Berkeley CSUA MOTD:2000:June:12 Monday <Sunday, Tuesday>
Berkeley CSUA MOTD
2000/6/12-14 [Computer/SW/Security] UID:18446 Activity:moderate
6/11    Anybody know if encryption routines (DES, IPsec related, etc) can
        be parallelized?  Does adding more CPUs and writing some parallel
        software speed things up?
        \_ Look at the source code.  Much of the time, what can be
           parallelized is done at a fine grain level (vector data,
           level, loop level, instruction level, etc...) in which
           case, adding CPU's won't do you any good.  If it's thread
           level paralellism, then yes.  Go to's web site and
           search for Krste Asonovic (he was a PhD student here w/
           Patterson).  His thesis has a good explaination.  Also look
           at the spring 2000 cs252 website.  I think someone did
           a project on encryption algorithms. -jeff
        \_ IPSEC isn't an encryption routine--IPSEC ESP just makes
           provision for tunnel encryption and key exchange for
           whatever crypto you're using.  -John the Nitpicker
        \_ No.
        \_ It depends on the feedback mode used.  If the cyper is running
           in ECB mode, yes, but it's a bad mode of operation otherwise.
           the most common mode, CFB mode, has a dependency between
           blocks and can't be parallelized.  -nweaver
           \_ is that a mathematically proven statement or a "can't _easily_
              be paralellized"?
                \_ Do you understand what you're talking about?  If step B
                   depends on the result of step A before it can be started,
                   IT'S IMPOSSIBLE TO RUN A & B IN PARALLEL.
                   \_ Do YOU understand what YOU are talking about? There's
                      more than one way to split a task into blocks, and
                      parallelism need not apply at global level to be useful.
                      A complete mathematical proof of nweaver's statement
                      would be quite difficult.
                   \_ Not so.  It's been done before in superscalar
                      processors using load value prediction and
                      trivial computation predictions.
                   \_ CFB can not be parallelized beyond the parallelism
                      inherant in the encryption of a single block, because
                      of the dependency.  CFB of block N is computed by
                      encrypting the value of N xor the last block.  -nweaver
                      \_ look, computation prediction is NOT trivial!!
                   \_ CFB encyption can NOT be parallelized beyond the
                      parallelimsm inherant in the encryption of a single
                      block, because of the cyclic dependency.  You need to
                      completely encrypt one block before you can begin
                      encrypting the next block.  CFB DECRYPTION however,
                      can be parallelized between blocks.  -nweaver
2000/6/12-14 [Computer/HW/CPU] UID:18447 Activity:high
6/12    What's the difference between a 386DX and a 80386?  I have a "386(TM)DX
        Microprocessor Programmer's Reference Manual" and a "80386 Programmer's
        Reference Manual", but they list different cycle counts for the same
        \_ It's the exact same thing.  One of your manuals is wrong.
        \_ 386DX is the "Deluxe" version of the 80386
           \_ Hmm, so the "Deluxe" requires more cycles?  "REP MOVS" is 8+4*CX
              cycles on the 386DX but 5+4*CX on the 80386?
                \_ No one ever accussed the Intel x86 designers of doing things
                   that made sense.  Just look at the whole segmented memory
           \_ 386SX is the "SuX" version of the 80386
        \_ I vaguely remember one has the co-processor built-in and the
           other doesn't.  But then, my memory is pretty faulty these days.
           BTW, it was the 486SX that didn't have a co-processor.  80386's
           never had them.
                \_ just don't mention anything like this when Kahan is around
           \_ 486SX had an 8-bit bus, whereas the DX has a 16-bit bus.
              \_ I'm sorry, but that was an incorrect answer.
        \_ a 386DX is an 80386DX without the annoying 80 part.
        \_ a 386DX is, in fact, a .25 micron copper-process 68030
           \_ so what's a 386SX?
              \_ .30 micron aluminum
              \_ .44 magnum
                 \_ *giggle*
                 \_ Or maybe it is a .8 CMOS -eric
        \_ the 386DX is the 80386, as god intended it to be. the DX
           to distinguish it from the later, lamer 386SX which is a
           386 crippled with a 16bit data bus (as opposed to 32 on the DX).
           Both are essentially programatically identical. To confuse
           matters further, there was a 486DX (normal) and 486SX but
           the difference there was that the former had a built-in maths
           coprocessor and the latter did not. Bus widths were the same.
           \_ And there's a 486GX with 16-bit data bus.
           \_ Then how come the manuals of 386DX and 80386 shows different
              cycle counts for "REP MOVS"?
                \_ Why do you care?  No one uses 386's anymore and the timings
                   on modern CPUs are far different.
