Asus A7V133: Promise ATA-100 slower than IDE

[smithpd]

I just assembled the following system:
Asus A7V133 rev. 1.5
Athlon Thunderbird 1.2
256 MB Crucial PC-133 CAS-2
IBM Deskstar 75GXP 60 GB (DTLA-30-70-60)
Geforce II Pro

I installed the latest bios and Promise drivers from Asus:
Asus 1004
Promise 1.60 (Build 33)

My problem is that the Deskstar runs SciSoft Sandra 2000 disk drive benchmark slower in the ATA-100 slot than in the normal IDE slot.
IDE score: 23063
ATA-100 score: 13270 before software updates
ATA-100 score: 16606 after bios/driver updates

Same hard drive, same benchmark in ASUS P3V4X w/ ATA-66: ~26000 !!

It is not just Sandra. I noticed a factor of two difference in load times for loading Thief II missions, which are mostly disk I/O.

Other than the above, the system runs fine. All other benchmarks (CPU, memory, 3-Dmarks) are up to snuff (3-DMark2000 = 7750).

Does anyone have any idea what is going on here?

 

[chainbolt]

<< I just assembled the following system:
Asus A7V133 rev. 1.5
Athlon Thunderbird 1.2
256 MB Crucial PC-133 CAS-2
IBM Deskstar 75GXP 60 GB (DTLA-30-70-60)
Geforce II Pro

I installed the latest bios and Promise drivers from Asus:
Asus 1004
Promise 1.60 (Build 33)

My problem is that the Deskstar runs SciSoft Sandra 2000 disk drive benchmark slower in the ATA-100 slot than in the normal IDE slot.
IDE score: 23063
ATA-100 score: 13270 before software updates
ATA-100 score: 16606 after bios/driver updates

Same hard drive, same benchmark in ASUS P3V4X w/ ATA-66: ~26000 !!

It is not just Sandra. I noticed a factor of two difference in load times for loading Thief II missions, which are mostly disk I/O.

Other than the above, the system runs fine. All other benchmarks (CPU, memory, 3-Dmarks) are up to snuff (3-DMark2000 = 7750).

Does anyone have any idea what is going on here? >>




LOLOLOLOLOLOL:

You know what? I have EXACTLY the same &quot;problem&quot;. I have checked it in many ways, with a Promies Fastrak 100 PCI , with a Promise Fastrak 100 TX PCI, and with the on-board controller of my
mobo. Sometines I get a little better preformance, sometimes even less preformance in RAID 0 versus ATA 100. I got quite frustrated here. My Sandra scores for the RAID 0 set up are ranging from 24,000 to 35,000.

And you know what? The problem is definitely Sandra, not your RAID set-up. Use Mem Tack or other benching software.

 

[smithpd]

chainbolt,

Thanks a lot. I will indeed try one of those other benchmarks. However, I am not using Raid 0. I am using straight ATA-100 (jumper set to default - I checked it). Would that make any difference in this bug?

How can you explain Thief II missions loading more slowly? That's what got me onto the benchmarks in the first place.

 

[AquamanA]

Check your performance with hdtach.

I found the following on my a7v133 1.04:

Disk performance (in WinMe) was about the same with promise/via controller; however, the cpu utilization was much higher with the promise controller.

I would be interested in your posting your results with hdtach

 

[chainbolt]

<< chainbolt,

Thanks a lot. I will indeed try one of those other benchmarks. However, I am not using Raid 0. I am using straight ATA-100 (jumper set to default - I checked it). Would that make any difference in this bug?

How can you explain Thief II missions loading more slowly? That's what got me onto the benchmarks in the first place. >>




Sandra is definitely not apporpriate to check HDD and in particualr RAID performance.
The recommended HDDTach is better/more precise.

Wait: you are using only 1 HDD? And that is connected to the RAID IDE connectors? I wouldn't do this.

 

[smithpd]

No, it is not raid 0. It is normal ATA-100 without raid, according to the manual. There is a jumper setting that controls which is selected, and mine is at the default setting, which is ATA-100 (no RAID).

I will try those other benchmarks tomorrow.

 

[seind]

chainbolt

Wait: you are using only 1 HDD? And that is connected to the RAID IDE connectors? I wouldn't do this.

Can you inform us the reason why? I'm using my hdd on the promise controller (single drive, ata100), and though i did not bench, it 'feels' quicker than the ide controller. Only annoying thing it's the time it takes to load the controller bios.

seagate ata66 30gb 7200

cheers

 

[seind]

I benched with hdtach 2.61

Only visible difference was a 0.4% more cpu utilization while the disk was on the promise controller. Otherwise, everything was almost identical.

cheers

 

[smithpd]

Here are my benchmark scores

ATA-100 port, single Deskstar drive, ATA-100 jumper enabled, after bios and driver update
CD-ROm normal IDE secondary master
SciSoft Sandra 2000 - 16449
HD Tach 2.61
38751 max, 13208 min, 29383 average, 13.1 access, CPU utilization 4.6%
Load Sabotage at Soulforge (Thief 2) first time, accesses CDROM - 28 s
Load Sabotage at Soulforge (Thief 2) from quick save - 20 s

Normal IDE port ports, Deskstar primary master, CDROM secondary master
SciSoft Sandra 2000 - 22563
HD Tach 2.61
38761 max, 18218 min, 29410 average, 13.1 access, CPU utilization 7.8%
Load Sabotage at Soulforge (Thief 2) first time, accesses CDROM - 20 s
Load Sabotage at Soulforge (Thief 2) from quick save - 18 s

OK, it looks like the Sandra is out of line, but not so much as before the bios and driver upgrade. Maybe Sandra includes write tests, which HD Tach does not, in unregistered version? Also, it looks like the normal IDE uses more CPU. OK, that could be bad, but in my applications (games) the CPU isn't doing anything useful anyway during the mission loading, so it may as well help the disk drive run faster.

The bottom line is that no matter how you measure it, the normal IDE is somewhat faster. That's what I am using, but I wish it were the other way around and I could get up to the 26000 Sandra 2000 that I got on the ASUS P3V4X.

Any comments?

 

[smithpd]

Has anyone else experienced a slower response on the Promise ATA-100 port than the normal IDE port, or am I alone in this? See my benchmarks above.

 

[chainbolt]

I'm playing around with a Promise Fasttrak 100, a Promise Fasttrak 100 TX, and I have the
Promise Fasttrak 100 Lite onboard controller in my GA-7DXR: and I also sometimes get slower scores when I connect a single HDD to the RAID IDE, instead of the regular IDE. That is the reason why I recommneded not to connect a single HDD to the RAID IDE. I don't know why, it's just trial and error. The Promise controller is marvelous fast though, when running in RAID 0 with 2 or more HDD. I'm still waiting to get my 4th WD Caviar
to set up a 4 HDD RAID 0.