Overclocking Problems

[lk_hltn]

Hi all

 

I've had my Lanparty UT NF4 SLI-DR Expert up and running now for a day. It was folding all last night with no problems and it has passed prime95 and memtest.

 

Today I tried to push it a bit harder. I pushed up the FSB gradually, keeping the RAM divider set to keep the memory below 250MHz (because I knew that this was stable). It ran prime95 at 270MHz without errors but the problem is that it sometimes hangs and restarts itself before the "Verifying DMI pool data..." message. I have tried adding more voltage to the CPU, LTD and Chipset with no success.

 

It can't be the RAM because it is running slower than the 250MHz that I know works. It can't be lack of voltage because I have tried adding more, and the temps are fine so it isn't that. The LTD/FSB multiplier was kept at 3x.

 

Also what is the difference between the Startup VID and the CPU VID?

 

Any ideas welcome, thanks in advance.

 

lk_hltn

[THEnumber337]

Hi all

 

I've had my Lanparty UT NF4 SLI-DR Expert up and running now for a day. It was folding all last night with no problems and it has passed prime95 and memtest.

 

Today I tried to push it a bit harder. I pushed up the FSB gradually, keeping the RAM divider set to keep the memory below 250MHz (because I knew that this was stable). It ran prime95 at 270MHz without errors but the problem is that it sometimes hangs and restarts itself before the "Verifying DMI pool data..." message. I have tried adding more voltage to the CPU, LTD and Chipset with no success.

 

It can't be the RAM because it is running slower than the 250MHz that I know works. It can't be lack of voltage because I have tried adding more, and the temps are fine so it isn't that. The LTD/FSB multiplier was kept at 3x.

 

Also what is the difference between the Startup VID and the CPU VID?

 

Any ideas welcome, thanks in advance.

 

lk_hltn

 

Take a look at the overclocking guide found here.

 

Also, some more information such as voltages your using and RAM timings would be helpful.

[Praz]

Some of the secondary timings will have to be readjusted as you raise the FSB.

[lk_hltn]

Thanks for the replies,

 

I've had a read of the overclocking guide and related articles - very useful.

 

Tomorrow I'll attempt to find the max CPU speed and report my results back. If I can reach a high clock then I assume we can narrow the problem I have been having today down to the memory.

 

I have read about the different BIOS versions. In the "DEFINITIVE DFI A64 OVERCLOCKING GUIDE" it says that -2 is the best all rounder and I have read here that -3 is optimised for TCCD/Micron (which I think my memory chips are) - which BIOS should I be using for my Crucial Ballistix Tracer 512mb modules and where can I download it from? Is there a guide to updating the BIOS - I'd like to make sure I'm doing that correctly?

 

Some of the secondary timings will have to be readjusted as you raise the FSB.

 

Is this the case even if I am using a memory divider and my memory is under-clocked?

 

Thanks for your help again

 

lk_hltn

[Praz]

Is this the case even if I am using a memory divider and my memory is under-clocked?

Yes. Two of the most important ones that are affected by an increase in the FSB is Max Async Latency and Read Preamble Time.

[lk_hltn]

I am now running at 2.7GHz (300MHz*9), 1.66V and it will complete the superpi 32M test without errors.

 

If I try to run it with a HTT of 305MHz it will not complete the superpi test, even if I apply more voltage (~1.7V). I would like to try and push it a bit further as my temps are still good (38C).

 

I have tried lowering the multiplier to 8x and increasing the HTT to 320MHz and it completes the superpi 32M test fine.

 

I tried running the OCCT test with the 300*9, 1.66V overclock and it fails after 2 mins, I increased the voltage and it failed again after 2 mins.

 

I thought that it might be the psu that was causing me problems because speedfan read the 12V rail as ~11.84V so I monitored the 12V rail with a Digital MM (from a molex connector) but it doesn't vary too much - it varies between 11.98V and 12.02V.

