Sounds like luck of the draw?

[MTSteel]

I isolated the board first, just dropped the memory to 100, and the CPU multi to 5. 300x5 booted no problem(I was not loooking for the limit of the board as 300 is more than enough for me)
300x6 booted 300x7 POST but looped back to POST :0

Since my goal was 2.4G with a 3000 I tried 265x9 memory divider still at 100 to elminate it, HTTx3.
It will boot up no problem, but fails prime within a minute at 1.425V
Upped V in .25V increments up to 1.6V. Still failed prime95 immediately.

Sound like a poor OC chip to you fellas?

I was able to prime for 12.5 hours(before I stopped it to try ohter settings) with the setup i am using in my sig.

Wish I had access to a different CPU to at least try one and see if I could go higher.

 

[imported_Kensai]

Bad luck. 🙁

 

[tr1kstanc3]

it may be your memory controller. try running your ram at 2T timing just to isolate the problem. i'm running my ram at 2T because 1T caused errors almost immediately in prime. i picked up a 0503 3000+ and it hit 2.6 @ 1.55v but its luck of the draw.

 

[edge929]

I'm in the same boat.

0504 3000+ Winnie
Failed Prime in less than a minute at 2.2Ghz @ 1.55VCore (according to CPU-Z)
3x HTT
250FSB
Memory at 100 to take out of equation (Corsair C254400 so according to Anandtech, it "should" run 1:1 all day)

I have heard that some rigs will fail Prime even BELOW stock settings. Can't confirm it though as I'm still tweaking my system.

 

[deadseasquirrel]

Originally posted by: MTSteel
I tried 265x9 memory divider still at 100 to elminate it, HTTx3.

Are you leaving other memory settings at default when you do that? A /100 divider makes the memory run at 100mhz, yes, but, if you leave all the other settings on default, they will default to much tighter timings. Especially with this DFI board, and with that RAM.

I have pretty much the same combo:

DFI SLI-D BIOS 2/17
A64 3000+
2x512 Corsair VS 2.5-3-3-7 2.8v

I suggest you manually put in the memory settings and don't leave ANYthing on default. This will assure that something is not trying to run faster even when UNDERclocked.

I'm at 313 * 8x with a /133 divider (ram at 208). Give the VS more voltage too. This board likes 2.8v for the ram.

We often times like to just dump off a less than great overclock as "(un)luck of the draw", but the fact is that with these A64s and this DFI mobo, it's not a simple task anymore. Nobody can give you an A, B, C list of steps to do to get the overclock. Because of the individual behavior of each memory controller on the A64s, what works for one may not work for another.

Try more combinations. I've seen 300*8 work fine and 257*9 fail immediately (with RAM being taken out of the equation both times). Some chip-mobo combos prefer 8x over 9x, etc.

It's taken me weeks to find a clock I'm happy with. I'd play around with multi's and HTTs until I found one that would get into windows, then I'd run an 8M SuperPI. If it passed, I'd run a 32M. Then, if that passed I'd, bump up higher. Kept doing that until I had to come back down or try a different combination. Then, I'd run Prime overnight. (Didn't need to run Memtest since I was keeping memory right around 200 all the time.)

 

[yliu]

I personally have found more success not always setting the memory to 100mhz (1:2 divider) but rather using a divider that makes my memory run as close as possible to 200Mhz without going over. Perhaps you guys could try that and see.

 

[CaBoOse999]

I'd say bad luck. Alot of people are getting similar results with the new Winchesters.

 

[GML3G0]

well, someone mentioned 2T command rate. but the memory controller isn't being stressed in the large FFT test is it? so lowering it to 2T won't help in that respect. besides, 2T will lower performance greatly, it's not worth it.