timings or fsb

these timings are pretty much the best i can do with this ram, i think i could get 2.5-2-2-5.

i can get the ram to 460mhz with crappy timings

should i loosen up my timings and go for highest fsb, or keep the fsb at 400 and have really good timings? i would bench and find out for myself but my rig is down while AMD processes my rma...

Test and find out.. and let us know

I cast a vote for high FSB, but that's just a hunch, and not based on fact at all!

hope your RMA gets processed quickly. that way we'll know for sure!

I have a thread bookmarked on my home computer where a guy asked the same thing. I said high FSB would be better and a couple people thought the tighter timings would be better. He eventually ran the benchmarks and found that higher FSB was indeed faster. I'll try to remember to post that link when I get home.

I guessed for higher fsb, as that should increase the bandwidth for between the CPU, NB and RAM, and there should be more bandwidth from dual channel ddr than the EVA6 bus...

but I kept my mouth shut as I never tested this out.

but seeing confirmation from batboy, would what I thought of be the reason for higher fsb to be better?

Alright thanks guys, I'll bench it when my 2500+ comes back and I'll post a definitive thread

If it's a difference of single digit MHz, like from 200 to 208, then timings would take it. Now, you're talking 200 to 230, that is a BIG jump. I'm almost sure you'd do better at 230 as long as you're not running 3-4-4-8. I'd go as low as 2.5-3-3-7 and see what you can get from there.

--Illah

i read an article a while ago, that stated, (with lots of benchie #'s) that latency timings have much less effect on ddr modules than they did with sdr modules. In conclusion, it was found that higher fsb's made much more difference than tighter timings in DDR. I wish i could find that article again, i would like to read through it again. But i was testing all day yesterday, and found that the diff between 6,2,2,2 and 8,4,3,2.5 @ the same fsb was about 50MBs according to sandra, which was easily made up for and some more by upping my fsb to 225 .


J.

ok i had this question too so i decided to see which would be better. i ran the sandra mem bandwidth test a while back but i cant remember the exact score but i know the round figures:

i ran my pc 2700 ram at 133fsb(266ddr) at the tightest timings of 2-2-2-5(sandra said 6) and got a score of 1930/1881
then i ran the same ram at 141fsb(282ddr) at the loosest timinigs of 2.5-3-3-6(sandra said 7) and got a score of 2062/2000

now my test involved a very slight fsb increase of 8 (like Illah said, its amazing because this happened by chance) and fsb scored higher so i figure it would score alot higher if u had 30mhz of fsb increase

if its not obvious i vote for fsb

Y.E

I think people take the memory timings a little bit TOO seriously. Seriously, change your CAS Latency to 2.5 and see if you notice ANY difference at all. Now bump the FSB up another 30mhz or so. You'll notice a difference.