• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Some Timings Vs. FSB Fun A64 Style

T

tsuehpsyde

Supreme [H]ardness
2FA
Joined
Oct 22, 2004
Messages
6,604
Idunno, being frustrated at my system being tapped out, I decided to look into a way to get the most out of my system at stock (or close to) settings....day to day, I don't run full core, as I need (and love) my computer very much. :D So, I decided to see which would be faster, the default 11x Multiplier and 200MHz @ 2-2-2-12 (12 tRAS is the ownage), or 10x Multiplier and 220MHz @ 2-3-3-12 (2-3-2 errors our on memory @ 220). The follow result was....

11x200MHz @ 2-2-2-12 1T 2.7v = 2.2GHz CG Claw 1:1
AM3 = 65, 638

10x220MHz @ 2-3-3-12 1T 2.7v = 2.2GHz CG Claw 1:1
AM3 = 65,911

I used AM3 as it's pretty quick, and pretty CPU bound....also, I kept my videocard at stock as well, no overclocking there. Running the latest official drivers. A small, but still notable increase with the FSB higher 1:1 even with looser timings...feel free to make some of your own comparisons and post them, I'm interested to see what's good or not. I'd like to do a comparison with dividers, but I'd have to do 240MHz to get 200MHz memory to keep it even, and I can't 1:1 240MHz with my memory (something is holding me back, I think it's the board but it could be bunk memory). Meh, at least it does 2-2-2 @ 200MHz. :) Just be sure to make even comparisons in clock speed, so we have something to work with (include stock speed bench).

Edit: Oh yea, and I rebooted between tests and kept all the settings the same so noone got an advantage in anything loaded into memory.
 
No input at all? For all the inquiries for this info you'd think someone else would try a few scenarios out....guess I'm solo on this.
 
nah, i'm with you. :D

i think the extra bandwidth helps out our single channel setups more than on the dual channel rigs that reviewers typically use.
though, just out of curiosity, can you get latency figures for both 220x10 and 200x11? :cool:
 
i beleive in tight timings first, and then bumping up the Mhz. I run mine @ 225mhz, at 2-3-2-6. I could go higher, but I prefer tight timings.
 
I like tight timings too, but speed is speed reguardless of how you get it. :D I'm just tired of the usual A64 threads that go something like this:

Should I raise my FSB and lower my multi? What about a divider (still always causes a debate)? Are tighter timings better?

I want some apple to apple comparisons to show the true effects, and would like some socket 939 input here as I'm still on my single channel roots atm. If I can get my memory to run 1:1 error free @ 240MHz I'll be showing 1:1 240MHz/1:1 200MHz/Divider 240/200 to show the real effect of all three. But I'm still at work so I can't do much. :p
 
have you seen this thread?
i too got fed up with everything and ran some benches proving that cpu speed owns ram speed, though music conversion benefits from higher ram speed more than other things.
 
No, I hadn't. Interesting....I'mma see what I can't do here for some more benches..

And right now (220x10) I'm at 47.4ns latency....I'll get 2-2-2-12's as well as any other settings as I go along. ;)
 
Huge forum at ANand about dividers..comclusion is that using one does not harm performance on an A64. =)
 
Some new results:

Memory running at 240MHz with timings of 2.5-3-3-12 2.8v and CPU 10x with 1.65v 2.4GHz even got a score of:
AM3: 69,038

While memory running at 219MHz with timings of 2-3-3 2.8v and CPU 11x with 1.65v at 2408MHz got a score of:
AM3: 67,208

So what's the conclusion? FSB > Overall speed and timings on socket 754, as the single channel memory puts it on a pinch. I doubt you'd see that notable of a difference on socket 939....I'm going to try again shortly to get a score for a divider running 200MHz 2-2-2-12 and 240 on the CPU to see if the divider/lower timings do better/worse/the same. The first round I accidently diddn't drop the multi and my computer posted (kinda) but would black screen on me....couldn't get into the bios/load windows/etc....just alive enough that the auto-reset feature diddn't kick in. But she's back now and on my first boot it wouldn't let me overclock for some reason...will try again shortly.
 
You must log in or register to reply here.
Back
Top