Ultra slow HDD -> HDD file copy

My computer has 2 HDDs. C is a 6 GB Udma33 capable Seagate Medallist. D
is a 2.5 GB Western Digital Caviar.

Creating a copy of a 100 MB file on the same drive is very quick as would
be expected. This is the case on both drives

However moving that 100 MB file from one drive to the other is incredibly
slow (30 minutes or more) and will also slow down other things happening
in the computer.

The drives are respectively the primary and secondary master devices and
there are no other devices on the IDE ports (at the moment).

Although this computer should really belong in a museum, it can burn a CD
at 16x without buffer underruns. There seems to be no problem filling the
process buffer fast enough for a burn and performance is back to what it
should be even though the burner (primary slave) is on the same IDE cable
as the drive containing the source ISO.

 

Do you have 80-wire UDMA cable on both primary and secondary cables?
Are your disk drives set to cable-select or master?

What OS are you running? Has DMA been enabled on both drives?

Your 2.5GB WD will probably not be UDMA. So that will slow things down.
WD are probably the worst possible brands in terms of comformity to
standards and compatability issues.

 

He may also have a mobo like the (SIS Chipset) PCChips M810....which has a
serious issue with it's I/O subsystem.

If you have enough RAM for a large cache and the hard drive has a reasonable
amount of cache, you don't often notice the problem.

But if you try to do something like copy a large file from one HD to another
- something much larger than the HD or RAM disk cache - you see the
bottleneck in all it's deeply flawed glory.

This is why these mobo were so cheap and often OEM'd as a "no-name"
M810....with no "PCChips" anywhere in sight.

I bought one....but have since gotten rid of it.

 

No they are not using the 80 wire cable. Does this require 80 pin MB
connectors?

Win98SE, doesn't seem to like DMA being enabled on the burner.

As it stands transfer rate is probably slower than a 10Mbps network card.
At work I regularly transfer 2 GB of data in about 30 minutes over a
wireless link of that speed.

 

Strangely, it now works much faster when both drives are on the same
cable (primary master and slave)

 

No it's an Asus TX97

Probably nothing to do with the chipset - PC chips sux

 

Do you have 80-wire UDMA cable on both primary and secondary cables?
Are your disk drives set to cable-select or master?
[/QUOTE]
80-wire only needed for udma 66 mode and above

 

Strangely, it now works much faster when both drives are on the same
cable (primary master and slave)
[/QUOTE]
If you had 1 drive coupled to an optical drive not in DMA mode that
would slow the HD down. Independant IDE timing requires both to be DMA
capable

 

80-wire cables are only needed for ATA66 and above. Neither of his drives
are rated that fast and I doubt his IDE controller is faster than ATA33.

 

If you had 1 drive coupled to an optical drive not in DMA mode that
would slow the HD down. Independant IDE timing requires both to be DMA
capable[/QUOTE]

In general it seems that devices on different ports have much slower
transfer of data between them.

When I tried the burner on the secondary with the HDDs both on the
primary the flextralink burnproofing kicked in a lot more often.

I would have expected the transfer rate to be faster between the primary
and secondary than between two devices on the same cable - unless maybe
most boards can't read the primary and write the secondary more or less
simultaenously.

 

I was asking if he had 80-wire cable.

 

80-wire only needed for udma 66 mode and above[/QUOTE]

I was asking if he had 80-wire cable.

 

It helps avoid un-necessary chatter if questions are briefly justified.
Justification also help the end user with their own troubleshooting steps
as it often triggers a different approach.
Ciao, Dave

 

Why?

 

I was asking all the questions I could think of because the post
was so devoid of detail. No mention of the os, etc.

 

pot kettle black!

 

Ok, it's just that the question was irrelevant. Regardless of OS, an 80-wire
cable won't make an ATA33 drive any faster.

 

It will cut out noise on the cable and possibly eliminate sporadic errors.
It provides better transmission characteristics, and better shielding
for electrical noise. An 80-wire cable has every second wire connected to
earth - if you want a high quality system then having 80-wire cable helps.

You say "... and 80-wire cable won't ...". Well it could do if the
cable was running right next to the high-frequency transformer of
the power supply!

Of course I could also have asked "how long is the cable" but that is
just one of many hundreds of things that a person needs to be cognisant
of when fault finding.

Perhaps I should have asked "does the cable run right next to a noisy
electrical component".

Perhaps I should have asked "is the drive recalibrating frequently"?

It is next to impossibly to diagnose a problem without vaste amounts
of information. "Why is my disk drive slow" is a silly question and
most likely will never be solved in discussion groups like this.

 

I know. However, from the OP it would seem that this is a less than
state-of-the-art machine.

Then why not suggest re-routing the cables instead of asking if he has
cables that weren't even around when that machine would have been built?

I have seen very few cables longer than the ATA (or whatever) max spec of
18" (I think)

See above.

I beg to differ. I hang in alt.comp.hardware and problems like this are
solved by the regulars frequently, at the rate of several a week at least.
Often the first replies ask for more detail but I've never seen anyone ask
if an old machine, running sub-5GB HDDs on an ATA33 system is fitted with
80-wire cables that (probably) weren't available at the time of manufacture.

 

Asus have made shonky mobos, too. I bought one a few years back that had an
I/O bottleneck of significant proportions in it's VIA Apollo chipset.