Overclock.net › Forums › Intel › Intel Motherboards › [Official] --The Asus P6X58D Premium & E Thread--
New Posts  All Forums:Forum Nav:

[Official] --The Asus P6X58D Premium & E Thread-- - Page 68

post #671 of 6382
At 21x200 with hyperthreading, Hyperpi 32M x 8 reliably completes with no errors with Vcore 1.3375v, QPI/Dram 1.23125v. Windows and applications seem to be okay. But LinX crashes quickly, sometimes not even completing one run.

Hyperpi and LinX both show 100% utilization on all eight CPU units, but CPU amps, power, and temps are much higher with LinX. Peak core temps are only around 80 with LinX so I don't think it is a heat-related failure; it just seems to work the core and IMC much harder and finding the instability faster.

I'm going to try again without hyperthreading to see if it makes any difference. Maybe LinX is more efficient at doubling each physical core's work than Hyperpi.
post #672 of 6382
Crashes - Like it BSOD's or it just errors out? How much memory selected in Linx?

If it BSOD's what type of BSOD?
post #673 of 6382
Quote:
Originally Posted by BlackDragon24 View Post
Crashes - Like it BSOD's or it just errors out? How much memory selected in Linx?

If it BSOD's what type of BSOD?
Either hits an error and stops the run, or BSOD (I've seen them all at various points... 101, 124, 9C, 1D, a). It is set for problem size of 25000 which is like 4766 MB, I think.

At extra 0.05v or so of Vcore (1.3375 > 1.3875) is needed to be stable in LinX (20 runs) or Prime95 (large or small FFT) compared with hyperpi. Temps with LinX or P95 are about 15C higher than hyperpi. Haven't tried with only four threads yet.

Also tried uncore:qpi of 1:1 3600Mhz:7.2MT/s (2xDram + 2) and I got BSOD within a few seconds of LinX, even Windows BSOD at one point. This was with QPI/Dram of 1.25. Obviously my particular setup does not like this ratio. May try 3400Mhz which would be 2xDram + 1. I tried with auto RTL and set manually to 65-66-66.

One thing I did notice, if manual RTLs are set, the actual values reported by CPU-Tweaker are each 1 clock cycle higher. Same with the BIOS reported values. Strange.

I'm going to be trying another CPU on Monday, a W3520. Hopefully it will be better. I didn't do to well in the CPU lottery with the 920 or 930
Edited by ekovalsky - 2/27/10 at 9:53am
post #674 of 6382
Given your circumstances, it is becoming apparent that you have a chip that does not have a stellar IMC. Hence the extreme instability at the slightly higher uncore frequency, and the need for so much more vcore when really stressing the memory controller in Linx. I've had a couple of those chips myself, some of which wouldn't even do 200 Bclk and 1600 memory.

Double check your memory settings and voltages to make sure everything is straight. Sometimes memory issues can be masked by pumping a bunch of vcore. I forget what type of memory you have?
post #675 of 6382
Hey guys, I'm trying to bring my beast up to 4.0 running stable with HT on. Right now I'm running at 4.2 Ghz stable but the temperature is like 90 under prime95 test. I use the same setting for 4.2 Ghz and lowered the BLCK to get 4.0 Ghz and that won't POST. Why's it like that
post #676 of 6382
Quote:
Originally Posted by BlackDragon24 View Post
Given your circumstances, it is becoming apparent that you have a chip that does not have a stellar IMC. Hence the extreme instability at the slightly higher uncore frequency, and the need for so much more vcore when really stressing the memory controller in Linx. I've had a couple of those chips myself, some of which wouldn't even do 200 Bclk and 1600 memory.

Double check your memory settings and voltages to make sure everything is straight. Sometimes memory issues can be masked by pumping a bunch of vcore. I forget what type of memory you have?
I think you're right about the IMC. Memory is Crucial Ballistix 6gb running at 8-8-8-1T, 1.68v with the 2:8 multiplier at bclk 200. It passed tests at DDR3-1603. I have tried repeating tests with the 2:6 multipler i.e. DDR3-1200, without any improvement in stability. Amazon had the Corsair C7D kit on sale recently and I ordered a set, but it was on backorder. When I got the Crucial, it was $100 cheaper than the C7D plus the LEDs look pretty cool in my caseless setup.

The high Vcore I need for stabilityy definitely may be compensating for the poor IMC. If hyperpi 32M x 8 runs with 1.3375v Vcore, that should be pretty close. The much larger amount of RAM utilized by LinX and P95 Large FFT must be causing the problem.

I justed booted at 4.2GHz with HT on with everything set to auto except for the RAM speed and voltage. Nice and stable in LinX at 1.425v Vcore and 1.375v QPI/Dram ! A little toasty though even with the NH-D14, my hottest core has hit 94C. I guess I should start working backwards from that trying to get temps down as much as possible.

Hopefully the third try will be a charm, and the W3520 will be a winner.
post #677 of 6382
Well, to tell you the truth the IMC could be fine and the memory could suck. I had a C0/C1 that wouldn't run 1600 with some OCZ rated for 1600 unless I gave it 1.35 QPI. I swapped out the memory and replaced it with elpida hyper and it did it with 1.22 QPI.

So it may not be the cpu. Especially if you are pushing 1.68v on the DRAM
post #678 of 6382
anyone help?
post #679 of 6382
Quote:
Originally Posted by sora1607 View Post
Hey guys, I'm trying to bring my beast up to 4.0 running stable with HT on. Right now I'm running at 4.2 Ghz stable but the temperature is like 90 under prime95 test. I use the same setting for 4.2 Ghz and lowered the BLCK to get 4.0 Ghz and that won't POST. Why's it like that
Try just disabling turbo and see if it works.
post #680 of 6382
okay. is it weird to run RAMs are like 1500 speed when it's slated for 1600?
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel Motherboards
Overclock.net › Forums › Intel › Intel Motherboards › [Official] --The Asus P6X58D Premium & E Thread--