Printable View
Napalm has done something no one else has done on this forum (expect low fat), he has demonstrated real life performance. You guys call out Napalm to run benchmarks and when he does you start to query the validity of them? :shakes::rolleyes:
Kudos to Napalm. :up:
+1
Testing real life performance is very hard, especially if critics demand a super high scientific methodology... there is none that people will agree upon. It has to be forced on us by some test site guru or forum "moderator" sort to say.
New LSI storage manager up on their site.
Grab it here
Personally I prefer the layout / functionality of the older version but the new one works fine too...
@Napalm - It's good to see some results we can compare between. Could I bother you though to run an ATTO and/or IOmeter as well?
LSI released new drivers, firmware and MSM yesterday.
YaYY!!Quote:
Originally Posted by Anvil
Download link :)
Ok, so I have the 2.30.03-0775 FW
I'm running on a X8DAH-+F Supermicro board with two Xeon 5560 24GB RAM (which will be 72GB when they send me the Registered RAM) and W7 x64 ultimate.
I'm running my SFF-8087 cable from my backplane on my Supermicro SC216 Chasis to my raid controller. My hard drives are Photofast v5 SSD 512GB hard drives. My raid setup:
RAID0
Read Policy: Adaptive Read Ahead
Write Policy: Always Write Back
IO Policy: Direct IO
Access Policy: RW
Disk Cache Policy: Enabled
Background Initialization: Enabled
on a 1M stripe.
My numbers are
READ:1504mb/s
WRITE:1736mb/s
How can I increase these numbers?
Sequential numbers, add more drives, faster drives, or possibly OC the PCIe bus.
The question is, what are your IOPS numbers, and do you NEED to increase storage performance? Or are you bottlenecked elsewhere?
Have you done a comprehensive search for bottlenecks in different situations?
You are aware even a 10GbE connection will be a bottleneck for (the sequential performance of) your array?
EDIT: you may get better results with a smaller stripe, f.ex. 64KB, 128KB, or 256KB.