Overclock.net banner

1 - 19 of 19 Posts

·
Premium Member
Joined
·
7,703 Posts
Discussion Starter #1
Here are my results with two Raptors in RAID 0. They aren't truly short-stroked, but I did cut the volume down to 200GB from 300GB, because I don't really need 300GB of space. Does this seem about right for these drives? The access time looks good, but the burst rate seems pretty low from what I've seen around here.

 

·
Premium Member
Joined
·
17,110 Posts
I'm pretty sure my raptors in raid 0 were much higher than that... I hope you didn't just buy those, or you got them for really cheap. My SSDs get 550MB/s reads...
 

·
Premium Member
Joined
·
7,703 Posts
Discussion Starter #3
No, I've had one for a couple of years. I just got another one on eBay for $60 bucks.
 

·
Beyond
Joined
·
4,493 Posts
Not what I'd expect for RAID 0 Raptors. Like you said the access time is really good but my RAID 0 WD 640GB AAKS drives get higher numbers with everything else. That can't be right at all. Did you test multiple times? Is it a pretty fresh OS install and have you done a defrag recently?
LL
 

·
Premium Member
Joined
·
7,703 Posts
Discussion Starter #6
It's a fresh install. I used 128K for the stripe. Should I have used a lower number?
 

·
Premium Member
Joined
·
7,703 Posts
Discussion Starter #8
Regular. I know the result is low for these drives, I just want to figure out some possible reasons why.
 

·
Premium Member
Joined
·
5,330 Posts
Turn on write-back caching.
 

·
Registered
Joined
·
865 Posts
Quote:

Originally Posted by the_beast View Post
Turn on write-back caching.
Write-Back Cache does improve burst rates and write speeds, however it uses your system memory instead of the hard drive cache to buffer "writes" to the hard drive.

I would suggest against this unless you have an uninterruptible power supply, because system memory is volatile and you will lost whatever you were transferring at the time if you lost power.

I'm not really sure what is up with those results. The highest reads/writes should be at the beginning and the lowest reads/writes at the end of the benchmark.
 

·
Premium Member
Joined
·
7,703 Posts
Discussion Starter #12
With or without write cache gave similar results. Just slightly lower without it. I might short stroke it down to about 100GB to see if it helps a little. I really don't want a smaller drive than that though.
 

·
Premium Member
Joined
·
5,330 Posts
do you have Matrix Storage Manager and the latest drivers installed?
 
  • Rep+
Reactions: losttsol

·
Premium Member
Joined
·
7,703 Posts
Discussion Starter #14
Quote:


Originally Posted by the_beast
View Post

do you have Matrix Storage Manager and the latest drivers installed?

Ya, I press ctrl-i and that is what comes up. I will check the drivers though. Thanks.
 

·
Premium Member
Joined
·
7,703 Posts
Discussion Starter #15
I changed the block size in the the settings to 128K which is what I used setting up my RAID. I read that this is how you are supposed to run the test. The results look a lot better.

 

·
Premium Member
Joined
·
5,330 Posts
Honestly they don't look any better - the graph should start high then go down.

Set the block size larger than that or you won't really be able to bench the max speeds of your array. Try 512.

Also MSM is a Windows app for managing your array. Without it you can't set/check some of the more advanced settings. Get it from Intel.
 
  • Rep+
Reactions: losttsol

·
Premium Member
Joined
·
7,703 Posts
Discussion Starter #18
Quote:


Originally Posted by the_beast
View Post

Honestly they don't look any better - the graph should start high then go down.

Set the block size larger than that or you won't really be able to bench the max speeds of your array. Try 512.

Also MSM is a Windows app for managing your array. Without it you can't set/check some of the more advanced settings. Get it from Intel.

Thanks. I found where you can enable write-back cache in MSM. I was just enabling write caching in disk management. Here are my new results. The burst rate went way up.

 

·
Premium Member
Joined
·
5,330 Posts
forget burst rate - for most things it is meaningless, and only has some use in diagnosing a few issues.

Now your results look fine for the disks you have.

As mentioned in this thread, running in write-back can risk data loss. But TBH there is little difference to running with your onboard drive cache enabled (as is the case by default) - a sudden power loss could cause issues even if you run in write-through. As you are well advised to have a backup of your system drive (whether running in RAID or not) you should be able to minimise the potential losses anyway. For 99% of desktop uses you get much better performance in write-back (for both writes AND reads), so it is usually best to deal with the slightly increased risk of corruption rather than the lacklustre performance.
 
1 - 19 of 19 Posts
Top