RAID 0 Speed Problem - 550 MBps Write Speed too high

[Moharram]

hi

i have 6 SATA Hard Disk with RAID Controller that create a drive with 850 GB Volume...

when i test the speed with PerformanceTest i gain the speed between 450 to 550 MBps Write Speed ...

but in my program i can't acheive for this rate...

i use CFile in MFC Application and try to write to hard and calculate speed.
in the best case my speed is 115 MBps..

is there any programming techniques ... ???

thanks

[David (PassMark)]

You have to be careful to make sure you are no just measuring the cache speed.

If you are getting 550 MBytes/Sec (as opposed to 550Mbit/sec) then I think you are measuing the disk cache speed and not the real disk speed.

[Moharram]

Thr Result of PerformanceTest is Below:
http://barnamenevis.org/forum/attachment.php?attachmentid=12333&d=1195399151


as shown in the graph : SPEED = 478 MBps

[David (PassMark)]

The link you have posted is to a private forum that no one has access to. So no one can see your image.

[Moharram]

Sorry
please check these urls:

http://link.imgshare.us/ajjJy0

http://link.imgshare.us/ajjJdu

[David (PassMark)]

What RAID level are you running?
How much RAM is in the machine?
What O/S are you using?
What type of drives are they?
What CPU do you have?

[Moharram]

RAID 0 (Stripping)
3 GB of RAM
Windows XP
on drive (850 GB) (6 * SATA)
AMD 64X2-DualCore 6000+

[David (PassMark)]

I agree the numbers look a bit strange as they exceed the SATA interface spoed (300MB/sec for SATA 3GBit). We don't have a 6 drive RAID config to test with, and we couldn't reproduce numbers like this with a single drive.

Do you get similar numbers with a smaller block size?
Do you get similar numbers with a larger test file (at least double your RAM size)?

[merlin]

RAID 0 Performance

Are you using a hardware SATA RAID controller something you brought and put into a PCI-e slot and not the one that came with the motherboard?

I agree with the Passmark admin It sounds like its the cached speeds that Passmark is reading.