Defiler

Terrible SCSI performance in Windows XP

Recommended Posts

Looks like I am in the same boat as most of you.

Not quite. The problem we're discussing here has only to do with write performance, but you also have slow reads! Note how both reads and writes are pegged firmly under 40 MB/s. Perhaps, your drive is not recognized as Ultra160 capable? Make sure to flash your Tekram with the latest firmware update (it fixes some recognition issues for X15 Cheetahs). Also check if your cable is Ultra160, and if you connect it to the correct channel on the card.

Leo

Share this post


Link to post
Share on other sites
The latest known SP1 build numbers are 1078 and 1079. I have neither.

XP SP1 1081 out. 2000 SP3 RTM leaked? .NET Server RC1 shipped to beta testers in Virginia, England, and Japan.

Share this post


Link to post
Share on other sites
Microsoft's official response:

"[...]

We will have a KB article on this very shortly. The fix associated with Q308219 is not related to this problem and we will update that KB article as well to provide better information."

Eugene,

When you have the KB number available, could you please post it here? Thank you.

Leo

Share this post


Link to post
Share on other sites

Hi,

This is my useless penny's worth.......I am using XPpro with a 19160, x15 36LP on an Asus A7M266-D. I had exactly the same problem, write speed of about 16mb/s (HDtach). I moved the card from the 4th pci slot to the 1st(64bit) and it sorted the problem...write speeds jumped to about 40mb/s.

Share this post


Link to post
Share on other sites

Hello,

I'm new to the forum. I've read pretty much the entire thread regarding SCSI and WinXP. I am already deeply invested in time, drivers, apps, setup, etc. in WinXP, so there's no going back to Win2K. I just got my first SCSI drive, which is an IBM DMVS 36V. I have an Adaptec 29160 U160 SCSI card. At bootup, the SCSI bios acknowledges 80MB/s active. I had trouble with both basic disc setting under XP, and with the native Microsoft driver for the Adaptec card. After changing to dynamic disc, and after installing Adaptec's beta release candidate 1 driver, I get mostly the same read speeds as write speeds. However, the absolute maximum read or write speed I get, accordint to ATTO, is about 28MB/sec. (Anything from about 16K transfer size to 1024K transfer size gets 28MB/sec).

It seems that I should be getting something more than that, but I don't know how much. My understanding is that I could burst to 80MB/sec, but that average throughput would be somewhat less.

I would greatly appreciate if any of you fine folks could give me an idea of what I should see from this setup. I do see folks in this thread with read speeds of 50MB/sec+, but not with the same hardware as I have.

At this point I can't tell if I have a hardware problem or if this is just a WinXP issue. Thanks in advance for any replies.

JGMusic

Share this post


Link to post
Share on other sites

Hello,

I am new to the forum and like most here am noticing slow performance with my SCSI drive while using XP.

When using ATTO benchmark utilities, I seem to be right on par with ZStation, my reads and writes being close and at the same speeds. Thirty - Five to 40 mb sec max.. I have never been able to achieve read speeds above 40 mb /sec in any OS . Should this drive be faster?

For Defiler

How are you getting those insane read speeds? Is the 36 gig X-15 that much faster than the 18 gig X 15 or is you using a next gen Cheetah? My model number is ST318451LW

Share this post


Link to post
Share on other sites
For Defiler

How are you getting those insane read speeds? Is the 36 gig X-15  that much faster than the 18 gig X 15 or is you using a next gen Cheetah? My model number is ST318451LW

I'm not Defiler, but take my word, the 2nd generation Cheetah is significantly (no, I can't really overemphasize it, SIGNIFICANTLY!!!!) faster than the 1st generation one. There is why it bothers me a lot when I see X15-36LP referred to as X15, 'cause the difference is enormous.

Note the rightmost digit before "LW" in the model number. It is the generation of the drive.

Leo

Share this post


Link to post
Share on other sites

There is commentary at 2cpu.com that the SP3 patch to win2k has introduced scsi performance hit similar to XP.

Can anyone here verify? This would be quite an interesting development.

I for one am not installing SP3 due to changes in EULA. I don't like spyware on my system.

