SiSandra & ATTO HD Scores Contradict

[Unimatrix X]

Ok I've got an IBM DTLA -307030 (30GB 7200RPM UDMA100) Hard Disk and I did a benchmark on the drive using SiSandra & ATTO.

As you can see SiSandra says one thing...


My Specs:-

AMD T'bird 1GHz "Type B"
Abit KT7A-RAID (266MHz FSB)
512Mb Crucial PC133 RAM (Cas2)
ATI Radeon 64MB VIVO
30.7GB IBM 75GXP
Hercules Game Theater XP
Plextor 24/10/40 CDRW
Pioneer 16xDVD Drive
Windows XP
DirectX 8.1

[Unimatrix X]

....And ATTO says another!

I'm hoping ATTO is right!

[NDD]

Man they both BAD !!!
Are you running this HD in UDMA-5/ATA100 mode ? With DMA enabled ?
I have almost the same system, same mobo and CPU running WinXP on 45GB 75GXP in a matter of fact, and my scores are AT LEAST 2 TIMES BETTER ! Although I got mine connected to first RAID port, connected to Primary IDE the scores are almost similar.

Sorry I can't perform any benches and present any pictures right now, got 72% of DiVX file encoded, and I won't be here when it's done, but I'll post mine as soon as possible.

[Unimatrix X]

What the hell I thought somthing weren't right.

I've connected it to the first RAID port too and I have no way of checking if it's running at UDMA5/ATA100 mode in XP.

I updated my Highpoint Drivers to 2.31.0.0 dated 25/01/2002.

Although when I enter the Highpoint BIOS/HD setup it does say its in UDMA5 mode.

So your getting about 50000 points then!

[NDD]

"One picture worth more than thousand words" they say, so here're some benches of mine ;)

Sandra 2002.6.8.97 (scores seem to vary from one version to another, I wouldn't take Sandra as a serious benchmarking program) and WinXP, IBM 75GXP 45GB, system is up for about 4 hours, mostly Inet surfing and MP3s listening

[NDD]

And ATTO 2.02 ...

[NDD]

Same system few months ago, Win98SE and Sandra 2001 :

[NDD]

And HDTach 2.61 ...

[NDD]

I've PMed otheos, he's a real PRO to take care of the things we have here ... 4ms access time doesn't seem right to me :rolleyes:

[Unimatrix X]

Nice work with the Pics ND, I'll try HDTach 2.61 and report back

[Unimatrix X]

Here it is....

[Unimatrix X]

Man your read burst speed is off the chart and mine just goes to 49.2mbps, what the hell is going on.

If your runing XP, did you install any extra software for the HD?

[Unimatrix X]

Any takers on this, are the scores out of wack!

[otheos]

Woa, ok quite a few different results here so I'll try to have a go :)

It is true that sandra is lousy for HD benchmarking. Usually you need HDTach and ATTO to determine if the drive works as it should.

Now Sandra, uses a "score" which is a mix of various performance figures one of which is Burst speed. Remember burst speed only measures how "wide" the data path is to the controller, and for IDE drives it makes little or no performance difference (hence ATA66/100/133 all perform the same).

Ultimatrix X first Sandra plot:

This is a PIO4/DMA2 mode plot. I don't know how you managed but data is capped at 15MB/s and if we had a CPU useage figure we could tell if it is PIO or DMA (the former uses far more cycles).

The ATTO benchmark looks normal but it's too flat for the various sizes that makes me wonder.... Is this the same setup as with the Sandra benchmark, and was it done immediately after the Sandra test? did you alter anything in between? Cause the results are definitely different (but you didn't need me to tell you this), there could be a reason however.

And then the HDTach plot comes which could enlighten us a bit. You say you have a DVD and a CDRW on this box? How are the drives connected? Is DMA enabled on all drives? This I think will explain the discrepancies.

