• 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.

Memory overclocking pointers needed

L

laserbeam

Limp Gawd
Joined
Jan 9, 2005
Messages
168
I've just started overclocking the memory (Corsair 4400C25PT matched pair) on my new system so I can find it's limits. Have throttled back the cpu to 8xFSB and raised the FSB to 225Mhz. The timings I'm running right now are 2-2-2-10 (memtest86 v1.51 reports tRAS 10 as optimal for mem bandwidth).

In order to get, stably, past 223Mhz FSB I conservatively raised DIMM voltage to 2.8V. All other voltages are default. Tonight, when I get home from work, I'll continue raising FSB until I become unstable again, at which point I expect to have to raise DIMM voltages again. Can someone give me an indicator of what is regarded as a "mainstream" memory overclock voltage? And is there a voltage I shouldn't go beyond?

Finally, when memtest86 reports errors, is there a way of recognising which memory timings I need to raise in order to fix things?


All help greatly appreciated :D
 
memtest crash usually means that you need more CPU voltage.

memtest errors usually mean that you need more chipset and/or ram voltage.

is there some particular reason that you are using such an old version of memtest?

tons of peoplesay that 3.3 volts is pretty standard, however unless you know that your sticks of memory use BH-5, BH-6 or CH-5 chips on them, then i really don't reccomend going that high if you want you memory to have some reasonable lifespan. i THINK that those sticks that you have use samsung TCCD on them, in which case, i think that 3.0 volts is about as high as you can safely go.

you should really put some system specs in your public profile before asking for help, though.
 
1.51 is the latest version isnt it?

DO NOT go to 3.3v on TCCD ram.... even 3.0v is pretty High..... 2.8-2.9 is usually max

Dont be aftaid to run a Divider on that ram to keep the timing tight. That ram is rated at 2.5-4-4-8 @2.75v I think to get to PC4400 Rating...... Which are terrible timings.....
 
One of the things I want to establish is what compromise of timings vs. mem. bandwidth results in the best performance. From what I've seen on sites like Anandtech and this forum too, lowest timings aren't necessarily the only route to the highest framerates.

I'm not too fussed by the advertised ratings of my mem. I'm more attracted by the fact It can do DDR400 at 2-2-2-5 (curently running at DDR 450) and can ramp up to DDR600+ at 1T - though I have to prove that. Somewhere in that spread I'm hoping to find a good sweetspot. And I have a feeling that it won't be found by sticking to minimalist timings at the expense of Mhz. Of course, time will tell :)
 
Basically, low timings > FSB in A64's usually....since the memory controller is in the CPU, it loves low latency.....so (i think anyway) 12x200=2.4 @ 2-2-2-5 is better than 10x240 @ 2.5-3-3-7. Clock to clock, low timings always run faster....but if you can't just kick up the multiplier (FX are only this lucky) then the only way to get an overclock is to raise the FSB, which requires lower timings (unless you use BH-5 or OCZ VX....but high voltages can fry A64's). So how many MHz do you need to make the looser timings compensate? Not much I don't think....on an athlon64 like 100MHz would be more than enough. And don't use memory dividers....ugh, performance killers.

Give the guide in my sig a read.....should help you out a bit. I'd say run your memory at 2.5-3-3-7 and see how fast you can get that memory to go.
 
You must log in or register to reply here.
Back
Top