DDR2-800 run at DDR2-667 w/ better timings

[Radeon962]

Hypothetically, if I have some DDR2-800 with 5-5-5-15 timings and run it with an e2160 overclocked to 9x333 = 3GHz, does it make more sense to run the memory as DDR2-800 (2.4 mult. in Gigabyte BIOS giving 1:1.2 FSB😀RAM) or run it at DDR2-667 (2 mult. in BIOS giving 1:1) with better timings of 4-4-4-12?

I did find some mathematical explanations about cycles/operation of latency and latencies/operation, but in the end it sounds like even though you get better latencies/operation with better timings, the faster thoroughput of the higher memory (i.e. DDR2-800 vs. DDR2-667) will push through more data.

Bill

 

[nefariouscaine]

DDR2-800 at CAS 5 > DDR2-667 CAS 4 ---> but not by much as the two wouldn't likely be fully exploited for all their bandwidth

Some chipsets offer a minor penalty for not running 1:1 but its not going to be much and will likely be only noticed in running a large archive (rar file) or in a memory benchmark

get a larger amount of cheaper slightly slower ram - but i would stay away from DDR2-667 CAS 5 sticks

 

[DSF]

The real world performance difference will be meaningless 99% of the time. Don't worry too much about it.

 

[Mondoman]

The crux is that all the data to/from RAM has to go through the FSB, so as long as the memory bus provides a datarate at least as fast as the FSB, the bottleneck is the FSB.

 

[Radeon962]

I'm running DDR2-667 5-5-5-15 with my e2160 at 3GHz and it works just fine. As with anything else, I want to try something more. That's where the question came from.

Bill