Benchmarking PrimoCache
-
- Level SS
- Posts: 477
- Joined: Wed Oct 06, 2010 11:10 pm
Re: Benchmarking PrimoCache
However I did notice CPU utilisation was quite high (80% on one core) during the Diskmark tests (except for the final QD32 testing which gave a lower load spread across 4-5 cores). Given we're using different CPU architectures, this might be the underlying cause so I have attached a copy of my Sandra CPU Multimedia benchmark results. I'm assuming these would be more relevant to PrimoCache than arithmetic benchmarks (as an aside Romex, are MMX/SSE instructions used at all by PrimoCache?).
- Attachments
-
- SandraCPUMM.png (70.48 KiB) Viewed 2339 times
Re: Benchmarking PrimoCache
I did some testing with FancyCache and the problem that I ran in to was actually due to the architecture of my MoBo.. MoBo manufacturers do a lot of shady money saving things, like sticking the storage controller on a single lane PCI bus (so they can save the lanes for video cards) or they daisy chain the buses, so you can have a 8 lanes running off of a 2 lane bus. The closest I could do to solving the problem was to stick to 3rd party SAS controllers but sadly one of them saturates the bandwidth so I can never get anywhere near the 6GBs I should see. Next time I'll use a server board instead, then you know you're getting what you paid for.
viewtopic.php?t=1278
viewtopic.php?t=1278