RAM MHz increases with CPU MHz!

Alright ive got an ASUS A8N-E mobo, great mobo i might add, for not being SLI... My CPU is an AMD Athlon 64 3000+ Venice (yay skt 939!) Anyways, I have my CPU set at 245MHz and i recently noticed that my RAM was running at 245MHz. This is a no no as i am using valueSelect from corsair Anyways, is there a way i can get my RAM speed to be at about the 200MHz area with my CPU running at 245MHz+?

 

So 245 Mhz is unstable ?? Just cas its cheap mem dont mean it wont OC well ..I had a 512MB stick of PC2100 that ran PC3200 fine and it was the cheapest stuff newegg had....

 

At 250 it becomes unstable... Unfortunately. Theres no way i can get my mem to run at "by SPD" on this mobo? I cant find the feature..

 

have you upgraded bios on that board? there may be a new version which lets you do this or gives option to specify ram speed

 

sorry i fail to see the problem your 5 MHz from your memorys max.... thats a good thing....are you trying to get more core and your mem is holding you back ?????

 

Exactly... My mem is holding me back and im looking for more core speed... on my old rig i just upped the cpu speed and kept the mem speed "By SPD"..

 

Thares no timing settings on the mobo ???? no memory ratios ???? Memory voltage ???

 

Theres timings and voltage... Not too sure about Ratios, ive never seen any options for that..

 

try to loosen the timings and maybe raise hhe voltage a bit.....

 

Go by the book/manual, page 2-21.

May be you are lucky with cheap pc3200, just very unlikely. One of the major advantages of NF4 is you only see minor performance loss if you keep ram at rated speed, or close by - or better oc it as much as you can but not more than any mem-test can be passed. Things work better with no ram-errors and right now I will bet you have tons. Valueselect normally does not run 245fsb stable. Sometimes huge difference between what appears to work, games run etc. and what can pass ie. Prime95, different ways of overclocking. As said you will not notice much loss so torturing ram because higher must be better is no good. Better to kill some latency if possible, not sure they react friendly to other timings than SPD but may 1T will work with ram/cpu/bios - a bit of a mess with that setting. Not as important as it was on NF3 though and again be sure you test properly. If 1T wont work no matter what you will get blue screens, perhaps worse. A quick memtest86 will give a hint.

Most speed gains come from cpu and its memory controller so lock that memory down to rated specs and see how much you can get out of cpu - test that as well. Then you go back to memory and work on that. There is still some gain in keeping 1:1 with cpu but that means $$$ with a 3000 cpu. Way too much for what it is worth.

If none here is know Asus bios go to some motherboard forum, get best bios version - see what others are doing. Set it up as recommended. Best result guranteed.

You did lower HTT frequency to 3 or 4 I hope? Just in case you have wonder ram/motherboard combination HTT can give problems, 3 or 4 dont matter for speed.

 

Yea... I tried the Hyper Transport at 5x... didnt work.. so ive got it at 4x and thats stable... will go check out the asus forums..

 

Ok... I set my RAM to run at 333 MHz and my processor at 250MHz at my RAM is running at 200MHz now... give or take.

 

Let them pass at least memtest86, shouldnt be hard. So now you have found 1 ratio you could find another so you get perhaps 210 or 220fsb. You oc or not? Not much gain though and if you have to slow down any latency setting at all it will be counterproductive. Try 1T and 200fsb with memtest86. 1T vs. 2T Ive not found to be a major question on NF4 but anything goes. From what Ive read about Valueselect they are not keen on better timings so perhaps 1T is what you can do. Should be testet once you are done with cpu.

I dont suggest Prime95 because I think you need computer for millitary purposes but because Ive often been able to pass memtest86, SuperPI and whatever and still get blue screens and instability. Not so with Prime95. Picky and hysterical but once passed anything else will too. Read "stress.txt" from download, says it all and you must not confuse success with proven to work in all games. As author says it does not test 3D what so ever. 100% stable cpu and ram should make it very easy to get 3D going though.

Much of this oc is about getting highest numbers no matter what so not all are too happy about programs like P95 It does kill wet dreams of world records thats for sure - just so you know... All a matter of degrees. 1 error after 5 hours will not blow up computer, 1 after 1 minute and computer is very unstable. P95 only give 1 error then it stops, heh.

A little trick you can use to provoke memory errors is to tick "Custom" in torture test, set min and max fft to 4096 and "memory to use" to half of total physical memory. That way you can at least surf while testing and it seems to tricker errors faster than with randomly sized fft. Read about it on some forum long ago, have found it to be true. For a real test of memory, like scandisk for hd, you should use memtest86. P95 is for testing stability while racing 100% not looking for bad "sectors" Bad module will give error of course but memtest86 much better suited for finding them though - tests almost all memory, P95 cant since it runs in OS. Memtest86 not allways foolproof about oc-stability, about minimum really but you soon find out.

 

Yea... Ive got my RAM set to 333 and my processor set to 255MHz and my RAMs only running at 208MHz now... If i can goose my processor up to 300MHz, then ill try my RAM at 133MHz.