PDA

View Full Version : Hard drive benchmark with HDTach



Johnny Ray
12th January 2001, 19:54
Here's my HDTach score for a Cheetah X15 on an Adaptec 2940U2W, with a PIII at 825 MHz:

http://home.earthlink.net/~carneyj/_uimages/Image5.jpg

What's your score?

John

P.S. I like my Cheetah

SCompRacer
13th January 2001, 21:32
Way cool Johhny Ray.

A third generation Cheetah...
http://www.enteract.com/~richol/forums/hd_tach_03Chta.jpg

I am pondering going ATA100 IDE with RAID rather than update all my SCSI gear. I'd need the card, the drive, $650 or so minimum.

Decisions, decisions...

------------------
ABIT KT7 non-RAID * 900MHz Athlon T'bird * 128MB Crucial 7E PC/133 RAM * Two Cheetah LVD's * Barracuda UW * DiamondMAX IDE * Plextor Ultraplex 40max/Plexwriter 12/4/32 * Hitachi IDE DVD * 2940U2W * SB Live * 3Com 905B-TX NIC * 3Com Courier V. Ext. * Hollywood + * Win 98SE, Win 2000 *


[This message has been edited by SCompRacer (edited 15 January 2001).]

Indiana
14th January 2001, 17:07
My old IBM Ultra-SCSI DDRS 39130D with an Initio UW-SCSI hostadapter - FAT32:

http://www.indiana.claranet.de/HDTachDDRS.png

My U-DMA RAID (RAID 0 with the ABit KT7-RAID's HighPoint controller and two IBM DTLA 307030) - NTFS:

<img SRC=http://www.indiana.claranet.de/HDTachRAID.png>


I love my RAID http://forums.murc.ws/ubb/biggrin.gif


[This message has been edited by Indiana (edited 03 April 2001).]

spazm_1999
14th January 2001, 20:08
My score is far from being as good as yours, but it's not that bad for a IDE Quantum 14g.

http://www.geocities.com/spazm_c/hdtest.JPG

Spazm

FrankDC
14th January 2001, 21:55
Johnny, your burst speed seems a bit low. I get 70.6mbps on my X15. What else do you have on the bus?

Johnny Ray
15th January 2001, 19:54
Thanks for the comment Frank. I agree, it does seem low. There's nothing else on that channel (i.e., it's the only device on the LVD channel). Is your 70 MBS on an 2940U2W adapter? Any suggestions on what to try to bump it up?

Thanks all in advance for the tips.


John

FrankDC
16th January 2001, 00:23
Yes. I recently upgraded from a 2940U2W to 29160, but the 70mbps is consistent on both. This is on a PIII @ 840MHz.

What OS are you using? I noticed a substantial performance boost once I installed W2K SP1, compared to Win98 SE. (Something to do with the fact that command queueing is supported under W2K but not 9x, from what I've been told.)

Guru
16th January 2001, 02:51
2xIBM 30GB raid 0 Abit bx 133 raid!
http://personal.inet.fi/clan/setimurc/HDTACH.jpg

Only 128MB mem in system + Seti runing in the background therfore the low score!

------------------
Join the MURC SETI team! (http://setiathome.ssl.berkeley.edu/cgi-bin/cgi?cmd=team_join_form&id=25678) | SETI @ MURC (http://www.setiatmurc.f2s.com/)

All extremists should be taken out and shot.

Johnny Ray
16th January 2001, 20:10
Frank, thanks for pointing out to me the obvious. My score is from WinMe. I should have indicated that in my original post. Out of curiousity, I did manage to get a few more MB/s by playing around with my PCI latency, but hardly worth the trouble.

And Guru, I'm surprised it's taken this long for someone with a RAID rig to post! http://forums.murc.ws/ubb/smile.gif Nice score.

Come on all, let's see some more! It's nice to see what's out there and what one can expect from the various hard drive set ups.


John

Shoyu
19th January 2001, 06:13
Just curious, do we all know that some of these scores are inaccurate due to "Advanced Size Check" not being checked?

HDTach only tests the first 7.7GBs if it isn't checked, as shown in some of the screenshots.

Electric Amish
19th January 2001, 07:54
I get BSOD's every time I try to run HDTach.

amish

Himself
19th January 2001, 08:33
http://www3.nf.sympatico.ca/byronm/images/hdtach.jpg

5400rpm quantum, ata 33, yee haw. http://forums.murc.ws/ubb/smile.gif

[This message has been edited by Himself (edited 19 January 2001).]

Ovi
19th January 2001, 09:49
<font face="Verdana, Arial, Helvetica" size="2">Originally posted by Electric Amish:
I get BSOD's every time I try to run HDTach.

amish</font>

This is a known problem with HD Tach ver 2.61. To eliminate this, you'll have to disable auto-insert notification for your CD Drive, or use a previous version of HD Tach.
Cheers
Ovi

SCompRacer
19th January 2001, 19:05
Just got my drives today and this is my Atto score out of the box for my RAID setup.
http://www.enteract.com/~richol/forums/atto_011901.jpg
http://www.enteract.com/~richol/forums/hdtach_0120.jpg
HD Tach 2.61 won't run with advanced test checked, and Ovi has provided the reason why, so I'll research my Archive CDR's for earlier version.