THe HDtach plot shows four things: the drive physically performs ok (sequential reads are up to spec). However the low burst rate means that it either is set in the bios in ATA66 mode, or that something is starving the bandwidth from it (I'll come to that). The high CPU useage clearly shows a driver issue that needs to be fixed (it shouldn't be more that 10% at most) that may explain everything else too, and finally the extremely high access times mean that something adds latency to the bus.

Now the KT7A uses the 686B southbridge which is ATA100 capable but is known not to switch speed very efficiently. Starting from the BIOS settings, all IDE ports should be set to AUTO or max or UDMA5, data prefect and 32bit access to enabled. Now after POST press pause and check the summary screen to see what the drives are reported by the BIOS.

Now if the drives are all connected to the onboard IDE and not the "Raid" controller, then at least one optical will be on the same channel as the HDD. Check in windows that both drives on the channel as set to DMA mode and NOT PIO. Is the DVD rom connected with the HDD? Cause it seems it is as the Pioneer 16 is an ATA66 (not 33) device and the HDTach plot seems to verify that you're stuch at 66MB/s max bandwidth. Not a problem otherwise. ATA66/100 should give identical ATTO benchmarks as there's plenty of room for the 35MB/s peak the drive is physically able to deliver.

Now this could explain the low bandwidth but does not explain the high CPU usage and high access time. It looks your driver is not behaving properly and at this stage with 2 known facts taht don't work as they should (IDE controller doesn't switch and driver is messed up) having weird results all over should be expected.

So take the following acction:

Leave the HDD as the only (master) device on primary.
Enter bios set IDE0 to UDMA5 and enable data prefetch and 32bit access mode.
Make sure your cables are not bent, cut or damaged and that all pins are inserted into the sockets. They should be 80wire 40pin cables, no longer than 12"
Check the summary screen and see if your hard drive is reported as UDMA5 (ATA100).
Boot to windows and run VIA4in1 to uninstall the IDE driver. Then run it again and leave the IDE driver unticked (i.e use MS driver). Reboot as necessary.
Check the device manager in the IDE controllers are running at DMA mode. They should either say DMA (UDMA) or N/A (if no device is connected).
Check the HDD in the device manager in case it is accessed in DOS mode (unlikely but check).
Reboot and run HDTach.
Reboot and run ATTO
Reboot and run Sandra

Post results here :)

[Unimatrix X]

Ok Thanx otheos,

My Drive status:-

Primary Master IDE0 = Plextor 24/10/40 CDRW (DMA on)
Secondary Master IDE1 = Pioneer 16xDVD (DMA on)
RAID IDE2 = IBM DTLA -307030

My properties are here, is it right for the IBM Drive to be listed as a SCSI device?

I'll do what you said and report back!

[NDD]

HD will be listed as "SCSI" if connected to RAID port. Mine is the same way.

[Unimatrix X]

Ok here are the scores...

[Unimatrix X]

HDtach

[Unimatrix X]

Sisandra

[otheos]

OK, you're back in business as far as I can see it.

Are you running anything in the background? CPU intensive cause your CPU usage seems a bit high for my liking.

As for the 67MB/s Burst speed, nothing to worry, as long as it shows it's more than 66MB/s (and it is) it means the drive works at ATA100 and will use the bandwidth when needed.

[Unimatrix X]

Ok otheos Thanx 4 your help!

one last question, could I use the CDRW/DVD on the secondary IDE as master/slave and use the HD on Primary IDE.
I ask because I though you couldn't use two CDROM drives on the channel!


Hmmm, that avatar aint you is it?

[NDD]

Having two ATAPI devices on the same cable shouldn't be a problem in your case, especially coz your Plextor is "Burn-Proof" capable. Be sure to leave it as "Master", though ...
But why would you want to do it ?! Your problems are seem to be solved :confused:

And it's not otheos himself, it's his baby son :)

[otheos]

And it's not otheos himself, it's his baby son

:D :D :D