Raid 0 2x 500gb 7200.12 too slow

vslo

Distinguished
Oct 9, 2009
61
0
18,630
Hi,
I bought 2x 500gb 7200.12 seagate to perform a Raid0 128kb

My results where slower than most (with ICH10R intel matrix).
my moob is a biostar x58A.
I didnt installed drivers during the windows 7 64 installation, but after it was done I downloaded the last intel matrix driver for ICH10R controller.
However, my results are below normal.
My avarege read with write back cache disabled, is about 170-180mb/s and my burst rate dont get more than 320mb/s while a lot of people with the same controller and hd`s as mine get avarage reads about 240-260! and burst rate over 360mb/s.
With the write back cache enabled, I get about 200-215mb/s in avarage read but still cant go through 1800mb/s while the normal for those 7200.12 is about 3400-3800.
Those results that I`m stabling as NORMAL for those HD`s are based in hundred od threads in many many forums!

Please, suggestions?
 
First, the burst rate with write-back enabled is entirely dependent on your RAM and chipset - it has nothing to do with the drives. Anyone getting burst rates of 3000+ probably has faster RAM than you, or a newer chipset (or both). Second, there's no way that you'd ever get 240-260MB/s with two of those drives - 200-220 sounds about right. If people are claiming 240-260, they're either talking about max speeds rather than average, or they're full of it.
 

vslo

Distinguished
Oct 9, 2009
61
0
18,630

what you mean by separately or tandem ?
there are 3 spots in my x58 board for SATA. well, one is in SATA 1 and the other at SATA 2. the cant be both on SATA 1, is that what you mean ?
 

sub mesa

Distinguished
Your results are fine, and the write caching option will cause fake additional sequential performance figures in "bad" benchmarks like HDTune's raw benchmark.

If you want to know the real performance level of ICHxR RAID drivers without the Write Caching option enabled, do a filesystem benchmark like HDTune's Files benchmark, SiSoftware Sandra Filesystem benchmark or ATTO benchmark.