Tom's Hardware > Forum > Storage > Hard Disks > WD800JB's in RAID 0 benchmarks

WD800JB's in RAID 0 benchmarks

Forum Storage : Hard Disks - WD800JB's in RAID 0 benchmarks

Tom's Hardware: Over 1.4 million members in 6 different countries available to answer all your high-tech questions. Sign up now! Its free!
Ask the community Add a reply
Word :    Username :           
 
johnnyx   10-01-2002 at 06:27:32 AM

Ok, I am finished with my benches. Here are the stats. The one on the left has a 32k stipe and cluster, the one on the right has a 32k stripe, and a 16k cluster.
AVG Access time 13.4ms 13.4 ms
Read Burst Speed 78540.8kb/s 78336 kb/s
Max Speed 61357kb/s 61843kb/s
Min Speed 36423kb/s 36478kb/s
Avr Speed 57512kb/s 57761kb/s
Test time 64:17:00 61:49:00
Partition size 131.78 GB 131.78 GB

I also tried 16k clusters with 32k and 16k stripe. The results were about half those of the other tests (like a 30-40 meg avgtransfer). Now compared to a single drive, like a wd1000bb the burst is about 10mb/s less. My question is this: would putting each drive on its own RAID channel, improve these specs at all? Once this is confirmed I will slap windows xp pro, and mandrake 8.2 on it and turn this puppy into a server.

Thanks


If an orange was driving a racecar would it peel out? www.jxfiles.com

Add a reply
Sponsored Links
Register or log in to remove.
lhgpoobaa   10-01-2002 at 07:36:47 AM
- 0 +

Yes. I would suggest that. Drives allways go better if they can hog the bandwidth of a channel all by themselves. :smile:
beside you have two IDE RAID channels, you may as well use them :)

<b>I'm Toms Hardware Guide Official Forum Strumpet! :cool: </b>

Reply to lhgpoobaa
HammerBot   10-01-2002 at 07:49:31 AM
- 0 +

Each drive should be set as master on its own cable (separate channel) for optimum performance (I told you :smile: ). But the difference is not large.

Im not quite sure on your configs. First you write some numbers for stripe/cluster; 32k/32k and 32k/16k. The numbers are virtually identical. How did you come up with those?

Next you state you also tried 32k/16k and 16k/16k, and got half the performance? Huh? You already tested 32k/16k above. Is this a typo? Im surprised you get half the performance. I wouldnt expect cluster/stripe in that range to have such a large impact. For completeness you should also try 8k stripe and cluster size. Note: Be careful how you benchmark. Some benchmarks may not show significant performance change, since they only tests with large files.

You compare your benchmark to a single WD drive transfer rate of about 50MB/s. Watch out here. Dont compare apples with bananas. The 50MB/s transfer rate of the drive is the absolute maximum sustained transfer rate of the drive. But your benchmark numbers are test of the drive formatted with a filesystem. You cant compare that. On a formatted drive you never even get close to those maximum transfer ratings. Try copying a 1GB file and time it with a stopwatch. If you are lucky your get about 20MB/s average read and write speed. Only compare numbers obtained using identical benchmarking tools.

Reply to HammerBot
HammerBot   10-01-2002 at 07:50:19 AM
- 0 +

Whats a Strumpet?

Reply to HammerBot
lhgpoobaa   10-01-2002 at 07:56:29 AM
- 0 +

HAHAHAHA
do some research :lol:
find out for yourself

<b>I'm Toms Hardware Guide Official Forum Strumpet! :cool: </b>

Reply to lhgpoobaa
HammerBot   10-01-2002 at 07:59:50 AM
- 0 +

Anything like this: <A HREF="http://www.winternet.com/~mikelr/flame40.html" target="_new">http://www.winternet.com/~mikelr/flame40.html</A>. I didnt know you were female :lol:

Reply to HammerBot
lhgpoobaa   10-01-2002 at 08:05:19 AM
- 0 +

haha. a good start my good man.
however for an accurate description please consult your dictionary.


<b>I'm Toms Hardware Guide Official Forum Strumpet! :cool: </b>

Reply to lhgpoobaa
Scotty35   10-01-2002 at 09:10:12 AM
- 0 +

Arrh, not always a female, can be a man in disguise. Read again. lol

<A HREF="http://www.btvillarin.com/phpBB/index.php" target="_new">A better place to be</A> :wink:

Reply to Scotty35
camieabz   10-01-2002 at 12:27:39 PM
- 0 +

Sounds like Wingding to me. :lol:

<b><font color=blue>~ <A HREF="http://www.btvillarin.com/phpBB/index.php" target="_new">Anyone looked in here yet?</A> ~<font color=blue></b> :wink:

Reply to camieabz
WingDing   10-01-2002 at 12:53:26 PM
- 0 +

Let's not drag me into a technical forum. I'll only lower the tone.

Interesting benchmarks. I'm planning to use a RAID 0 setup myself when I put my new rig together (not until Q1 03 :frown: ).

