New LSI 9200 series controllers: 6Gb/s, 2.88 GB/s seq. reads, 1.87 GB/s seq. writes

[WonderSausage]

It doesn't look impressive at all.

Tom's has no idea how to configure hardware RAID, and they tested the LSI on the ATI board, not the Core i7.

[WonderSausage]

I'm waiting for the Intel Gen2 SSDs to become available again, but in the meantime, my LSI 9260-8i arrived and I decided to do some initial testing to make sure that it's getting allocated enough lanes and so forth.

I just threw on a couple of ancient Raptor 36GB in RAID 0 and set up benches with small working sets to test cached performance only (with write-back forced on because my battery hasn't arrived yet).

In my previous experience with X58 and other RAID controllers, they tend to get 4 lanes allocated. Of course, all previous controllers have been PCI-e 1.0 and the 9260 is 2.0. There is no way to tell the difference between 4 lanes @ 2.0 and 8 lanes @ 1.0 but due to the previous experience I'm going to assume 4 lanes.

Clearly the LSI 9260-8i is getting 4 lanes @ 2.0 on the P6T Deluxe given these benches. Hell, it's above the theoretical real-world maximum (theoretical ultimate maximum is 2000, real-world is suppoed to be ~1800). It's in the bottom X16 slot on the P6T Deluxe. I put an Intel PCI-e x1 NIC in the top x1 slot as well as a 4870X2 in the first x16 slot.

Let me just say in caps for those skimming this thread, THIS IS INTENTIONALLY TESTING CACHED PERFORMANCE.



It's obvious that the card has enough throughput for any drive configuration, and that the cache is fast enough to not hold back any SSD RAID0 setup.

I'm not going to bother to post the ATTO results which as you probably know are all over the place once you get over a few hundred MBps, but the max reached was similar in that bench.

WS