What's better: ATA/33 w DMA or ATA/66 w/o DMA ?

[Michel S.]

Hi !

I'm working on one friend's "old" computer (Compaq Presario 5304),
running Windows 98 se.

It has a Quantum Fireball CR 4.3 Gb hard disk.

On Maxtor site, there's a utility offered for this disk, allowing to
fix a problem with some BIOS (including mine), where the "DMA" box in
the disk properties can't be set properly.

The program allow to change the disk from Ultra ATA/66 to Ultra ATA/33
and back.

Under Ultra ATA/66, the DMA check box can be set, but it will be
uncheck on next reboot. Under Ultra ATA/33, it stays checked.

I didn't noticed significant performance difference between the two
modes, but I didn't made extensive testing.

In theory, which mode is better ?

Thanks.

 

[J. Clarke]

Hi !

I'm working on one friend's "old" computer (Compaq Presario 5304),
running Windows 98 se.

It has a Quantum Fireball CR 4.3 Gb hard disk.

On Maxtor site, there's a utility offered for this disk, allowing to
fix a problem with some BIOS (including mine), where the "DMA" box in
the disk properties can't be set properly.

The program allow to change the disk from Ultra ATA/66 to Ultra ATA/33
and back.

Under Ultra ATA/66, the DMA check box can be set, but it will be
uncheck on next reboot. Under Ultra ATA/33, it stays checked.

I didn't noticed significant performance difference between the two
modes, but I didn't made extensive testing.

In theory, which mode is better ?

Can your disk at the electromechanical level transfer data faster than 33
MB/sec? If not then 66 is irrelevant. DMA doesn't have a large effect on
disk throughput, what it _does_ affect is CPU utilization--watch that with
and without DMA enabled and I think you'll see a marked difference.

 

[Egil Solberg]

Hi !

I'm working on one friend's "old" computer (Compaq Presario 5304),
running Windows 98 se.

It has a Quantum Fireball CR 4.3 Gb hard disk.

On Maxtor site, there's a utility offered for this disk, allowing to
fix a problem with some BIOS (including mine), where the "DMA" box in
the disk properties can't be set properly.

The program allow to change the disk from Ultra ATA/66 to Ultra ATA/33
and back.

Under Ultra ATA/66, the DMA check box can be set, but it will be
uncheck on next reboot. Under Ultra ATA/33, it stays checked.

I didn't noticed significant performance difference between the two
modes, but I didn't made extensive testing.

In theory, which mode is better ?

What matters is which transfer mode is really used. If you get PIOmode4 in
the case that the checkbox does not stay checked, it is best to try ATA33
(UDMA2) instead.
The disk's transfer rate is probably so low in any case that you may not
notice it, unless you need your CPU for other stuff. Running PIO4 is
seriously more taxing on the CPU than UDMA2.
Check both scenarios with HdTach.

 

[Folkert Rienstra]

Can your disk at the electromechanical level transfer data faster than 33
MB/sec? If not then 66 is irrelevant.

Not with RAID it isn't.

DMA doesn't have a large effect on disk throughput,

Oh yes, it does.
PIO is limited to 16MB/s and in practice you won't even get close to that.

 

[Folkert Rienstra]

What matters is which transfer mode is really used. If you get PIOmode4 in
the case that the checkbox does not stay checked, it is best to try ATA33
(UDMA2) instead.

The disk's transfer rate is probably so low in any case

If it is capable of ATA66 then it is able to do at least 16MB/s or more,
30MB/s at most.

that you may not notice it, unless you need your CPU for other stuff.

Can't follow that. What has that got to do with UDMA2/ATA33?

Running PIO4 is seriously more taxing on the CPU than UDMA2.

PIO is also much less efficient in using the bus with all the handshaking so
you won't get the same transfer rate as with DMA. (And that's DMA mode 2
not UDMA mode 2). DMA mode 2 is 16MB/s, PIO4 is even less, more like 8-10.