RAID 0 benchmark

[serialb]

I have a RAID 0 setup with 2 Quantum AS drives and a Promise FastTrak 66. However, when I benchmark it with Sandra, the score is nowhere near Sandra's RAID 0 reference. I am positive I setup the RAID to stripping when I install the card. That makes me question the reliability of Sandra's HDD benchmark. Anyone know other programs/menthod that I can benchmark my RAID config? Thanks.

serialb

 

[bacillus]

you could try HDTach!

 

[Carceri]

Use a real life benchmark such as IOMeter. All those syntetic benchmarks such as Sandra, HDTach and Atto does not show what you actually experience in real life. High transfer rates are not the most important thing IMO, but instead the number of IO's processed per second.

 

[WHipLAsh13]

I use Sandra's hd benchmark and test out well above the benchmark for a raid 0. I have 2 IBM 60GXP's. The drivers you are using may not be correct or fully updated. When I first ran my test I forgot to load the latest drivers for my highpoint controller and got sub-benchmark scores but once I updated them my scores changed significantly.

 

[LostHiWay]

DON'T USE HDTACH. It will destroy a RAID if you use the advanced test. If you upgrade to the latest version of Sandra it should be pretty accurate.

Now if your using a single disk HDTach is the way to go but NOT for RAID 0

 

[woolmilk]

Make sure u took the U66/100 cables and not the old U33.

I ve recently set up a raid0 with FT66 and 2 maxtor 7200rpm 60GB. A single drive got me 16000, two on one channel with raid0 22000, two with one channel and cable for each got 24000. Then, I formatted the rest and made 4 partitions and still got 24000 on the first, but funny 29000 on the last one. Thats still less then sandra suggests for a comparable setup. Maybe U100 would give me another 10-20%.

(oh, forgot some 0 )

 

[sharkeeper]

<< Use a real life benchmark such as IOMeter. >>



IOMeter will show very poor results with most IDE drives because it is a server oriented benchmark. SCSI disks benchmark much higher with IOMeter.

Forget Sandra, it's worthless. ATTO is very accurate for measuring STR at various block sizes. It seems to have a problem with small block sizes with Windows XP however. HDTach is ok, but not very accurate with striped arrays. The most accurate metric of STR would be the Winbench99 disk inspection test.

Cheers!

 

[serialb]

I am using the latest driver and bios for FT66. I am also using ATA-100 cables on both channels. The array is set to stripping with 64kb stripe size. So, I don't think it has anything to do with my RAID configuration.

I am with sharkeeper on this one as I am never a big fan of Sandra. I'll give ATTO a shot and see how it turns out. Thanks.

serialb

 

[serialb]

Just tried out ATTO and ran a bench on it. Here is the result.

In the FastTrak BIOS, it says the drives are in UDMA-4 mode, which means the RAID card recognizes the drives as ATA-66 drives. However, from the ATTO bench results, the transfer rate seems to be capped at 28MB/sec. What is going on here?

serialb

 

[sharkeeper]

It appears that the drives are running in PIO mode.

You should connect each drive to the onboard IDE controller and benchmark it with ATTO. This could be a Fastrak problem too.

Cheers!

 

[serialb]

If the drives are in RAID 0, I don't think it's possible to boot into Windows Any other suggestion such that I can test the drive(s)?

 

[sharkeeper]

Normally, you can find another system that's working and swap one of your drives with the CD rom, for example. Boot into Windows and install a testing program and test it that way.

Cheers!

 

[serialb]

I plugged one of the drives from the RAID to another system but it seems to be write protected and unreadable. Still unable to bench it with ATTO.

 

[sharkeeper]

The drive can only be tested by ATTO if it's formatted and writable. Since this disk was part of a striped pair, it's basically unformatted to the operating system. HD Tach, however will still measure it.

Cheers!