Hard drive benching, Intel App Accelerator

[theDuDe]

Hey guys, just did a round of testing on some new drives and thought I'd share my finds with you fine people...

I've got 2 new drives in the new ANUS: an 80gig Maxtor D740X with fluid bearings, and the new Western Digital 120gig special edition with 8MB buffer.

According to Sandra on win98SE, (maxtor as boot) looks like the WD beats the Maxtor in every department. Both drives formatted for 32k clusters. With only Intel Chipset Installation Utility installed:

WD: aggregate score = 32259
Read buffered= 82 MB/s
Read sequencial= 47 MB/s
Read random= 8 MB/s
Write buffered= 75 MB/s
Write sequencial= 43 MB/s
Write random= 15 MB/s
est access time= 6ms

Maxtor: aggregate score = 25722
Read buffered= 73
Read sequencial= 38
Read random= 7
Write buffered= 71
Write sequencial= 30
Write random= 10
est access time= 7ms

I then also installed UltraATA drivers and benched, then I tried the new Intel Application Accelerator (v2.2). In a nutshell, the benched numbers fell slightly, or at best held the same in some areas. But! ...

I then tried some 'real world' testing, and things look very different. I took a 1 gig folder and copied it from one drive to the other and back, while watching TaskInfo2000 report on the cpu usage and FAT read/write k. The results:

With Intel Application Accelerator 2.2 the cpu usage was way lower (in the range of 15-20%) and the actual FAT read/writes were very stable in the area of 22MB/sec.

On the other hand, with the standard Intel drivers, cpu usage jumped around wildly in the range around 30-60% and read/write fluctuated roughly in the area of 16-20MB/sec.

Finally, the UltraATA drivers (which came before the Intel Application Accelerator are are considered updated by it) did slightly better than the standard drivers, but not as efficient or as constant... the numbers jumped around a lot.

Wish I had a benching utility that would give true max streaming throughput AND measure cpu usage and fluctuation! But based on this, I'd recommend the WD over Maxtor. And, at least speaking for 98SE, IAA is a good thing although the benchmarks may not reveal it's advantage.

theDuDe

 

[Tommy P.]

Excellent work! I always wondered how good that 8Mb buffered WD drive was. The Maxtors usually bench a little higher than that, I wonder if the fluid models are slower. Here, try the demo version of HD Tach its always worked well [="http://www.tcdlabs.com/hdtach.htm"]TC Labs[/]

Tommy P.

The demo only runs in Win98

 

[SonOfSmawg]

Hmmm ... so it would be best to use a 80GB WD SE (8Mb buffer) for the system drive, and a 120 Gb WD SE (8Mb buffer) for the audio drive. A little pricey, but apparently worth it.

 

[Opus2000]

Ummm...guys..hate to rain on your parade here but to be honest Sandra SUCKS for hard drive bench marks...go to http://www.voodoofiles.com and download the free program called FreshDiagnose...that will truly tell you how well the drive performs...do that and come back with those results...also the IAA doesn't do much..it actually inhibits you from changing anything and does drop your read and write performance by 5MB/s on your read and write time.,.....
food for thought..
otherwise gald you finally got your ANUS working properly!!
Opus

 

[theDuDe]

Well, as far as IAA, true it doesn't do a lot in terms of throughput. But it does a huge amount in terms of reducing CPU overhead. And that's not Sandra... that's watching real cpu usage while copying large data transfers.

But for throughput I'll try FreshDiagnose... think I've got it here already.

theDuDe