           \_ Didn't the 486SX actually have the coprocessor on-chip but
              was disabled by an extra pin?
        \_ speaking of processors, what distinguishes a 0.18 and a
           0.25 micron processor (such as when some articles discussing
           the dreamcast processors chips vs. the upcoming ps2 processor
           chips?  the 0.18 is supposed to be better than the 0.25
           but harder to manufacture...what makes the 0.18 better, what
           does that number refer too?   -- no-clue
           \_ 0.18 and 0.25u refers to the minimum channel length of a
              transistor.  Obviously, the shorter the channel length,
              the less resistance there exists in the pull up and pull
              down network of a CMOS circuit.  Changes in process
              technology also effects other factors like the reduction
              of gate oxide thickness which increases the gate capacitance
              per area of transistor and, therefore, increases the gain
              as a result of applying gate voltage.  Also, among many
              other things, the threshold voltage changes.  Take 105
              or 141 and they'll tell you all about it (but by all
              means, avoid Neureuther). -jeff
                \_ Width of 'wiring'.  Smaller = less heat from lower
                   resistence but harder to make.  Less heat = can run
                   faster before dying from heat problems.
                       \_ thanks, i didn't know what that number referred to.
2000/6/12-14 [Computer/SW/OS/FreeBSD] UID:18448 Activity:high
6/12    ANyone know if FreeBSD uses an inode per symbolic link?  --PeterM
        \_ Use the source luke, or build a freebsd system and try it yourself.
           \_ which is much more efficient than asking someone who might know.
        \_ I answered your question on wall already peterm (which is where
           you asked it originally). --jon
           \_ Missed your wall, jon.  Doh.  Thanks though. --PeterM
        \_ yes.  pretty sure all operating systems do.  -dpetrou
        \_ a symlink is a file. files use inodes.
            \_ Reiserfs can store very small files (like symlinks) without
               using an inode.  I wasn't sure FreeBSD didn't do that.  --PeterM
                \_ peterm sounds like a typical super bright wanna-be-tech
                   non-tech major
                        \_ gee I didn't know plasma simulations programming
                           was a non-tech "major".
                   \_ Yeah.  Maybe someday he'll wise up and go back to
                      hanging out with all his fratboy friends back at Haas,
                      where he belongs!
2000/6/12-15 [Reference/BayArea, Reference/RealEstate] UID:18449 Activity:low
6/12    Looking for 2-3 bedroom house in Berkeley/Albany/Emeryville/Oakland
        area that I can be very very noisy in w/o disturbing neighbors.
        Under $2K a sky if you know of any and I shall shower
        you with gratitude.  -sky
        \_ hopefully not of the amber-colored kind
           \_ My urine is glow in the dark orange -sky
              \_ "drink my yellow spray!"
        \_ I don't know any home owners in that area I dislike.
           \_ Ok.  But how about areas where you want the property value
              to depreciate?  -sky
                \_ Liability is a big concern for me.
        \_ I think there are probably a few charming places within walking
           distance of West Oakland BART.  -John
2000/6/12-14 [Industry/Startup, Industry/Jobs] UID:18450 Activity:high
6/12    This is prob. been asked many times already, but what is the
        standard scale for converting from hourly/contractor to yearly/
        employee?  Have choice btwn stock option or bonus structure but
        I think i will take bonus structure...  -sky
        \_ Hourly->Yearly is multiply by 2000, but contractor->employee is
           much harder.
        \_ hourly->yearly is correct.  but for contractor->employee i
           usually like to do hourly * 1000.  so if you're making
           $120/hr, that would be $240,000 annual.  but if you want to be an
           employee, it would give around $120,000 or maybe slightly more.
           \_ Wow!  Are you saying that a company will pay twice as much
              to a contractor as opposed to an employee?  Why?  I do realize
              to a contractor as opposed to an employee?
              \_ no, he's saying that contractors have to pay more of their
                 money to the gubmint as taxes
              Why?  I do realize
              that as a contractor you have to pay self-employment
              tax, health insurance, retirement benefits, vacation, etc.
              When I account for all this it seems to me that contractors
              should get about 1.4--1.5 times as much as salaried employees.
              How do you come up with your x2 number.  Thanks. -emin
              \_ yes, they will pay twice as much for a contractor over a
                 salaried employee.  why?
                 1.  usually contractors are better skilled
                        \_ Hahahahhahahhhahaa.. cough.. gasp... wheeze....
                                \_ contrary to popular belief, not everyone
                                   (other than you) is stupid.  every day
                                   our group is looked at to make sure that
                                   we are adding value to the client's
                                   projects.  and every day they realize
                                   that for every one of our guys we do the
                                   work of at least 3 of their employees.