:eek: Wingding - proof of the need for genetic screening :eek:

Reply to WingDing
johnnyx   10-01-2002 at 05:49:29 PM
- 0 +

ok, yes, that is a typo, i meant to put 16k stripe and 32k cluster. The other thing is my soundblaster takes up so many irqs that i cant put each hdd on its own channel. I have tried that in the past it it just screwed things up. I think i will just run them on the same channel, unless for sure it would increase perfromance by 10% or more. probably not though. One other thinge, is it worth the time to do many more benchmarks?

If an orange was driving a racecar would it peel out? www.jxfiles.com

Reply to johnnyx
HammerBot   10-01-2002 at 07:04:01 PM
- 0 +

Quote :

ok, yes, that is a typo, i meant to put 16k stripe and 32k cluster


So what have you tested now? Stripe/cluster: 32k/16k, 32k/32k and then 16k/16k, 16k/32k which somehow halfed your performance? That cant be right. You shouldnt get such large impact by reducing stripesize from 32k to 16k. I suggest you repeat that test. And by the way. How do you test? Performance numbers are pretty much useless unless they are accompanied by some sort of test/benchmark setup story.

Quote :

The other thing is my soundblaster takes up so many irqs that i cant put each hdd on its own channel


Huh? Lets get one thing straight here: An IDE channel has nothing to do with IRQ. You should have no problem with putting each HD on its own IDE channel regardless of the IRQ of your soundcard. Further your RAID controller does not use more IRQs if you move one disk to the secondary channel.

Quote :

One other thinge, is it worth the time to do many more benchmarks?


Since you apperently got a very large difference with your previous mentioned setups, I (at least) think it would be interesting enough to repeat those. If they really are correct it appears that you get a very large performance increase by increasing stripesize. So why not test 64k stripesize as well. But remember larger stripesize have a good influence on transfer speed for large files but not for small files. So the optimum setup for YOU depends on where you need the most performance. Further Ive read in another forum that WD drives 'likes' 8k or 16k cluster size. So I would also test 8k clustersize.
Happy benchmarking :smile: ... and remember: NOW is the only time you have the opportunity to do those test. Once you put data on those drive you cannot change stripesize, and if you use NTFS you cant change cluster size either! (Well, at least I know of no tool that can do that, but if someone does please post it!)

Reply to HammerBot
HammerBot   10-01-2002 at 09:32:16 PM
- 0 +

Quote :

haha. a good start my good man.
however for an accurate description please consult your dictionary.


Then, what about this one: <A HREF="http://www.dictionary.com/search?q=strumpet" target="_new">http://www.dictionary.com/search?q=strumpet</A>. Doesn't sound right either. Perhaps you better explain, or do another one of those 'Is my sig too offensive' polls! :smile:

Reply to HammerBot
lhgpoobaa   10-02-2002 at 05:36:45 AM
- 0 +

It means i like answering questions. I am free with my gifts. I like people using my...errr... knowledge. :smile:

<b>I'm Toms Hardware Guide Official Forum Strumpet! :cool: </b>

Reply to lhgpoobaa
johnnyx   10-03-2002 at 02:56:05 AM
- 0 +

Ok, two things: So you are saying that, to benchmark, I should make a folder with varying file sizes (depending on what file sizes I use most)and then just copy it from one partition on my array to another, right? Because if i copy it from one physical drive to another, it would just be limited by my slowest drive. Also, what does it mean when I dedicate a channel for each hard drive and get static when i am formatting them? (there is no static when when they are both on the same channel). Thanks

If an orange was driving a racecar would it peel out? www.jxfiles.com

Reply to johnnyx
HammerBot   10-03-2002 at 07:33:51 AM
- 0 +

Quote :

Ok, two things: So you are saying that, to benchmark, I should make a folder with varying file sizes (depending on what file sizes I use most)and then just copy it from one partition on my array to another, right


No, Im not saying that (how did you get that impression) - Read again. Im simply asking how you test? since benchmark results are useless without a test setup. Further I warn you to be carefull with the tests, since stripe/clustersize combinations may have different impact on transfer speed for large and small files.
However, such a test does give you some idea of real life performance. But the performance is practically impossible to compare due to the many unknown parameters.

Quote :

Also, what does it mean when I dedicate a channel for each hard drive and get static when i am formatting them?


I dont know. What do you mean by getting 'static' when formatting?

Reply to HammerBot
Ask the community Add a reply
Tom's Hardware > Forum > Storage > Hard Disks > WD800JB's in RAID 0 benchmarks
Go to:

There are 666 identified and unidentified users. To see the list of identified users, Click here.

Forum map
Bestofmedia Forum ©
2000-2009 Bestofmedia Group
Page generated in 0.342 seconds
Please mind

You are about to answer a thread that has been inactive for more than 6 months.
If you still wish to proceed, please ensure that your posting is original and does not duplicate or overlap any prior responses to this thread.

Add a reply Cancel
Sponsored links
  • Ask the community now
  • Publish
Ad
Related Topics
See all relatives topics
They won a badge
Join us in greeting them