouse, they have to keep obselete hardware for years because some plane flying still uses it. -- daveh \_ Cool. We can break in like on that Buffy episode where Xander and Cordelia snuck into the base to steal a bazooka. All we need now is an attractive woman to distra-- Oh... nevermind. That won't work around here. \_ Maybe we can hire one of those out-of-work MCB major chicks \_ They have to be good for *something*. \_ What kind of RAM chip? You could also get a databook (handbook) from various OEMs (TI, Motorola, various obscure chipmakers), and look it up in their memory section. Some of these handbooks are hundreds of pages -- must be in there _somewhere_... |
1998/6/9 [Computer/HW/Memory, Computer/HW/CPU] UID:14188 Activity:nil |
6/8 See the machine that is the legacy of the late Seymour Cray: <DEAD>gopher.CCS.ORNL.GOV/SRC<DEAD> 6 Intel Xeon Processors, cray-like memory system, and FPGA-based reconfigurable elements. --jon - GEEKS!! \_ Don't wanna read geek-talk? Get the hell out then. \_ Da CSUA baby! Love it or leave it. - tpc \_ how aobut you leave instead? |
1998/5/28 [Computer/HW/Memory, Computer/SW/OS/FreeBSD] UID:14148 Activity:nil |
5/27 How heavily cached is the file system? If I read a file over and over again, how much (in terms of mb) is it usually in memory, in systems like FreeBSD? \_ Your hard drive has a cache as well. \_ Note that this matters less and less as you get a big software cache -- in the worst case, if your hardware and software caches were the same size and used the same policy, you wouldn't get any benefit from the hardware cache at all. \_ You fail to mention hard drives that prefetch into their cache. This is pretty much unrelated to the original question, whose answer is "usually most of it; run top to see how much of your physical memory is allocated to cache". \_ If the file you're reading over and over again is a read-only file, or changes in small incremental, it may be wise to use a database who's sole task is "smart disk cache". \_ Shut up, cmlee. \_ Good call. \_ FreeBSD has a unified vm/buffer cache, meaning that file caching and process memory both contend for the same physical memory (in other words, the memory isn't segmented between the two). so really, the answer to your question depends on how loaded the system is, both vm and i/o-wise. now, if you don't want your file cached (because you're only reading sequentially, or something), look at madvise(). |
1998/5/22 [Computer/SW/Languages, Computer/HW/Memory] UID:14130 Activity:kinda low |
5/21 Five reasons to believe computers are female: 1. No one but the Creator understands their internal logic. 2. The native language they use to communicate with other computers is incomprehensible to everyone else. 3. The message "Bad command or file name" is about as informative as, "If you don't know why I'm mad at you, then I'm certainly not going to tell you". 4. Even your smallest mistakes are stored in long-term memory for later retrieval. 5. As soon as you make a commitment to one, you find yourself spending half your paycheck on accessories for it. \_ only half? |
1998/4/4 [Computer/Theory, Computer/HW/Memory] UID:13899 Activity:nil |
4/3 How do you solve: 1) summation of 1/n where n = (1 to infinity) 2) summation of 1/n^2 where n = (1 to infinity) THIS IS A SERIOUS QUESTION, THANKS!!! \_ If you sum from 1 to infinity instead, the first one, at least, still diverges to infinity. \_THANKS! How do you solve by induction w/base case? \_ Is your homework due at 5 PM? \_ Already due. Actually I'm at work and I don't have my calculus book. \_ I'm trying to solve the interview question: How do you write a program that can randomly and evenly pick ANY line in a huge text file, in one parse, where you have very little memory? |
1998/3/19-20 [Computer/HW/Memory] UID:13843 Activity:low |
3/19 Memory is getting cheaper lately. Anyone know when notebook memories will get cheaper? Ie. When a 32 EDO RAM for a notebook will not cost $300 like it is now? \_ http://www.thechipmerchant.com \_ Errr, notebook memory prices aren't *that* high... 64MB for my notebook came out to $210. -nevman |
1994/4/11-17 [Computer/HW/Memory] UID:31552 Activity:nil |
4/9 D'oww. 16 meg memory board died. Soda's down to 64 megs of memory total. Increased swappage is likely. --dpassage |
12/24 |