Frequency Vs Timings Benchmarking

[merlinbadman]

My Core i7 setup defaulted my RAM to 1066Mhz and gave me timings of 6-6-6-20 1T. When I manually set the RAM to 1333Mhz the timings increase to 9-9-9-24 1T.

What benchmarking suites are good and would allow me to guage what is better?

 

[outlw6669]

If you are unable to manually set the timings where they should be, running your RAM at 1066Mhz with 6-6-6-20 timings will give you the best performance.

http://www.thetechrepository.com/showthread.php?t=160

 

[outlw6669]

Sorry, CL9 is a little off the end of that chart.
This one will give you data through CL10

 

[merlinbadman]

Cheers for that info, very helpfull.

So @ 1066 CAS 6 my RAM is doing it all in roughly 11ns ish

@ 1333 CAS 9 its doing its thing in 13.49ns

Is that correct?

 

[merlinbadman]

Cant edit my post, Grrrr

Forgot to post that my RAM states that it should be capable of 7-7-7-20 1T @ 1333 which equates to 10.49ns.

So..... am I correct in thinking the best way forward for me is to set it to 1333 and manually change the timings in the BIOS to 7-7-7-20 1T ?

 

[Zenthar]

While on the topic, maybe someone could tell me the quantifiable impact of running RAM asynchronously (not 1:1 ratio)?

When I built my system last march, I saw no performance difference in benchmarks (3DMark 06 & Crysis) between running my RAM at 667 or 800 MHz when my CPU runs at default 1333FSB.