Built up a new A64 rig recently, based off an opty 144 and DFI SLI-D.. To be frank its been a mixed bag... Not had any issues with bsods, reboots, no boots in the slightest. Its been good fun to play with... However having huge issues getting a stable OC. The previous chip owner had 2.9ghz from it, while he didnt run prime etc I would expect to see 2.8-2.9 fully stable from it. However this isn't the case. Finally got the ram sorted, it'll run memtest (tests 5 & 8) all day long without a problem (tested @ 310 1:1 3,4,4,7, 305 3,4,4,7 and 300 3,4,3,7 also run 275 2.5,3,3,7 and 280 3,3,3,7 mixed 1:1 and 9:10 ratios) It'll also run windows memtest for 500% runs at each of the above (and combo's in between) Obviously the Alpha's vary from 2.5,3,3,7 to 3,4,4,7 but again the memory tests pass fine. To me this would suggest the memory isnt an issue here. SP2004 tends to bomb out on the higher fft tests (ie 648k or above) with the most common being 896 and 786k (iirc) This suggests memory/chipset as a likely cause. Played with varying LDT and chipset voltages, but again things don't seem to stabilise a big deal. Superpi 32m runs fine, and does so when SP/OCCT fail after 20 seconds! Bios is the 702-bta modded which seems to be best, running ram in orange slots and the orders been swapped to find the most stable ram config. Yellow seemed a little worse but need retesting. LDT multi - 4 seems most stable, although theres not much difference using 3, LDT voltage has been varied from 1.3 - 1.5 with 1.5 giving best results. Fans over PWN and ram area, CPU's cooled via DD TDX with dual 120.1's Load temps are under 40 deg. So, anyone got any ideas on this? Can anyone spot anything I'm missing or havent tried as of yet? Cheers
Yup run 1.6 on previous occasions... Need to re-test tonight with modified timings and 1.6v... Might whack it up to 1.7 and see if that'll help too...
Do you want it to die? Theres an issue in the silicon that will cause the CPUs to fail above 1.65V, period. Unless you have word that this has been sorted, I would think twice about throwing that much voltage at it.
Have you tried a different BIOS Highland3r? Maybe that one's just throwing a wobbly... I have to agree with the Chipset voltage adjustment suggestion. It seems a bit of a strange one to me. I've been having strange issues with my rig recently, it just seems to be fudging around with me. Must be DFI boards with Opty's tbh! What stepping is your 144 out of interest?
Its a CABNE, will get the full code when I get home.. Tried various bioses, the 072BTA seemed to be the best of the bunch...
Hmm... curious... Tried fishing through Dfi-street.com? I'm sure there's someone out there who's had a similar experience to you.
Leave all the settings as they are and lower the multi. Test Prime, OCCT, and Super Pi. MemTest stable doesn't necessarily mean that the RAM is stable. You could also test with a divider on the RAM, to make it almost stock speeds. This of course, just would test the CPU and board.
FYI, 1.7v did nothing more for me on mine. Just in case Could be a dodgy memory controller if it doesn't like being stressed 1:1. Like mine seems to be. Makes sense, since the last guy never primed it.
It runs 1:1 better than it does a divider Its currently priming at 300*9 1:1 so we'll see how this goes... Dropping to 8* multi made things worse, as did 1.7 NB... The CMOS reloaded thing on these boards is a god-send, tis great being able to jump back to a saved config as and when
See, now to me thats odd! But mines the same sometimes. Overclocking a64s is so strange sometimes. 9x300 is nothing bad tbh 2.7Ghz. You might find the memory controller might be hitting its "peak of stress" so to speak at 300. Hence why it doesn't like 1:1+Stress any higher.
Yea thought the same, but looser timings + 2t + bank interleave off didnt help things, which you'd expect if the memory controller was maxing.. Might be able to borrow a 144 CAB2E or 3700 CAB2E to test, otherwise need to get the 3500 out of the Neo2 and give it a shot... should make 2.7ghz too and it doesnt have an issue with dividers...
did the previous owner give you proof it was at 2.9 ... quite easily ask him and he might say opps sorry typo nevermind ..
I'd be tempted to not care.. if it never crashes, reboots, or BSODs, then who cars if it wont prime? I used to have a rig that wouldn't pass prime, or memtest, but never ONCE crashed in real life, so I left it. I had it for years, and it never once gave me any trouble. BTW.. what kind, and size of RAM are you using to hit 300?
Inclined to go with that, need to test some more 3D on it though... IIRC the last 3dmark loop crashed the pc lol! Its PQI TCCD.. Timings are a bit poo, but speed wise its ~ the same as 275 2.5,3,3,7