Share this post


Link to post
Share on other sites

You need to change the Total Length option to 32MB. Those 90+ numbers you are getting are running from cache.

As for the low scores at the start, that's extremely low. Have you noticed a performance drop after installing SP3? Do you have another disk you can run a timed file copy to?

Share this post


Link to post
Share on other sites

Ok, I changed the length and reran the test and got more reasonable numbers. As for the low scores at the top, I dual boot the system with 2 win2k installs. luckily on the second, i hadnt got around to installing sp3 yet. i ran atto in win2k sp2, and the scores at the top were significantly higher.

the disk is highly fragmented so i'm in the process of defraggin now. when thats done i'm going to make sure i have the latest via4in1s, latest network card, and latest adaptec drivers (there was an update for adaptec drivers on windows update at some point i believe?, well if there was, ill make sure both installs have it.) for both installs. i dont have another disk in the system. but i do i have some ibm 75gxps in another system over gigabit ethernet.

anyway, i'll rerun atto, and time some file copys, and post my sp2vs.sp3 results here. so far they look to be consistent with the post in the 2cpu forums someone made:

"And to throw one more thing in the mix... it seems as if Win2k SP3 'breaks' the basic disk the same way XP basic disk does. My storage sub-system went in the toilet after applying SP3!", "My disk performance after SP3 with basic disk was so bad that ATTO could hardly register small disk writes and file copies took way longer then before SP3.... upgrading a disk to dynamic solved the problems... but I despise dynamic disk!"

btw, what version of atto should i have?

Share this post


Link to post
Share on other sites

well, i've done the tests. sadly, i'm uncertain of the validity of the transfer tests because its possible they were limited by the cpu, of the machine they were transfered to and from (550mhz athlon, 384mb pc133 ram, ibm 75gxp, winxppro). its proc was at 90% to 96% percent during transfer, but it never hits 100 leading me to think its possible the disks were maxed, and the cpu was almost but not actually a bottleneck. the athlonxp1900 system was in 10% to 15% proc usage range. I think high proc usage is due to the drivers for the nic in winxp by microsoft. if dlink releases xp drivers maybe it would help. anywho...

dir is a directory misc files, txt, jpg, gif, png. 122MB 695 files, 38 folders

img is a cd image file, .bin 446MB

SP3  READ  WRITE    SP2  READ   WRITE 

dir  24"   31"           23"    31"

img  39"   1'07"         39"    1'04"

virtually identical. with margin of error, identical. and the atto stuff,

attosp3.gif

attosp2.gif

Share this post


Link to post
Share on other sites
Hi,

This is my useless penny's worth.......I am using XPpro with a 19160, x15 36LP on an Asus A7M266-D. I had exactly the same problem, write speed of about 16mb/s (HDtach). I moved the card from the 4th pci slot to the 1st(64bit) and it sorted the problem...write speeds jumped to about 40mb/s.

40MB/s, the problem is not solved.

The X15 36LP must have rates approx. 58-63MB/s(depends on controller).

Dino

Share this post


Link to post
Share on other sites

Well IDE or not I feel that I am suffereing from the same problem.

ddriveatto2.jpg

My reads are great but the writes suck big time.

I did try converting to dynamic disk and I did notice a difference no pic for that yet however I will do it and post the result.

My set is 3ware Esclade 7450 (As close to SCSI raid as you can get in IDE) and 2 WD 100 JB drives.

Supermicro P4DCE+ dual Xeons 2.2ghz (intel i860)

XP Pro

So so far the solution is either wait for SP-1 (prolly won't fix it) convert to Win 2K, or convert to dynamic disks (only works for some) is that about how you see it?

"g"

Share this post


Link to post
Share on other sites

norxh, they're interesting results. Identical real-world results, but quite different for the ATTO results. Many people have questioned the validity of ATTO's results, this just adds more to that.

Kidfrost, dino has a good point. What test are you using that's giving you 40MB/s?

Share this post


Link to post
Share on other sites
Wow what a difference

Now I just have to figure out where to go from here?

"g"

Is that after converting to Dynamic Disks? I'm running a 4 drive RAID0 on an Escalade 6400 myself, and have performance numbers like your first set of benches. How did you convert to dynamic disk if that's what you did, and did it destroy your data?

