Jump to content

Problems overclocking new system


Velvet45

Recommended Posts

I am unable to get this set up to 2.4 Ghz with either 11 x 220, or 10 x 240 settings. My processor has an 11 multiplier -CG stepping Clawhammer. I have increased the DRAM voltage to 2.7 and not yet increased the chipset or core voltage. I have relaxed my DRAM timings from 3-3-2-8 (stock) to 3-3-3-10. I have gotten as far as 11 X 215 -2365 Mhz booting into windows without problem. At 2400 Mhz (10 x 240) and dropping the LDT/FSB to x3 it will post into windows but the mouse froze and windows said it had recovered from a serious error. It is not prime stable at even 230 FSB- any suggestions? Should I bump up the Vdimm to 2.8 or 2.9 volts- I have heard this micron memory loves the volts. I am still using 8/27 Bios as I have not had any temp problems- Temps are 32C (CPU) and 29C (system) at idle- 47C while priming at stock settings.

 

Thanks

Share this post


Link to post
Share on other sites

Cyberfossil, No I have not tried one stick as of yet. I am not having any problems with the memory -passed Memtest86 at stock setting without problems. So if memory is the problem- I need to knkow what I can do to be able to run them both stable- more Vdimm.....etc. Thanks.

 

Anybody here?? I expected some kind of feedback but not getting much so far.

Share this post


Link to post
Share on other sites

Go see AG's signature> "bios screens galore" NF3 UT bios second page and you will see exactly how his set up is for EB stuff. Nobody is still using EB hardly so you ain't going to hear much and really it is so simple most are not going to stop and hold hands much.

 

His cpu volts and Vdimm and Vdd are all shown in his captures of the bios screens and that was with EB3700 and worked fine. You duplicate and you should be too. Of course use your multiplier and not the 9 he was using.

 

 

RGone...

Share this post


Link to post
Share on other sites

Lower your multiplier and increase your HTT/FBS. The idea is to find your memory's limits independant of what the CPU can do. Once you find how fast you can tweak your memory you can lower the HTT and raise the multiplier to find where the CPU craps out taking care not to to raise the mem freq over the value you deturmined in the first test. It may take using a memory multiplier (limit) to keep the mem in spec as you increase the FBS/HTT.

Share this post


Link to post
Share on other sites

Well, after spending about 24 hours working on this I am hitting the wall with this processor. Settings: 235 x 10 x 4 (LDT/FSB). Prime stable for 8 hours. Vcore- 1.55 x 104%, Vdimm 2.9V, chipset 1.66V. Currently memory at 3-2-2-8. Everthing 1:1, 1T.

 

I was able to run 250 x 9 x 3 prime stable with looser Ram timings- 3-3-3-8, but was not prime stable at 255 x 9 x 3. I guess the upper limit FSB for this memory is 250 at 2.9V:( The only thing I haven't tried yet is upping the Vdimm to 3V. I am not sure if that would help me achieve a higher overclock with this processor- I could not get it prime stable at 2.4 Ghz with any multiplier -tried 240 x 10, 218 x 11 with relaxed timings and was not prime stable. Going to 3V might get me to higher FSB, but unless I can run a 10 multiplier or reach atleast 265 FSB x 9, it will not result in faster clock. I still might try it though. Is 3V safe for OCZ PC3700EB without extra cooling on the DRAM modules? The only thing I know is my XP90 HSF is oriented so that it hangs part way over them providing some extra cooling- perhaps?

Share this post


Link to post
Share on other sites

Sorry to post this without waiting but I increased to 3V Vdimm and am stable at 250 x 9 x 3. memory at 3-2-2-8. Will not do 255 stable so I guess 250 is limit for me. I am not prepared to push beyond 3V for now as would require additional cooling.

 

The 3Dmark03 scores at 250 x 9 and 235 x 10 are very close with the clock speed advantage of the latter winning by a narrow margin. I guess I am not sure which to run- both are prime stable but the higher bus requires increasing from 2.9 to 3V for little gain. My processor is maxed out at 1.64V and will not do over 2350. What a dog- CG stepping clawhammer too! Memory is satisfactory though at 250 -tight timings:)

Share this post


Link to post
Share on other sites

Sorry you struck out with your CPU ):

I would opt for the slower FBS higher multiplier option myself. I've heard of memory modules running 3.0+ volts just dying after a few weeks of use. As your benchmarks prove, there is not a lot of benefit in running memory at higher freqs with the A64.

If your full load CPU temps aren't too high I would try running 1.50 X 110% and see how it goes. If it's not stable at 2400 with that then you can say you truly hit the wall.

Share this post


Link to post
Share on other sites

Ztech1, Thanks for the help. I did try 1.5V x 110% last night but it only primed for 41 min before fail. I guess you are right in that I have hit the wall with this processor. I guess I will try running the lower bus/higher multiplier and back off to 2.9V on the memory. This particular OCZ memory is guaranteed for life at up to 3V but .....I am not sure what the long term affects of heavily volting are 24/7. I want stability and longevity as well.

 

I may still try some additional cooling with the processor. My temps priming are at 50C at 1.6V and 53C at 1.7V so I dont really think it is a heat issue. The Panaflow is a 57CFM fan blowing directly on the XP-90 which should handle the cooling chores for this setup.

 

I still wonder if I might get better results with a more recent bios- I am still running the shipping bios (8/27) but have not flashed as RGone advised against flashing the bios just to achieve better overclock.

Share this post


Link to post
Share on other sites

It's possible you have a thermal issue. Sometimes a change of even 1 degree can be the threshold of running prime stable or not stable.

While it's not to easy to lower your temps to check for better stability, it's easy to raise your CPU temp and check for reduced stability. I do this with a hair drier on low directed on the HSF unit while watching the CPU temp readout. By modulating the flow to the HSF you can raise and hold the CPU temp where ever you wish. If prime fails at XX temp in 40 mins, a thermal bound CPU should fail sooner with a higher temp. If you raise your full load temp 5 - 10 deg and prime fails no sooner then at the lower temp your CPU is not thermaly bound and increased air cooling will not help. This test is also good if you have a stable CPU and want to find the temp where it becomes unstable. This will let you know if warmer summer time temps will be a problem, remember, for every 1 deg rise in ambient room temp, the CPU temp also rises around 1 deg - thats ballpark for average system/room temps, actual deltas plot on a curve but that's another story.

 

Unless a BIOS has more relaxed(slower) timings, it will not offer better stability. DFI BIOS's are pretty relaxed in the default mode to allow for better overclocking but tend to run a bit slower clock for clock than some other boards that are "pre-tweaked". I could not imagine DFI offering a "slower" BIOS.

Share this post


Link to post
Share on other sites

Ztech1, you may be right in that I may have a thermal issue. Last night I rebuilt some of the cooling components, placed an 80mm fan directly over the RAM and increased the speed on two of the fans (I had two of them running at only 5 volts- now back at 12 volts). I reset to 240 X 10 and was still only able to prime for 48 min before failure. My temps while priming dropped by almost 3 deg from 51 to 48 deg. but it appears to have made little additional difference. I will try to run a higher FSB with a lower multi instead, maybe this ram with the additional cooling will perform better- it should for $325! (Yah I know- what was I thinking? -at the time back in Sept that stuff (micron chips) was da Sh-- but not as good as what people seem to be using now (TCCD). I will report my results after some benching. Maybe I will hit the RAM with some more and take her on up towards 3.2V.

 

Thanks

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...