Sandra 2009 vs Everest Ultimate 4.60 reliability

[Neb]

I was wondering about your opinion on which one of these programs are most reliable regarding RAM perfomance. This question becouse I see large differences between results.

I was testing running my RAM at 667MHz 1:1 ratio vs system bus 1333MHz.


Sandra reported read 5,34GB/sec, write 5,32/sec and 112ns.

Everest reported read 6.3GB/sec, write 7GB/sec and 80ns.


Seems to me like Sandra measures single-channel perfomance and Everest dual-channel perfomance?

Any thoughts?



EDIT: Also is the system bus limit the absolute limit for RAM bandwidth. Say if my system bus bandwidth is 10.41GB/sec then it doesn't mather if my RAM has 16GB/sec or better since it wont be utilised?

Just thinking if the GPU would have direct access for trashing and could utilise the extra bandwidth when I exceed VRAM usage.

 

[chavvdarrr]

EDIT: Also is the system bus limit the absolute limit for RAM bandwidth. Say if my system bus bandwidth is 10.41GB/sec then it doesn't mather if my RAM has 16GB/sec or better since it wont be utilised?

yes.

I was wondering about your opinion on which one of these programs are most reliable regarding RAM perfomance. This question becouse I see large differences between results.

I was testing running my RAM at 667MHz 1:1 ratio vs system bus 1333MHz.


Sandra reported read 5,34GB/sec, write 5,32/sec and 112ns.

Everest reported read 6.3GB/sec, write 7GB/sec and 80ns.


Seems to me like Sandra measures single-channel perfomance and Everest dual-channel perfomance?

Any thoughts?

Who can say what these programs measure?!

 

[Moloch]

I generally trust sandra the least of any of the popular benchmarks, but that's just me, mostly because of how unreliable some of the tests are, like the HDD one.
really any of the benchmarks are only good for comparing performanc of overclocks imo.

 

[Neb]

Well they measure something!

Atleast they give some kind of guideline for the system. About Sandra I agree basing it on past experience where a 300MHz Celeron without cache got same perfomance index as a PII 300MHz. Kinda far off from reality.

But then to summarize, faster RAM throughoutput than system bus is wasted. Meaning my RAM at 667MHz or 1000MHz should perform equally with same timings?

Also anyone of you know a good guide to RAM and chipset memory timings, especially the chipset memory timings and how they relate to each other?

 

[swaaye]

They try to come up with the right combo of assembly instructions that will squeeze as much speed out of each platform. That's obviously very synthetic. I think it's most useful for seeing how much various BIOS tweaks get you.

 

[Neb]

Guess my 500MHz (1000MHz effective) is only good for 2000MHz FSB. Atleast i can tighten the timings more than it's specified 5-5-5-12 for 500MHz.