HDD Xfer Rate Problem

[Ritual]

Just bought Maxtor 160GB PATA drive to replace my old drive that went
bad. I burned a DVD for the first time today and had it fail. I burned
the DVD at 8x speed, which is ~11Mbit/sec, half the speed the burner
is capable of. I tested the transfer rate of the drive to find it
average 15Mbit/sec on a 4GB file. This seems way slow. I can't find
average transfer rates for PATA drives on the net. What is average?
I'm pretty sure DMA is working but can't be sure. DMA is enabled in
BIOS and shows up for the floppy in SystemInformation in WinXP SP2. P4
2.4Ghz w/256megs RAM. I'm starting to wonder if something is wrong
with the Asrock motherboard, as half my DDR400 memory disappeared,
leaving me with 256MB.

If anyone can tell me what sustained transfer rates I should expect
with a relatively modern motherboard and a new PATA hard drive, please
do. If anyone has suggestions on how I could fix the problem, please
do.

- Ritual

 

[Rod Speed]

Just bought Maxtor 160GB PATA drive to replace my old drive that
went bad. I burned a DVD for the first time today and had it fail. I
burned the DVD at 8x speed, which is ~11Mbit/sec, half the speed
the burner is capable of. I tested the transfer rate of the drive to
find it average 15Mbit/sec on a 4GB file. This seems way slow.

How did you measure that ? Try HDTach.

I can't find average transfer rates for PATA drives on the net. What is average?

Should be a lot higher than that.

I'm pretty sure DMA is working but can't be sure. DMA is enabled in
BIOS and shows up for the floppy in SystemInformation in WinXP SP2.
P4 2.4Ghz w/256megs RAM.

I'm starting to wonder if something is wrong with the Asrock motherboard,
as half my DDR400 memory disappeared, leaving me with 256MB.

Yeah, that's likely the problem.

If anyone can tell me what sustained transfer rates I should expect with
a relatively modern motherboard and a new PATA hard drive, please do.

If anyone has suggestions on how I could fix the problem, please do.

Check for bad caps on the motherboard. These are the usually
blue or black plastic covered post like things that stick up
vertically from the motherboard surface. The tops should be
flat. If any have bulged or have leaked, thats a bad cap.

 

[Ritual]

How did you measure that ? Try HDTach.

I measured it using a freeware program called DiskBench. Its simple..
from the looks of it, it just transfers a prompted file and
destination and measures the time used. HDTach loads special modes
that bypass the filesystem, which I don't think DVD burning software
does. Still... even with the XP overhead, I should be getting better
than 15Mbit/sec.

The program I used prompted for a file and a destination. It copied my
4.5GB file and measured the time and reported the results. I kind of
like the fact that the program had no fancy install that almost every
program out there has now. I love the elegance of small code. Utorrent
has quite an impression on me as a former programmer.

Should be a lot higher than that.

Tell me more. Be more specific. I'd run HDTach if I could see a
database of other ppls' results. I can't find anywhere on the net
where I get a straight answer of how long it takes to read a big file
from a hard drive.

Yeah, that's likely the problem.



Check for bad caps on the motherboard. These are the usually
blue or black plastic covered post like things that stick up
vertically from the motherboard surface. The tops should be
flat. If any have bulged or have leaked, thats a bad cap.

By caps, I guess you mean capacitors. I don't know of many capacitors
filled with liquid, but my electrical knowledge is limited. I have
been meaning to go in there and start pulling RAM and shit but the
thought makes me sick.


Thx for reply.

Still... what do YOU get for hard drive Xfer rates on a big file?



- Ritual

 

[Rod Speed]

Ritual wrote

I measured it using a freeware program called DiskBench. Its simple..
from the looks of it, it just transfers a prompted file and destination
and measures the time used. HDTach loads special modes that
bypass the filesystem, which I don't think DVD burning software does.

Its more complicated than that, but what matters is what the hardware
can do, to separate a hardware problem from a problem somewhere else.

Still... even with the XP overhead, I should be getting better than 15Mbit/sec.

Yes, certainly as measured with HDTach you should.

The program I used prompted for a file and a destination. It copied
my 4.5GB file and measured the time and reported the results. I kind
of like the fact that the program had no fancy install that almost every
program out there has now. I love the elegance of small code.
Utorrent has quite an impression on me as a former programmer.

The trouble is that most of us know what HDTach reports for a
drive thats properly installed and not what that particular app reports.

Tell me more. Be more specific.

It varys with what you are using to report that thruput.

I'd run HDTach if I could see a database of other ppls' results.

You'll get plenty of reports of what HDTach gets using groups.google and google.

I can't find anywhere on the net where I get a straight answer
of how long it takes to read a big file from a hard drive.

Thats why most use HDTach, its reports are at least using a commonly used test.

By caps, I guess you mean capacitors.
Yep.

I don't know of many capacitors filled with liquid,

Electrolytics do have an electrolyte, thats why they are called electrolytics,
and the industy has seen one hell of a problem with an attempt to reverse
engineer a proprietary approach to the electrolyte that produced one hell
of a problem with bad caps industry wide.

but my electrical knowledge is limited. I have been meaning to go in
there and start pulling RAM and shit but the thought makes me sick.

No need to retch when just having a look at the caps.

Thx for reply.

Still... what do YOU get for hard drive Xfer rates on a big file?

Like I said, it varys with how that is measured and its desirable to see
what HDTach gets because that eliminates some OS config problems etc.
If the numbers it gets look fine, that is a good indication of an OS config
problem and the best next check is a clean install of the OS on a spare
hard drive etc to prove that its an OS config problem etc.

 

[Bob Willard]

Tell me more. Be more specific. I'd run HDTach if I could see a
database of other ppls' results. I can't find anywhere on the net
where I get a straight answer of how long it takes to read a big file
from a hard drive.

There is a database with >100 HDs built-in to HDtach. DL it, run it,
then click on the Compare button to see how your HD on your PC compared
with other HDs on other PCs.

The advantage of HDtach is that, since it bypasses the Winduhs filesystem,
it measures the speed of hardware, not the OS. Of course, you do need to
run HDtach nearly standalone, so that other apps do not corrupt the data by
stealing CPU cycles.

DiskBench does not measure file read speed, it measures file copy speed,
which can be much much much less than half of file read speed. To each
his own; I am not impressed by DiskBench.

To measure file read speed under Winduhs, first create a really big file;
call it, say, BigFile.tmp. Then fire up a command window (pseudoDOS) and
use your watch to time: Copy BigFile.tmp nul. Since that copy command
just copies to the bitbucket, it is pretty close to a pure file read under
Winduhs. On this PC, I got ~30 MB/s on innermore cylinders of a HD, an area
that is good for (by eyeball) 32-35 MB/s (as per HDtach).

As to your Q of how long it takes to read a big file from a HD, the A is
that it depends:
- it depends on the HD: a huge variable
- it sometimes (but rarely on newer PCs) depends on the PATA/SATA/etc. bus
- it sometimes depends on what else is using the disk bus, and how much
- it depends on whether the HBA/chipset uses DMA or PIO
- it depends on the CPU and the chipset and the RAM
- it depends on the OS and its params, and the location of the pagefile
- it depends on what other apps are doing
- it depends -- a lot -- on the app reading the file
- and so on

Again, use HDtach. Some (many?) of us in this NG will trust nothing else.