do x2 a64 cpus have problems with memory speed the higher they go?

[iklebrit]

my processor works well at 2600mhz, 260x10. (dfi nf4 ultra 6-23 bios)
it also works fine at 2400mhz 300x8. (memory gskill 4400le)
but pushing to 270x10 iexplorer crashes or prime crashes. unless i use a memory divider. (same with 280x10 almost exact same type of crash)
(300x9 300mem using memtest passes 50times test5, but will bsod on bootup)

when i use the 9/10 divider, 2700mhz is stable using 300x9 (270 memory) which is odd.

i noticed on the ocz forums that tony had reported that using an x2 cpu the higher you clock, the less ** memory will overclock. is this an amd implemented thing do you think? with a 1/1 divider only???

im testing 311x9 with 9/10 divider and its running prime95 (280mem)
not sure if its 100% stable but its been running quite abit longer.

anyone else experienced similar problems?

 

[Ladforce]

could be the divider thats unstable

 

[Jimbo Mahoney]

A lot of people have experienced similar, myself included.

It's because the on-die memory controller is often the 'weakest link' and since it runs at CPU clockspeed, the higher you push the CPU, the higher you push the memory controller. Giving it high clockspeed memory to deal with can cause stability / prime errors etc.

I always use a divider to keep the memory at around 200 Mhz when clocking a 939 CPU for the first time, then when I find the limit of the CPU, I change the dividers until my memory is running as fast as possible too. ie CPU speed >> memory speed.