                                   we work longer hours and more
                                   importantly, we're more productive.
                                   that's why they can justify to their
                                   C-levels the amount of money that is
                                   being spent.  start your own company
                                   and maybe you'll learn. -uctt
                                   \_ Sounding pretty defensive.  I'm sure
                                      each of you is a tremendous value-add
                                      and >insert buzzwords 4 through 16 here<
                                      Yep.  You're worth any 10 normal men!
                 2.  you can get them quickly.  if you need 2 highly skilled
                     java EJB developers with 2+ years of java experience,
                     it will take you at least 2 months to interview 10+
                     people and find 2 good ones and have them start.  with
                     contractors, you can have 2 in your office in 1-2
                 3.  if you only need them for 3-6 months, you can get them
                     to agree to that.  try telling a salaried employee that
                     "your salary is $120k but you may have to get fired in
                     3 months cuz we won't need you then".
                 4.  they usually do not give them stock, they usually can
                     make them work, and they can usually get rid of them
                     relatively painlessly whenever they want to.
                 short answer?  flexibility.  that flexibility is worth
                 paying 2x.  the 2x comes from experience of hiring
                 contractors, hiring employees, and hiring
                 contractors->employees. -uctt
        \_ uctt, are you a contractor? what do you do?
                \_ contract.
                        \_ No wonder he thinks contractors are more skilled.
        \_ full time employee at my own startup now. never contracted in my
           life.  but i did spent some time in big 5 consulting managing
           various e-Commerce projects and our team of consultants from big
           5 was always much better than the employees in the client company
           (that's why they were paying us $275-$350/hr to be there) -uctt
                \_ No.  They were paying that so management could have someone
                   to blame when it all falls to shit.  No one pays you more
                   because you're more skilled.  They don't have a basis to
                   judge your skill level.
                        \_ of course that's a big part of it.  we'll take
                           the blame if it all "falls to shit".  it's called
                           taking RESPONSIBILITY and i'll take the
                           responsiblity for a project in exchange for the
                           big bucks that they have to pay us.  i've
                           met morons in consulting.  i've met a lot more
                           morons that work as full time employees for the
                           client. -uctt
                           \_ You know why that is, right? The companies
                              that don't need your expertise don't hire
                              you. If there's a better staff in house then
                              likely they'll go with that. Granted,
                              sometimes management does stupid things
                              anyway, but I bet this helps explain your
                              experiences. --dim
                                \_ Holy cow!  I find myself agreeing 100% with
                                   dim.  Hell has just frozen over.
                                   \_ I think it's a great idea to hire
                                      consultants when the current staff
                                      is better.  What's the matter with
2000/6/12-14 [Uncategorized] UID:18451 Activity:kinda low
6/12    Help. For some reason I've been getting call spammed by recruiters
        in the past few days. They have somehow gotten my home phone number.
        How do I make them stop?
        \_ sufficiently related:
        \_ tell them in no uncertain terms to go eat shit and die.  If they
           call again tell them you will find out where they live.
        \_ Hassling phone calls from a single organization are subject to $500
           fines.  You have to tell them that you want to be placed on their
           do not call lists, get their name and record the time.  If they
           call you again you can nail them.  I forget how though- I've never
           had to go that far.
        \_ "What's your name?  What company?  One second while I add you to
            my list of people I'll be certain to never work through again."
        \_ Same way to stop other spam. Change your phone number.
2000/6/12-14 [Uncategorized] UID:18452 Activity:low
6/12    So i am looking at a packet, the header info seems right, it gets
        an ack packet from where it is being sent but it doesn't do what
        it is supposed to when it gets there.  Ether-Peek doesn't seem to
        let me read what is inside the packet (except in hex).  Where can
        I get a hex reader? OR is there way to do this in Etherpeek that
        i just haven't found? (or gnusniff or whatever, i'm not particular)
        \_ use snoop on solaris.
                \_ tcpdump -vvv will do this too, I think.  -John
        \_ M-x hexl-mode
        \_ Convert hex in your head, like a real man.
                \_ Convert?  You mean you don't already think in hex?
2000/6/12-14 [Transportation/Motorcycle] UID:18453 Activity:nil
6/12    Where can i get cheap motorcycle insurance that is NOT McGraw?
        \_ progressive came in pretty cheap for me -shac
        \_ Call felder -  I'm using some offshoot of
           Sullivan&Sullivan at under $300 for full on my boxer.  -jor
Berkeley CSUA MOTD:2000:June:12 Monday <Sunday, Tuesday>