Good morning all, I still have a couple inches of snow yet to shovel - all told we got more than 30 inches in some places of Virginia. Lots of sore muscles and blisters - yuk.
Newegg delivered the controller on time – day before yeseterday.
All iometer runs in this post are on a gigabyte X58 extreme mobo with an i7 965 proc running at stock.
First a few complaints – it looks like the controller f/w rounds down each disk to the nearest GB. For larger disk drives this is generaly not an issue but for acard ram disks, every MB is precious – it would be nice if this could be fixed. Then – looks like the latest W7 driver on the website might have some problems - at least with my set up anyway.
The iometer tests run in this post are on a Gigabyte X58 extreme mobo with an i7 965 proc running at stock.
So the first test - this is the LSI 9211-8i with 8xacardR0 using LSI 9211 driver 20270 dated 20Feb09 - I think this is the driver version that shipped with W7 beta bld 7100:
Results file located here - http://dump.no/files/f24c8eba30cd/2_02_70_20Feb09.xls
Then rerun the same test - this is the LSI 9211-8i with 8xacardR0 using LSI 9211 driver 20817 dated 17Jun09 – this is the latest version on LSI’s website:
Results file located here - http://dump.no/files/f24c8eba30cd/2_...n09_driver.xls
Notice the huge hit in seq reads? I will stick with the 20270 driver for the remainder of this post.
NOTE - THIS MAY HAVE BEEN A BAD POWER CABLE - DISCOVERED BAD CABLE ON WED, 30DEC
So lets see how seq reads vary based on transfer size:
Results file located here - http://dump.no/files/f24c8eba30cd/92...x_read__bw.xls
Somewhere around 128K - seq reads get pretty good – 1284MB/s. Firmware and/or driver need to be fixed to hold that kind of performance al the way up to 4MB+.
Next, I used acronis true image to move the same w7 partition to the LSI/acard array and run pcmark vantage
Tht pcmark vantage run below is LSI 9211-8i with 8xacardR0, on a gigabyte X58 extreme mobo with an i7 965 proc running at 4.35 (43x132) with mem at 1848 8-8-8:
The current top 20 -
This run would rank 13th all time – not too bad considering the state of the firmware and the driver. If LSI can fix their s/w, for ~$230. this might be a real winner!
Bookmarks