Which of these configurations would be better?

[AD-15]

hello,

Which of these system performance settings would be better in terms of performance, stability, and of course, heat:

- 400 FSB, x8 CPU multiplier, 1:1.2 FSB: DRAM (3.2GHZ CPU, 960MHz RAM)
- 360 FSB, x9 CPU multiplier, 1:1.25 FSB: DRAM (3.24GHz CPU, 900MHz RAM)

I know I've been making loads of threads recently, sorry about being a nuisance!

Thanks in advance!

[Hicks12]

360FSB will give the motherboard less strain, go with that one. Keep the FSB: RAM 1:1 so 720 or 800mhz is best, 900 might cause issues unless your ram is pc8500+
edit: ok forgot to look at spec but afaik 1 to 1 is best but having 4x1 ram you will need more volts so keeping lower could be better for stability.

[AD-15]

hello,

I've just checked, and the 400 option seems to give me slightly lower temperatures when using coretemp. Does 40MHz really make any difference? I was getting 56 max rather than 58C max in intel burn test.

[Hicks12]

Not sure but it would put less strain on the motherboard not the cpu, the multiplyer is higher so cpu is worked higher(i think thats how it works). How hot is the mobo? if its 2C difference then its not really an issue and id say test with orthos and see if its stable, flip a coin xD.

[GSte]

There will be very little difference between the two, nothing worth worrying about. No offence Hicks12, people do say it a lot, but 1:1 FSB:RAM does not matter either.

[Webby]

1:1 FSB : DRAM is great for pushing up FSB to see what your max OC is and particularly when your pushing the FSB high (500+) as other available dividers may put the RAM past its limit, but you should aim for the highest RAM speed possible so once you have a stable CPU speed play with the dividers to see if you can get the RAM going a bit faster.

[Hicks12]

Ah, well thats given me some information to lol. Would 800-> 900 give a noticable difference?

[Webby]

In benchmarks yes, in real life who knows will likely depend on how memory intensive the programs you work with are, of course it offers greater bragging rights DDR2-1150 FTW

[AD-15]

hello,

Thanks guys.

Do you think it would be worth swapping to the 360 FSB setting, and as the RAM speed is lower, tighten the timings to 4-4-4-12?

[GSte]

Not really, the performance difference will be negligable. I did find an example of RAM speed affecting something the other day, AVG virus scan ran about 10% faster when going from DDR800cas3/DDR1000 cas5 --> DDR1200 cas5, but generally the difference between different RAM speeds/timings is nothing noticeable.

[arthurleung]

Not really, the performance difference will be negligable. I did find an example of RAM speed affecting something the other day, AVG virus scan ran about 10% faster when going from DDR800cas3/DDR1000 cas5 --> DDR1200 cas5, but generally the difference between different RAM speeds/timings is nothing noticeable.

Funny enough I had ram that would run 1:1 stable but fail at 5:4 at same FSB (Probably just the auto timing messed up)

Back when I was using Pentium D going from DDR2-745 CL5 to DDR2-900 CL4 gave me a 15% boost in x264 encoding speed

I think the general rule is just do Ramspeed/CL and the bigger the value the faster it is. Correct me if I'm wrong.

[GSte]

It seems to depend on many things, the programs you are running, the motherboard chipset, the memory strap can have an effect too as you say.... I personally don't think it's worth worrying about one bit, CPU and possibly GFX, and the amount of RAM I'd agree with, but timings and speed have never had much impact on performance for me, apart from in synthetic benchmarks. Until that AVG Scan.