------------------
ABIT KT7A, RAID 0 * 900MHz Athlon T'bird * 256MB Crucial 7E PC/133 RAM * Two 30GB IBM 75GXP * Two Cheetah LVD's * Barracuda UW * DiamondMAX IDE * Plextor Ultraplex 40max/Plexwriter 12/4/32 * Hitachi IDE DVD * 2940U2W * SB Live * 3Com 905B-TX NIC * 3Com Courier V. Ext. * Hollywood + * Win 98SE, Win 2000 *


[This message has been edited by SCompRacer (edited 20 January 2001).]

xortam
27th April 2001, 10:15
Indiana ... I have the same drive as you, IBM DDRS 39130D U2W LVD 9GB formatted with FAT32, which I just benched under Win98. I got slightly higher numbers than you using my on-board U2W SCSI (ASUS P2B-S):

Read max: 13,585 kbps
Read min: 11,201 kbps
Read ave: 12,632.9 kbps
Read burst: 46.9 Mbps
Random access: 9.9 ms
CPU: 1.7%

Advanced size results:

Read max: 13,585 kbps
Read min: 8,162 kbps
Read ave: 10822.8 kbps
Read burst: 43.6 Mbps
Random access: 12.4 ms
CPU: 1.4%

Interesting that you reported zero CPU usage. My random access and read burst were much better than yours.


SCompRacer ... were those results from a RAID 0 running two 30GB IBM 75GXPs? Which OS were you running? Which FS was installed?

Indiana
27th April 2001, 12:54
<font face="Verdana, Arial, Helvetica" size="2">Originally posted by xortam:
Indiana ... I have the same drive as you, IBM DDRS 39130D U2W LVD 9GB formatted with FAT32, which I just benched under Win98. I got slightly higher numbers than you using my on-board U2W SCSI (ASUS P2B-S):Interesting that you reported zero CPU usage.
My random access and read burst were much better than yours.</font>

Burst speed and most probably also access are better because you have the LVD version on a U2W controller while mine is an old narrow Ultra-device on a UW-controller.
But yes, I'm quite happy with the Initio, especially looking at the price. I had the IBM connected to a friends Adatec2940UW and he even got slightly worse results than mine.
Plus the Initio works well with everything I've connected including a Mustek scanner (they really have BAD incompatible SCSI-ports) in both, Win98 and 2k.



<font face="Verdana, Arial, Helvetica" size="2">SCompRacer ... were those results from a RAID 0 running two 30GB IBM 75GXPs? Which OS were you running? Which FS was installed?</font>

If you're searching for results of this combo, just look at the second bench I provided: Two 30Gig IBM75GXPs on a RAID0 array (ABits onboard HighPoint) in Win2k with NTFS.

Edit: I've falsely written Inito instead of Initio TWICE, doh! Even though I'm not really sure if the chip comes from Initio or Domex....

[This message has been edited by Indiana (edited 27 April 2001).]

xortam
27th April 2001, 15:03
<font face="Verdana, Arial, Helvetica" size="2">Originally posted by Indiana:
Burst speed and most probably also access are better because you have the LVD version on a U2W controller while mine is an old narrow Ultra-device on a UW-controller.</font>
I see they make five variants of the DDRS-39130: three SE (50, 68, and 80) and two LVD (68 and 80).


<font face="Verdana, Arial, Helvetica" size="2">But yes, I'm quite happy with the Inito, especially looking at the price.</font>
I'm always on the lookout for SCSI controller companies that do quality work. I think Adaptec products are way overpriced. My SCSI RAID controller was made by Mylex (now IBM).


<font face="Verdana, Arial, Helvetica" size="2">I had the IBM connected to a friends Adatec2940UW and he even got slightly worse results than mine.</font>
Interesting, the P2B-S uses the Adaptec AIC 7890 which is the same as the 2940U2W board.


<font face="Verdana, Arial, Helvetica" size="2">If you're searching for results of this combo, just look at the second bench I provided: Two 30Gig IBM75GXPs on a RAID0 array (ABits onboard HighPoint) in Win2k with NTFS.</font>
I saw your RAID results. My friend has a pair of the 75GB IBM 75GXPs. He got the second one keeping in mind to RAID them some day.

I was asking about SCompRacer's setup because a) unlike you, he didn't detail his specs and b) I'm looking to compare against my SCSI RAID 0 setup under Win98. I have two of these DDRS-39130s already RAIDed with my UW Mylex controller. I just picked up a used DDRS-39130 from a friend ... cheap. I was checking out the performance of my UW RAID 0 versus the same drive connected on the on-board U2W controller. I've been debating about leaving the third drive as a separate boot drive versus adding it to my RAID array. I finally decided to go ahead and add it to my array and get a fast Ultra160 drive to use as a removable boot drive (using mobile rack). I can't believe how slow these DDRS-39130 drives have gotten relative to current drives. These drives are only two years old which is considered a baby in my equipment arsenal.

