• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

God, could choosing memory be any more difficult?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Qin

Registered
Joined
Mar 3, 2004
I'm man enough to admit I need some help here guys. I just put together a new system about four weeks back and have since had to RMA my memory twice. The setup:

Intel 2.4C M0
Abit IC7
Ati 9800 Pro
Western Digital 36GB Raptor

My first set of memory was a pair of Mushkin PC3500 Level One modules, I found out only after getting them that the CH-5 chips they used had major issues with my Abit IC7 motherboard (can't go past 250mhz when using a 5:4 or 3:2 ratio). I then did some research on Abit's forums and found a lot of people were satisfied with GEIL's Golden Dragon line, so I then purchased a pair of PC3500 Golden Dragon modules.

Upon recieving them I ran a few tests, everything looked good (I could use them at a 5:4 and 3:2 ratio above 250mhz), and so I sent back my Mushkin. Running my Golden Dragons at 1:1 and at 2.8Vs I found I was able to get all the way up to 250mhz with the stock timings (2.5-3-3) before I started getting errors in memtest86. Impressed with their ability to run with at a 35mhz overclock with their stock timings I set out to now find my CPU's limit.

So I put my ratio at 5:4 and found that around 270mhz I had to raise my vcore to 1.5V in order keep Prime95 error-free and at 275mhz I had to raise it to 1.65V. I have no intention of raising my vcore above 1.65V and so anything above 275mhz was no-mans land for me. I brought my FSB down to 270mhz and just to be sure my memory wasn't causing me problem I ran memtest86 for a few hours and found it error-free. I now considered my CPU as what was holding me back, my logic being that my memory ran memtest86 error-free at 270mhz (5:4) and that increasing my CPU's vcore was the only way to now achieve a higher overclock.

But something kept bothering me, my chip's stepping (M0) is supposed to be the best of all 2.4Cs', and even the worst ones are hitting at least 280mhz with their default voltage.. but I was having to increase my vcore to what I consider my chip's maximum (1.65V) in order to just hit 275mhz. Finaly I decided to check something, I set my vcore to 1.65V, my fsb to 280mhz and decided to run memtest86 to see if my memory was what indeed was holding me back after 275mhz. I mean I had just assumed that since it could run 1:1 at 250mhz, it should be good at a 5:4 ratio to, at the very least, 290mhz right?

Well, I was wrong. I found at 280, 279, 278, 277, and even 276mhz I would immediatley get thousands of errors in memtest86.. and at 275-271ish I get hundreds of errors. So apparently when using 5:4 (and 3:2) ratios my memory just craps out at anything above 270mhz, kind of strange considering what it can do at 1:1, but at least now I know that it has in fact been my second set of memory holding me back and now my CPU.

So now I'm asking those of you with experience running 2.4Cs over 280mhz on Abit boards, what memory acctualy works? Please don't just say 'BH-5', a lot of memory now a days which used to use BH-5 chips now uses BH-6 or even CH-5 chips. I need to know exactly which memory, brand name and model, I can order and have it be guaranteed to work.

TIA.
 
Last edited:
As I've already suggested in your other thread, you may just have a rotten chip...
But if you really want to try out other memory, then how about Mushkin PC3500 Level Two modules? Should still be guaranteed BH-5...
 
As I've already suggested in your other thread, you may just have a rotten chip...

Acctualy, up untill just a few minutes ago I had always just assumed that memtest86 only tests your memory, but would running your CPU past it's limit also cause errors to arise?
 
interesting,i would assume a failing cpu could cause errors in memtest, but since i dont use it, i cant say for sure. secondly 275 is on the low side for a 2.4c but in my opinion still within a normal overclock for this chip.I also suspect running in async is probably more critical for memory timings than 1-1. have u tried using 3-4-4-8 timings just for testing?
 
Yes, you can get memtest errors from pushing a CPU too hard. The way computers work, you cannot avoid using the CPU when using you memory (data is stored in the cache before it is stored in memory). Memtest coding does its best to avoid the cache as much as possible, but your CPU will effect memtest results, no doubt about that.
 
Back