Overclock.net banner

1 - 14 of 14 Posts

·
Registered
Joined
·
1,830 Posts
Discussion Starter #1
My current OS is only on one drive right now of my two 250GB due to XP requiring a floppy drive to install RAID and I couldn't find a working floppy disk drive and and disk. Now however since I want to upgrade to Windows 7 64 bit, I want to get my RAID 0 working again. With my EVGA 750i RAID I was able to reach an average of 90MB/s read speed which I guess is pretty decent for older drives the normally should only do like 50-60. However I want in the 100's at least and I read something about short stripping or something? I have 1 TB file server right next to me so I'm not worried about space so much. I just want my RAID 0 to be faster hah.
 

·
Registered
Joined
·
1,664 Posts
There might be issues in using different brands and types of drives. It might slow you down a little bit. Not sure there... Stripe size is the only other thing I can think of that might affect speed. Oh yeah, don't forget to enable advanced performance on the stripe in windows ... my computer -> c: drive (whatever drive is the stripe -> hardware tab -> you should see "nvidia stripe" in the list of drives, click properties of that drive -> check enable write caching and advanced performance boxes ... I'm not sure how much increase but you should definitely see some increase afterward.
 

·
Registered
Joined
·
1,730 Posts
Quote:


Originally Posted by Sora1421
View Post

My current OS is only on one drive right now of my two 250GB due to XP requiring a floppy drive to install RAID and I couldn't find a working floppy disk drive and and disk. Now however since I want to upgrade to Windows 7 64 bit, I want to get my RAID 0 working again. With my EVGA 750i RAID I was able to reach an average of 90MB/s read speed which I guess is pretty decent for older drives the normally should only do like 50-60. However I want in the 100's at least and I read something about short stripping or something? I have 1 TB file server right next to me so I'm not worried about space so much. I just want my RAID 0 to be faster hah.

Sure, Short Stroking is easy. Just make a new raid0 array and assign like 25% percent of the usable size and it will be short stroked. Leave the rest of the available space free. Also enable write back caching when you have installed the new windows.
 

·
Registered
Joined
·
1,664 Posts
Quote:


Originally Posted by Izvire
View Post

Sure, Short Stroking is easy. Just make a new raid0 array and assign like 25% percent of the usable size and it will be short stroked. Leave the rest of the available space free. Also enable write back caching when you have installed the new windows.

Is your array short stroked now? I wonder what kinds of numbers you get in hdtune? I have the same drives, different raid controller but should produce similar numbers. Thanks for the idea of short stroking, I will have to try it out.
 

·
Registered
Joined
·
1,730 Posts
Quote:


Originally Posted by dmreeves
View Post

Is your array short stroked now? I wonder what kinds of numbers you get in hdtune? I have the same drives, different raid controller but should produce similar numbers. Thanks for the idea of short stroking, I will have to try it out.

 

·
Registered
Joined
·
1,664 Posts
Very nice... that's a slightly older version of hdtune... that must be read speed then? I am getting MUCH lower numbers... my average is about 150mb / sec ... will post picture soon.

 

·
Premium Member
Joined
·
5,330 Posts
Short stroking does not increase your sequential read speeds at all.

It will reduce average access times though, which will make your system more responsive.
 

·
Premium Member
Joined
·
65,162 Posts
Quote:


Originally Posted by the_beast
View Post

Short stroking does not increase your sequential read speeds at all.

It will reduce average access times though, which will make your system more responsive.

It does increase average sequential read speeds. This is a result of limiting benchmark data to the the faster sections of the array. It does not improve the actual performance. However, you should be able to achieve the same sequential speeds with a well-defragged array.
 
  • Rep+
Reactions: c00lkatz

·
Guess I'm Back
Joined
·
3,667 Posts
I don't mean to threadjack, but does short stroking show any real world difference other than benchmarks? I'm at about 12ms access time right now with my 640 AAKS RAID 0 non-short-stroked, and they are the only two drives I have. I have 3 partitions right now, one for Vista (main partition), one for Windows 7 (which I just recently installed) and one for pure storage. Would I see a real gain that I can feel if I got a new 1TB WD Black for pure storage and short-stroked my 640 AAKS RAID 0 array? Is it really worth the extra $100 for the new drive?

Again, I don't mean to threadjack, but I figure this is something the OP might want to hear as well.
 

·
Premium Member
Joined
·
65,162 Posts
Quote:


Originally Posted by c00lkatz
View Post

I don't mean to threadjack, but does short stroking show any real world difference other than benchmarks? I'm at about 12ms access time right now with my 640 AAKS RAID 0 non-short-stroked, and they are the only two drives I have. I have 3 partitions right now, one for Vista (main partition), one for Windows 7 (which I just recently installed) and one for pure storage. Would I see a real gain that I can feel if I got a new 1TB WD Black for pure storage and short-stroked my 640 AAKS RAID 0 array? Is it really worth the extra $100 for the new drive?

Again, I don't mean to threadjack, but I figure this is something the OP might want to hear as well.

You already are short-stroking with your Vista partition.


While it isn't the only partition, it still is on the outer edge of the platters.
 

·
Premium Member
Joined
·
5,330 Posts
Quote:


Originally Posted by DuckieHo
View Post

It does increase average sequential read speeds. This is a result of limiting benchmark data to the the faster sections of the array. It does not improve the actual performance. However, you should be able to achieve the same sequential speeds with a well-defragged array.