 

RAM divider is 2:1, HTT multiplier is at 2x.

 

Anyone have any ideas on how I can further increase my clocks (or get 2.7GHz stable)?

 

lk_hltn

[marcopolo]

I have exactly the same problem as you. I know my sig says i'm at 2700 but i've backed it down to 2600 because i had the same poblem as you.

 

To this date there is no solution to this problem and no-body knows why it does this. Here or at xtreme forums or at DFI support.

 

I passed every F***i*g bench test out there, once into windows and every game imaginable. My PC was essentially 100% load stable BUT.... just occasionally at boot up "Verifying DMI pool data..." message it would do the same as yours.

 

I do not believe it is a stability issue as once into windows everything works fine. I would love to know the exact code that is getting executed and the order the the BIOS enables all of the devices but so far no one seems to have this information: it is doing something different at boot up and I would love to know what aswell so I could fix it.

 

I realise Praz that you have a serious amount of experience on PC's but if the secondary timings were an issue at high HTT/FSB then just one of the multitude of torture tests i've run would show up the problem............not at boot up though, surely???

 

I have one suspicion I would like to hear some comments on: As the PC is just starting to boot from the HD at this point what if the SATA controller is not up to the job at high HTT/FSB - this might cause the exhibited results. Stab in the dark.

 

Marco.

[HITandRUN]

I would also check out this and look at your PSU! What are the exact specs of it? :confused:

[marcopolo]

In my case It's nothing to do with the PSU.

 

Torture tests tax the PSU more than anything at boot up. Iv'e used one of those watt meters to read power consumed at various stages.

 

lk_hltn's PSU is actually below recomened watts (480W recomended) , but as you say he will need to check out current ratings on the rails.

[lk_hltn]

Thanks

 

I think praz is correct - when I was fiddling around with the Max Async Latency and Read Preamble Time, if I had them too low, then it wouldn't boot, but if I increased them - it booted fine.

 

All I can say is give it a go!

 

I think my psu is now limiting my clocks - I'm at 290x9 and I can get any higher stable - even with more voltage applied. My psu is connected with a 24-pin connector - no adaptors and it also has an 8-pin connector. I see it is not quite powerfull enough though. Will it damage the components if it isn't powerful enough? Could you recommend a psu for less than £50 ($100).

[Praz]

I passed every F***i*g bench test out there, once into windows and every game imaginable. My PC was essentially 100% load stable BUT.... just occasionally at boot up "Verifying DMI pool data..." message it would do the same as yours.

 

I do not believe it is a stability issue as once into windows everything works fine.

 

but if the secondary timings were an issue at high HTT/FSB then just one of the multitude of torture tests i've run would show up the problem............not at boot up though, surely???

 

This is a classic symptom of improper timings. That's why world records are set using programs to set speeds and timings from within Windows. At those settings the system would never boot.

 

In my case It's nothing to do with the PSU.

 

Torture tests tax the PSU more than anything at boot up. Iv'e used one of those watt meters to read power consumed at various stages.

I don't believe a consumer grade watt meter exists that can react fast enough to display the maximum inrush current at the time the system is powered up.

 

Stress testing a system puts the greatest load on it over a prolonged period of time but maximum load, although brief, occurs at system startup.

[marcopolo]

You are right Praz; the watt-meter either seems to average over a few seconds or only reads every few seconds, so as you say it won't pick up transient spikes.

 

When I was setting my secondary timings I payed particular attention to MAL and RP. I had to get these spot on to even be able to get through memtest!

See my sig! Mal 8.5 or 7.5 was no good! RP 3.5 or 4.5 was no good! It had to be exactly 8.0 and 4.0 to get past memtest to pass with Corsair on 140 divider.

 

I will give up soon anyway as I have some 2GB sticks of UCCC or TCCC coming so I can clock higher on the 150 divider which never gives me any problems ever!

 

Good luck lk_hltn.

 

Marco.