[Neuromancer]

I would rather have 1333 6-6-6 over 1800 8-8-8 anyday. But I am an AMD RAM guy mostly

I never noticed much of a differnce on any of them for speed. Once you have enough bandwidth more does not help, only latencies. it is usually easier to get tighter than it is faster IME to lower them.

[Hondacity]

Originally Posted by Neuromancer

I would rather have 1333 6-6-6 over 1800 8-8-8 anyday. But I am an AMD RAM guy mostly

I never noticed much of a differnce on any of them for speed. Once you have enough bandwidth more does not help, only latencies. it is usually easier to get tighter than it is faster IME to lower them.

you should try sp32m for comparison

[Cecil]

Originally Posted by Neuromancer

I would rather have 1333 6-6-6 over 1800 8-8-8 anyday. But I am an AMD RAM guy mostly

I never noticed much of a differnce on any of them for speed. Once you have enough bandwidth more does not help, only latencies. it is usually easier to get tighter than it is faster IME to lower them.

Even with AMD higher speed helps if you NB speed is high enough. I think the NB bottlenecks your mem bandwith untill its double the speed of the ram or something like that. Its been a while, but that sounds right.
I did a comparison a while back, and there was much of a difference with the timings or speed untill I upped the NB to 2600 to ease up on the bottleneck. Fastest Spi32 I got was 1600Mhz 6-6-6-18 with NB at 2600.

[Neuromancer]

Originally Posted by Cecil

Even with AMD higher speed helps if you NB speed is high enough. I think the NB bottlenecks your mem bandwith untill its double the speed of the ram or something like that. Its been a while, but that sounds right.
I did a comparison a while back, and there was much of a difference with the timings or speed untill I upped the NB to 2600 to ease up on the bottleneck. Fastest Spi32 I got was 1600Mhz 6-6-6-18 with NB at 2600.

Yes its triple effective speed.

IE DDR3 1600 = 800MHz = 2400MHz CPUNB speed.

And yes SPI32 is a memory everything intensive app.

I was talking for use on a non-bench rig. Tighter latency over higher speed is my recommendation.