Exactly - this really only makes your benchmark look better, but does nothing to your actual transfers.

Quote:


Originally Posted by c00lkatz
View Post

I don't mean to threadjack, but does short stroking show any real world difference other than benchmarks? I'm at about 12ms access time right now with my 640 AAKS RAID 0 non-short-stroked, and they are the only two drives I have. I have 3 partitions right now, one for Vista (main partition), one for Windows 7 (which I just recently installed) and one for pure storage. Would I see a real gain that I can feel if I got a new 1TB WD Black for pure storage and short-stroked my 640 AAKS RAID 0 array? Is it really worth the extra $100 for the new drive?

Again, I don't mean to threadjack, but I figure this is something the OP might want to hear as well.

Short stroking can make a huge difference to performance if you are struggling with lots of random access slowing down your disks. If you aren't, then it won't do anything that a good defragger won't do.

If you were to buy a new 1-1.5TB disk for storage, then I would say move everything off your main array & short stroke, leaving just your OS & apps on the main disk. But honestly, unless you notice slowdowns during heavy disk access & can hear the drive heads flying around I wouldn't recommend buying a disk just so you can do it.

EDIT: only after reading Duckie's post did I re-read yours and realise you already have 2 partitions.

Basically, if you use stuff from both partitions at the same time, you will take a performance hit, as the heads have to seek back & forth continuously. So if you do things like photo or video editing you would probably be better off with a new disk. If not then the benefits are less clear cut - you will get a performance hit even if you just play an mp3 from your storage disk, but whether it is noticable is another matter. I guess it depends how much money $100 is to you - some people can drop that without thinking about it (but then if you have that kind of cash go SSD!).

Hope this helps.
 
  • Rep+
Reactions: c00lkatz

·
Guess I'm Back
Joined
·
3,667 Posts
Thanks for the replies. I was under the assumption that a true short-stroke was setup while setting up your actual array and reformatting. I basically have a full 1.16TB array, but I did partition, but everything is still on the RAID 0 array. In HDTune and other benchmarks, I am not getting results even close to others who have setup their array for only a small portion of disk space, losing the rest of the drive space.

I just didn't want to throw down the cash, reformat, reinstall everything, lose a full TB of drive space and not notice a nice difference.

My PC isn't slow by any means, but I know every little bit helps. I mostly just listen to music, surf the net, watch movies, and game. I don't do any encoding or anything like that.
 

·
Registered
Joined
·
643 Posts
Quote:


Originally Posted by c00lkatz
View Post

Thanks for the replies. I was under the assumption that a true short-stroke was setup while setting up your actual array and reformatting. I basically have a full 1.16TB array, but I did partition, but everything is still on the RAID 0 array. In HDTune and other benchmarks, I am not getting results even close to others who have setup their array for only a small portion of disk space, losing the rest of the drive space.

I just didn't want to throw down the cash, reformat, reinstall everything, lose a full TB of drive space and not notice a nice difference.

My PC isn't slow by any means, but I know every little bit helps. I mostly just listen to music, surf the net, watch movies, and game. I don't do any encoding or anything like that.

The way I understand it. It works kind of like this.

The first 1/4 of your benchmark = 100% of a short-stroked benchmark.

So basically by short-stroking, you are limiting HD Tune to only bench a portion of the outer edge of the platters.

Your HDTune benches start from the outer edge, but then go past the cutoff point that someone would have short stroked too. When your bench gets past that point where someone would have short-stroked to, that's the other 3/4 of your benchmark where your speeds start to get slower.

So basically if you partition C:\\ first on your raid array (on the outer edges) and you have a well defragged drive - and your drive is only reading/writing from/to C:\\ then you will see the performance of a short stroked drive.

As soon as your drive starts reading past the first 10-25% of your drive (where most people stop on a short stroke) then that's when your graph starts to slope downwards.

Benefits - (of short stroking)
Faster access times - by limiting the area that the heads have to travel back and forth from.
Faster "On Average" Seq Read/Writes only by way of denying yourself access to slower sections of the disk

Cons -
Can only use 10-25% of your total storage space.
 

·
Guess I'm Back
Joined
·
3,667 Posts
Quote:

Originally Posted by Evermind View Post
The way I understand it. It works kind of like this.

The first 1/3 of your benchmark = 100% of a short-stroked benchmark.

So basically by short-stroking, you are limiting HD Tune to only bench a portion of the outer edge of the platters.

Your HDTune benches start from the outer edge, but then go past the cutoff point that someone would have short stroked too. When your bench gets past that point where someone would have short-stroked to, that's the other 2/3 of your benchmark where your speeds start to get slower.

So basically if you partition C:\\ first on your raid array (on the outer edges) and you have a well defragged drive - and your drive is only reading/writing from/to C:\\ then you will see the performance of a short stroked drive.

As soon as your drive starts reading past the first 10-25% of your drive (where most people stop on a short stroke) then that's when your graph starts to slope downwards.

Benefits - (of short stroking)
Faster access times - by limiting the area that the heads have to travel back and forth from.
Faster "On Average" Seq Read/Writes only by way of denying yourself access to slower sections of the disk

Cons -
Can only use 10-25% of your total storage space.
Looking at the graphs, that makes sense. The performance is averaged after all. I feel a lot better about my array now lol
 
1 - 14 of 14 Posts
Top