1/17 My DMA is in fact turned on, but I'm hardly getting the 133 rate
in my ATA133 as claimed by my motherboard and hard-drive. What
are some culprits in the lack of performance? ok thx.
\_ How are you measuring the throughput?
\_ You know nothing can sustain ATA133 throughput, right?
\_ And you're running which OS?
\_ Bottleneck is hard drive heads reading data off the spinning
platter. I think you max out around 20-30 MB / second these days
for a 7200rpm drive for not too fragmented large files.
Heck, notebooks have ATA133, why are their hard drives slow? Duh.
\_ I've tried transfering 50Gig of content (mostly movies,
very little HD head movements) and according to the XP
estimation it will take 160 min. So that means the
rate is approx 5.2M/sec. That is nowhere close to 20-30MB
rate you're saying. I'm very dissappointed...
\_ if your source hard drive is defragged and your dest
hard drive is empty, you are not doing this across a 100Mbps
network or some crappy FireWire / USB 2.0 controller, and
5.2 MB/s turned out to be the user-measured result (as opposed
to what Windoze estimates), I'd be disappointed too.
\_ yes. Drive 1 is Maxtor 250Gig drive, empty.
Drive 2 is WD 200Gig drive, no fragmentation.
Both are on the same ATA133 cable. Transfering
from 2 to 1 is about 5MB/sec. WHAT IS GOING ON?
Why is it so slow???
\_ That *is* slow. Do you have any anti-virus apps doing
on-the-fly scanning?
\_ nope! Disabled the "real-time" scan. ARGGGGGGGGGGG
\_ Okay, now why didn't you give this information from the
start? If they're on the same cable, you're running in
half-duplex. So your theoretical best is half the max.
\_ no. 133ATA means 130000Mb/sec, or 16.6MB/sec.
So theoretical limit is 16.6, and assuming
half-duplex, it's 8.3MB/sec. I'm getting 5.2MB/sec.
I guess 5.2 out of 8.3 is not too bad, but still
very disappointing. -op
\_ 133 means 133 MB/s
http://en.wikipedia.org/wiki/List_of_device_bandwidths
Anyway, re-install Windoze and measure again.
\_ Is your NTFS drive compressed?
\_ NOPE -op
\_ Did you check your primary/secondary IDE channel settings
(in Manage Hardware) and verify that DMA is in fact
selected?
\_ I have run performance tests on literally dozens of
drives and they are all like that. 5MB is kind of slow,
but you are never going to see 10MB/sec.
\_ I just timed a dump of a 2 GB DVD image from drive
to /dev/null on my Thinkpad T41. Average 20 MB/s, and
watching a realtime monitor it was more like 24 MB/s
and then a pause at 0 MB/s while Linux did some
poorly timed swapping. I am sure I get better speeds
than OP copying from drive to Ipod over USB2.
\_ There's also seek time when the sectors you try to read are non-
contiguous in the same track, and when the sectors are in
different tracks.
\_ I believe it's called "latency" when the data resides in
the same track. Seek time is moving heads, as you noted.
Access time is the overall result.
\_ The only time you'll see 133 is when it's reading data that's in the
drive's memory buffer.
\_ if the little pixies are on strike there will be some slow down
\_ Dude where are the true SCSI-files? ATA has always make outrageous
claims. It is a combination of the fact ATA just not "robust"
enough to support that many drives and Windoze sucks at FS. |