Hi, I would be pleased of any comments about my situation
My system is an E4300 with a Q649 batch, P5B Deluxe, 2Gb Corsaire XMS2 TWIN2X1024A-6400 (4x 512Mb 2 by 2 matched modules) 5-5-5-15 timings set, Zalman 9700 AT (4 pins instead of LED brand with its 3 pins for better auto adjustment, even if marginal lower performance at maximum settings)
Stability is for me an Orthos 16h Large FFT session at priority 1 to keep the system functional for other tasks. I don't use TAT loader since it is too theorical and agressive
All mb AI, speedstep... stuff is off following everywhere guides for this mb
I have a good cooled case, as the system temperature (motherboard sensor) is around 40-41 degrees all the time
I tried to overclock with some issues:
1- Bios versions: I can get my stable overclocks for the 0910 bios revision. 1004 is less stable, and 1101 won't post at any of my rock solid overclocks on 0910 bios
2- My best stable overclock is FSB 326*9 = E4300 @2934 Mhz, so pair with the X6800 :-) for a vcore of 1.45 bios (1.40 load on CPU-Z) and all chipsets/FSB voltages minimal fixed value
TAT temperatures are around 60-64°C depending on room temperature (24-26°C)
I can get 333*9 stable but with vcore over 1.47 bios and TAT temps hitting 69-71°C
3- My DDR2 is rated 800 Mhz. All my overclocks are stable only at 1:1, so that my memory must be underclocked to 2*326 (652Mhz) to be stable. RAM modules are rated at 800Mhz for 1.9v and are cooled with heatspreaders. I keep by default the vDIMM on 2.0v. Every attempt to increase my memory speed tends to be unstable even at something like 750Mhz. If I set my CPU and all to default settings, my modules are rock stable at their default 800Mhz speed. So if I increase the FSB and desynchronize memory, it tends to be unstable even at low speeds (like and FSB limit for the RAM). I tried with increasing memory voltage up-to 2.15v without success.
My questions are: - For point 1: Anyone encountred or can explain my bios issues with the P5B Deluxe and E4300
- For Point 3: I heard once, and never seen elsewhere, that oveclocking is less stable on 4 RAM modules than with 2 modules. Could it be the cause of my overclocking instability if I try to run DDR2 close to its rated speeds?
Many thanks for all your useful comments on any of these points
More about :overclocking e4300 temperature ram questions
I have already tested it. It won't post. I also tried 401-402 and some more FSB, it won't post even on lowest multiplyer. Don't know if MB issue (vFSB and v chipsets maxed for testing) or CPU FSB ceiling (strange though) or memory issue with FSB
Well, I see, I could try to remove two RAM modules from their banks...
I'm stable at FSB 326*9 = E4300 @2934 Mhz, so pair with the X6800, 1.40 bios vcore, so, yes, I'm already lucky
Any further increase in speed, needs too much volts
I'm testing hardly 1.55 vcore bios (1.50 real) for 9*345: 3105 Mhz, it's going on... testing orthos
But, coretemp/TAT are high: 74C.
As it doesn't throttle, is it risky to run these temperatures / volts on air cooling?
Also, at idle FSB 200*6 vcore 1.10, Coretemp/TAT both indicate Tjunction 35 for Tambiant 25C. Can I assume TAT real is TAT shown minus 10C? In other words, at low volts/speeds, Tcase=Tjunction=Tambiant or Tjunction is stil 10-15C above Tcase on idle low speed / volts?
I can't believe it, I tried to reseat my HSF many times, not a degree gain on Temperature.
Finally, I read on a french forum users feedbacks on applying the thermal compounds. I have the MX-1, which is shown on many tests to be better than AS5. Well, on that french forum, people noticed important temperature drops when applying larger amounts of compoun than the recommended grain of rice in the center.
So, I used a soft lint free cloth tinted with the very thick compund (for whom know the MX-1) just to put a quiet virtual layer of the compound on all HSF and CPU surfaces. This should be just enough to fill micropores of surfaces.
Then, I applied a line of compound on the 2/3 middle part of CPU, the line is not larger than a grain of rice, but it is long enough (2/3 of CPU width)
Then, I applied the HSF, screwed it, and twisted it some few degrees , about 3 times.
I fired my system: got a 10°C load drop in cores temperature with TAT and coretemp for same room temperature
I know 10° is much, but that's what I got
Maybe I have a really concave CPU IHS (Zalman surface has a mirror finish by default)
This decrease also explains my 35°C cores temperature on 6x200 FSB and 1.10 vcore (10° above ambiant)
Now, my E4300 is testing Orthos at 1.525 vcore, TAT 65°C, for a 9*345: 3105 Mhz
Will post my results later.
I didn't test removing memory modules. In fact, I think I'm hitting the roof. I can gain more FSB, but vcore would be too high
My hint for all:
To check if your HSF is well seated: set the CPU at lowest multiplyer and lowest vcore. Max out your CPU fan (this is important). You should have a TAT reading mearly equals to ambiant +/- 3-5 degrees. If above, your heatsink can be improved