On die memory controller heavily affected by cooling

[Oc-cba]

I hope my english writing skills are good enough to make my point, I guess maybe just writing step by step my expierence will help:

1-a643200 on DFI nf4 - under xp120 got stable @ 2,45ghz / 245 mem (bh6)

2 -same chip went under custom phase (limited atm to -20c so atm it's been modded)

3- after the phase experience, i got a 3000 CBBHD to try, at firs would crash @ even 2,05ghz, I thought it was a dog so old chip (3200+) goes back to the same config as before and could not go @ anything over 2,25

4- thought it could be the sink seatting bad, so I got multiplier down to 5x and try, same reset at anything above 230 HT speed

5 - posted for help (n00bness atack heheh) thinking the DFI was damaged while under phase (as if it made sense some damaged board would not overclock as well, I've always believed that if it's gona break it breaks to dead and not only to worse oc)

6 - Just to try i set the mem to 100mhz in genie bios, ran memtest and it worked all the way up to 400ht , so the board was not dead, it really CAN be the way i seated xp120..

7 - reseated hs over the at firs bad 3000 CBBHD (with standar thermal grease instead of as5 as it was faster to spread, plus tomorrow maybe i have my phase back) and could go up to 312ht with multi @ 8x as posted in my other thread...


from all this I conclude that even with low final mhz on the cpu, the try to go for 220 or 230 mem with bad CPU cooling was impossible, so the way you cool the cpu WILL affect your mems oc even @ low cpu mhz.

hope this helps someone (as bachus_anonym in his quest to defeat Onepagebook ) as reading that thread inspired me into posting this. And after my brain burning to be able to write this I hope I'm not telling an old storie (damn lazy to check but not to write hehehhe).

[Magnj]

So with all these crappy CBBID's that don't OC well ( I assume most of the results are on stock cooling). If They were under a better Heatsink, maybe they wouldn't be terrbile after all?:hopeful:

[RaptorRaider]

I think he's saying that they need to be under a worse heatsink.

Those Winchesters don't like cold. What he's proven, is that it is the memory controller which makes them overclock so bad.

[Oc-cba]

I think he's saying that they need to be under a worse heatsink.

Those Winchesters don't like cold. What he's proven, is that it is the memory controller which makes them overclock so bad.

Maybe my english is to hard to read, but what I meant is that cooling will affect a lot of how much mem oc you get, as well as final cpu ghz...

So in no way i said they have to be under a worse cooler....

Offtopic: i had 3 winchesters and the 3 of them would boot fine subzero, was it only Neo2 bug or am I lucky with the cpus i got???

[saaya]

So with all these crappy CBBID's that don't OC well ( I assume most of the results are on stock cooling). If They were under a better Heatsink, maybe they wouldn't be terrbile after all?:hopeful:

some people just hear what they want to hear hehehe :P

Maybe my english is to hard to read, but what I meant is that cooling will affect a lot of how much mem oc you get, as well as final cpu ghz...

So in no way i said they have to be under a worse cooler....

Offtopic: i had 3 winchesters and the 3 of them would boot fine subzero, was it only Neo2 bug or am I lucky with the cpus i got???

i think it was a general bug that dfi fixed on their board/bios

[$0m#0n#]

some people just hear what they want to hear hehehe :P

[Magnj]

I can hope.

[flutie98]

hmm well ill besure to tes this out on my water situation

[saaya]

I can hope.

hey, no offense man

[ozzimark]

just tested that theory with my clawhammer.. no go, 50c load give same memory stable as when i blow cold air from outside over the heatsink to drop the temps to about 15c.. i didn't check overall overclock cause i'm feeling lazy

[Magnj]

blah. I think the only answer to Good winny Ocing is to steal other peoples 0448 SPAWs

[Ackbar]

Could this be a cause for the winnie "cold bug"? It seems like if this is true that cooling too much makes the memory controller go down the sh1ter. A better memory controller might fix the cold bug!

[funkdog]

Isn't that what is really the problem? The memory controller has gotten worse later in production. Stable fsb at higher frequency was much better on the 0446-0450 chips. I can't go above 320fsb even with a divider on my 0451... looks like those chips were the best of the bunch so far.

And what I say by "worse" really isn't worse. I mean paying for a chip that runs at 1.8ghz and being able to get it to 2.6ghz really isn't a "crappy" chip... since you actually just paid for a chip that is only supposed to run at 1.8ghz..

they just produced a great bunch of chips in late '04 and now they don't.