hard drives & raid - benchmark and compare!

[tweakmaster420]

Just set up my two WD740-FLCO in Raid 0 Array on SATA 1&2 Ports

and also a second RAID 0 Array with my SATA II WD3200KS SE16 on Ports 3&4.

16k Stripe/4kb Cluster on both arrays.

I also have 2 seagates to play with (7200.9 SATAII 16mb cache w/NCQ 300GB&400GB) That bench faster then the WD3200KS's as single drives so I might remove the 320GB WD's and try those in an array. Or maybe the seagates and WD3200KS in a x4 RAID 0 array:nod: . Just don't want to mess up my x2 raptors boot stripe by removing them.

Or would I be able to Plug them right in if I wanted to Switch back without having to rebuild the array and reinstall winblows?

 

Heres the benches:

 

[soundx98]

Can't tell you how much I have enjoyed the last three posts!

 

Wevsspot brought the same questions that have been running through my head for a week.

 

ExRoadie brought some sanity to the party as usual. Clear. consise and nicely put Mike!

The only way to know which stripe/cluster works best is to test on your rig with your applications. Drive hardware, interface, number of drives and OS all have an effect on performance.

 

In general, for a two drive RAID-0 array, 16/4 works good for the Windows XP OS partition. 64/16 works good for large files like audio and video.

And that's exactly what I settled on - 16/4 OS and 64/16 storage.

 

Tweakmaster50 brought the empirical evidence and those oh so pretty pics.

Really nice job m8 and I know how time consuming that can be. Kudos

 

I've been doing the same thing on x64 Edition which is my Primary OS.

Here's a link to my benchs/results/findings/questions and sheet

http://www.planetamd64.com/index.php?showtopic=21897

[wevsspot]

The WD740's are so consistent in a RAID 0 array. I could lay my benchmarks almost right on top of Tweakmaster's and the results would be almost identical. I use the 16/4 arrangement on my personal rig for OS and applications. I use the 64/16 for storage and audio/video editing. I've always been happy with that.

 

The funny thing (sometimes it hurts to be a computer geek) is that in everyday normal use I can't tell the difference if my arrays are reading at 100mbs or 130mbs. But oh how we love those benchmarks

[wevsspot]

Oh yes. Shameless product plug here. Acronis True Image. Complete clone and restore of a 139gb array in less than 10 minutes. Free to try for 15 days (home only) then 29.95 if you want to keep it.

[suspekt]

Im just thinking here.. is it possible to get a faster array if you were to use 2 SATA drives and 2 IDE drives instead of all 4 of them being connected to the SATA ports?

 

I know coverters are bad to use when you have SCSI disks (they increase the resistance and cause more interference), but could it work better with the SATA/IDE interface?

 

I was thinking something like this:

http://www.granitedigital.com/catalog/pg52...ridgeboards.htm (random link from google)

 

Then I could put 2 of my SATA Raptors on the SATA ports and then use SATA-2-IDE converters on the other 2 Raptors. (Running them as Primary Master and Secondary Master on the IDE).

 

Thoughts, suggestions, input ?

[wevsspot]

I used two Rocket IDE to SATA converters for over a year on a pair of Maxtor Diamond Max9 IDE drives. The performance difference between that setup and native SATA drives was neglible. So i don't see any reason you couldn't go the otherway and expect similar results.

[tweakmaster420]

Ok, I rebuilt the WD3200's this time with 64stripe/16k cluster...

it dont look quite as good:

 

maybe i'll try 32/8 now

[red930]

Just a reminder!

 

The "recommended" stripe/cluster size is based upon a two drive array. The more drives you put in the array the larger the stripe to cluster ratio can be.

 

For instance, a four drive array may like a 128/16 stripe/cluster setup.

[tweakmaster420]

Or a 32/8... im really itchin to do a 4 drive array with the seagates..

These new seagate aint the 7200.7's I remeber, they really improved in speed and access time. the 300GB uses 3 100GB platters and the 400 uses 3 133 GB platters.

 

 

 

I think these will do better in a 2 drive RAID 0 array than the WD3200's.

The WD3200's avg is 10mb/s lower than the 300Gb seagate and the Peak only hits 63Mb/s in Single drive Bench.

I have to clear some data off them to DVD's so I can try them out in an array.:cool:

[tweakmaster420]

Tried this combo for the hell of it... boosted read perf.

64/512 allocation size...

 

[soundx98]

3 x WD2500KS SE16 (750G) Raid 0 Array Stripe 64, Cluster 16.

Zoom, Zoom, Zoom

 

 

my tests are continuing HERE

[suspekt]

I used two Rocket IDE to SATA converters for over a year on a pair of Maxtor Diamond Max9 IDE drives. The performance difference between that setup and native SATA drives was neglible. So i don't see any reason you couldn't go the otherway and expect similar results.

Was thinking that putting half of the drives on the IDE interface would releave some "stress" from the SATA ports, like having everything on it's own "channel". (Remember this is not a dedicated hardware controller we are talking about here, as it behaves very much different and better). But then again SATA ports doesn't steal bandwith from eachother, like for instance the IDE interface. Hmm

 

I know running a single SATA as IDE won't benefit anything, but with this NVRAID maybe you will have more throughput with 2 SATA's and 2 IDE's?

 

soundx98:

Nice read speed there on big sizes