Indiana
27th April 2001, 15:40
<font face="Verdana, Arial, Helvetica" size="2">Originally posted by xortam:
I see they make five variants of the DDRS-39130: three SE (50, 68, and 80) and two LVD (68 and 80).

<snip>

Interesting, the P2B-S uses the Adaptec AIC 7890 which is the same as the 2940U2W board.
[/B]</font>

That's the whole point: you've got a LVD version of the DDRS while I've got the 50 pin ultra version. So regardless of the used SCSI hostadapter only Ultra-SCSI can be used that is 20MB/s, with your U2W HD you can get max. 80MB/s with an U2W hostadapter (which you have).
Hence the higher burst-performance, just force the Adaptec into narrow Ultra-mode and then compare the results with the ones of the Initio(Domex): you should get the same values for burst as I got or even slightly lower ones.

[This message has been edited by Indiana (edited 27 April 2001).]

SuRGV
27th April 2001, 15:45
Here is mine:
http://pages.infinit.net/ttddhh/hdtest.jpg
It's Quantum Fireball AS 30 GB ATA100 @ATA33 on ASUS P3B-F. Win2000 Pro.
It seems mine is worst here.. Time to consider RAID now.
Trung.

xortam
27th April 2001, 15:51
<font face="Verdana, Arial, Helvetica" size="2">Originally posted by Indiana:
That's the whole point: you've got a LVD version of the DDRS while I've got the 50 pin ultra version</font>
Yes, of course I understand. http://forums.murc.ws/ubb/smile.gif I was just pointing out that the DDRS-39130 nomenclature was insufficient since they make five variants of that drive, with very different perfomance metrics.

SCompRacer
31st May 2001, 22:22
xortam, sorry I had not checked back here sooner and didn't see your question.

That was Win98SE with two 30GB IBM 75GXP's in RAID 0.

http://www.enteract.com/~richol/bench_content/atto_win98se.jpg

The above image was my best bench with Win98SE, two 30GB IBM 75GXP's in RAID 0.
http://www.enteract.com/~richol/bench_content/atto_win2000.jpg

The above image is my best bench with Win2000 SP1 and two 45GB IBM 75GXP's in RAID 0.


[This message has been edited by SCompRacer (edited 01 June 2001).]

xortam
31st May 2001, 22:49
Thanks SCompRacer. Again, let me ask what file system were you testing?

I've totally changed my system around since I posted, including a reinstall of Win98 and install of W2K-AS on removable boot disks. I kept my two drive array and used the third 9ES as a share drive for my OSes. The two drive array performed better for me than a three drive array and my Mylex RAID isn't supported under W2K so it made sense to leave the third 9ES separate for a shared drive. I also changed my stripe size from 16K to 64K. I did quite a bit of testing with Sandra, HD-Tach and ZD WinBench to determine the best setup. I found the synthetic benches to be practically useless and much prefer to use the real-world benching that ZD WinBench provides. WinBench allowed me to tune my system for how I actually use it versus some meaningless raw I/O metrics.

SCompRacer
1st June 2001, 12:20
Fat 32 for both to allow full access. 16K block size, formatted with the /z:32. Tried 32K and 64K with no improvement as I recall. Those benches were with HPT370 controller on the Abit board, and I never beat them with the Promise I have now though they are close.
I also tried a four drive IDE RAID 0 array and it was diminishing returns, maybe a 16% improvement over a two drive RAID 0.

I've never used ZD WinBench. Mostly I used ATTO as that is what was used at the Icrontic forum I frequented to help others out with their KT7A RAID boards. My goal was to show that ACPI was not the performance killer many there proclaimed it to be. Only once was I beaten there in a Win2000 ATTO bench on a RAID 0 array, but they had a four drive RAID 0 and didn't say until pressed for system specs.

Is ZD WinBench a freeware bench? I am really not much of a tweaker as I am a user, though I bench just to see where I am at compared to others.



------------------
ASUS A7M266, 1333MHz/266FSB, 512MB Crucial PC2100 DDR, Two RAID 0's, Win2000/Win98SE
The Rest (http://www.enteract.com/~richol/latest_computer.htm)

xortam
1st June 2001, 12:54
Thanks. I'll pour over your results later after I deal with my new burner.


<font face="Verdana, Arial, Helvetica" size="2">... Is ZD WinBench a freeware bench? ...</font>
Yes, though you need to buy the CDs (shipping/handling charge only) for the most complete coverage on some of their benches. I've always used the free stuff but I just bought the a) 3D WinBench, CD WinBench (only available on CD, obviously), and Audio WinBench CD and b) the Business Winstone and Content Creation Winstone CD. MURCers would be interested in the 3D WinBench as it will point out various shortcomings in the G4xx cards/drivers. The ZD benches are what PC Magazine uses to test and report on new HW. You can download various free benches at ZD eTesting Labs (http://www.zdnet.com/etestinglabs/filters/benchmarks/) (previously, ZDBOp [Benchmark Operations]).