BTW what kind of transfers do you get with timed copy or TotalCopy? I'm barely making 3MB/sec on a partition to partition move which is pathetic considering FreeBSD on another partition scores in the 60s when I'm using it.

Share this post


Link to post
Share on other sites
Is that after converting to Dynamic Disks? I'm running a 4 drive RAID0 on an Escalade 6400 myself, and have performance numbers like your first set of benches. How did you convert to dynamic disk if that's what you did, and did it destroy your data? 

BTW what kind of transfers do you get with timed copy or TotalCopy? I'm barely making 3MB/sec on a partition to partition move which is pathetic considering FreeBSD on another partition scores in the 60s when I'm using it.

Yes, I just did the conversion on a blank (newly created) array, that way I can do many different combinations.

I haven't used TotalCopy until now/this thread I have never heard of it.

I haven't thought that far ahead to consider partition to partition transfers.

"g"

Share this post


Link to post
Share on other sites

Why would SP 3 be slower?

I noticed a bit of a difference, going from 110mb/sec sp 2 in Adaptecs bm, sustained i/o, 256k, to around 90 with SP 3. Just uninstalled SP 3, and it went back up to about 108.

I've got atto, and they seem ok.

Question is, is anything in SP 3 so important that it's worth the performance hit?

s

8O

Share this post


Link to post
Share on other sites
Is that after converting to Dynamic Disks? I'm running a 4 drive RAID0 on an Escalade 6400 myself, and have performance numbers like your first set of benches. How did you convert to dynamic disk if that's what you did, and did it destroy your data? 

BTW what kind of transfers do you get with timed copy or TotalCopy? I'm barely making 3MB/sec on a partition to partition move which is pathetic considering FreeBSD on another partition scores in the 60s when I'm using it.

Yes, I just did the conversion on a blank (newly created) array, that way I can do many different combinations.

I haven't used TotalCopy until now/this thread I have never heard of it.

I haven't thought that far ahead to consider partition to partition transfers.

"g"

I hadn't heard of it till this thread either, so I searched for it and went out and got it. I knew I was getting slow copies in the 3-4MB/sec range just by timing how many seconds it takes to copy files. I wanted to know if dynamic disk speeds this up because if it doesn't then what's the point of worrying about it other than to see nice numbers in atto? The partition to partition transfer thing was so it didn't just update the FAT or whatever instead of actually moving the file.

Also it seems dynamic has a few drawbacks, like never allowing any other OS to access the disk. I might be able to live with that but my main concern is can I still have XP on a seperate partition from my apps so I can nuke it when it gets corrupted by oh say... realplayer or some other misbehaved app and I need to reinstall. Backing up 180GB of data every few months would be a pain.

Share this post


Link to post
Share on other sites

Yep, I'm trying to learn about the same thing as you.

I just got a reply to a post over at 2cpus on this very issue.

It indicated that you might be able to have the OS crap out on you to the point that you would have to do fresh install that you would be able just reinstall the OS and then import your dynamic volumes. I guess that is something that you could not do in NT 4.0.

I'm trying to follow up on that issue right now. If it is true then heck with it I'll just do the conversion and go with it. I'm not looking to run any other OS right now anyway, so as long as I can get back to my files saved on different vloumes on the same drive or on different disks or arrays then I'm good.

Ture it might be wrong for MS not to fix this but I like XP and it's not that big a deal to me so I'll just do it.

"g"

Share this post


Link to post
Share on other sites
Why would SP 3 be slower? 

I noticed a bit of a difference, going from 110mb/sec sp 2 in Adaptecs bm, sustained i/o, 256k, to around 90 with SP 3.  Just uninstalled SP 3, and it went back up to about 108.

I've got atto, and they seem ok.

Question is, is anything in SP 3 so important that it's worth the performance hit?

s

8O

As was said earlier in the thread atto tries to disable cache for the drive but a bug prior SP3 ignores that, so you get better benchmark numbers. With SP3 you get the speed without cache. Try to disable the write cache with SP2 you might get 90 as you got with SP3.

Maybe ATTO should allow cache just for simplicity... now the benchmarks cant even be compared without knowing OS and SP versions...

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now