**- Official Asus P5K Thread -**

Hi,
Have now checked with GPU-Z and the suggestion from another thread about the PCI_E speed was right, it gets set to x1 instead of x16 at this FSB. I then reduced the FSB down to 385 before it would get set back to x16. I also dropped the multiplier just to make sure the CPU was kept very low and was not the cause of the problem.
While I was still at 460 FSB I tried a few other things with voltages ending up with the following:
FSB 460
Multi x 7
mem 920MHz
Vcore 1.45v
Northbridge Voltage 1.55v
FSB termination voltage 1.22v

Even with all this it would still not allow x16 PCI-E. Dropping the FSB down to 400 had no effect either.

So, what is the route cause of the system deciding that any FSB above 385 cannot use the PCI-E at x16?? Many many people are getting high FSB's and high clock rates, surely they are not also all reducing their PCI-E speeds, I assume they would have noticed that. There must be something else I am not doing that is stopping it.

Does anyone have any other suggestions?

Thanks again for help so far.
 
Hibbyman said:
Hi Guys

Just decided to give my present OC a more thorough test on Prime and it stays great throughtout small fft's yet starts giving me errors when I use blend after a couple of minutes. I would assume that it has a problem with my ram settings if that's the case, as it's tested more in Blend?

Settings on the ram during Blend are 5,5,5,15 2T, 2.1v. Memtest passed with no problems and all the other settings are as recommended in the overclock guide for 3.2Ghz (400x8). Any ideas?

New to OCing.

Thanks!


Antec 900
Ocz GameXstream 700w
Q6600 (oc 3.2Ghz so far)
Asus P5K-E (Bios 0602)
Corsair 2x1gb XMS6400c4
Leadtek OC 8800GTB 512Mb
Click to expand...

Not sure if anyone has responded to this yet. Nothing obvious appears wrong with your set up. The only two things I can suggest are:
- checking the memory divider in your BIOS or via CPU-Z (could be overclocking your RAM but unlikely unless you've manually set it);
- increasing your CPU voltage a little (though strange the small ffts test didn't fail).

Hope this helps!
 
theoldtannery said:
So, what is the route cause of the system deciding that any FSB above 385 cannot use the PCI-E at x16?? Many many people are getting high FSB's and high clock rates, surely they are not also all reducing their PCI-E speeds
Click to expand...

asusp5kchotspotsdv0.jpg


The issue you are describing has been around for a year or so and has effected most of the P965/P35 chipset users at some point. It happened to me once or twice when running higher than 400MHz-FSB.

I have a hunch that the cause is *HEAT* related, something to do with either the northbridge or southbridge getting to hot and fuzzing out. You would think it was more likely something to do with the northbridge as that handles graphics PCI-E x16 data but a lot of people who have this problem you are experiencing have pumped their SouthBridge voltages.

Have a check of the two chipset heatsinks and see how hot they are running, even with the voltages on manual/minimum they can get a bit *Toasty* and will probably need some additional airflow, if you are running the NB/SB volts on auto or have them manually pumped up high and the case doesn't have much airflow then I bet the NB/SB heatsinks are *SCORCHING* hot!

The Northbridge heatsink should be a little less hot because it has a heatpipe and it's near to the CPU cooler but the poor old little southbridge has just a weeny little copper plated heatsink.

Anyway thats my Theory and I'm not quite sure how an overheating Southbridge can make the mobo drop the PCI-E from x16 to x1 but since keeping both the NB & SB well cooled I have never had that problem since?

Good Luck! :)
 
theoldtannery said:
Hi,
Have now checked with GPU-Z and the suggestion from another thread about the PCI_E speed was right, it gets set to x1 instead of x16 at this FSB. I then reduced the FSB down to 385 before it would get set back to x16. I also dropped the multiplier just to make sure the CPU was kept very low and was not the cause of the problem.
While I was still at 460 FSB I tried a few other things with voltages ending up with the following:
FSB 460
Multi x 7
mem 920MHz
Vcore 1.45v
Northbridge Voltage 1.55v
FSB termination voltage 1.22v

Even with all this it would still not allow x16 PCI-E. Dropping the FSB down to 400 had no effect either.

So, what is the route cause of the system deciding that any FSB above 385 cannot use the PCI-E at x16?? Many many people are getting high FSB's and high clock rates, surely they are not also all reducing their PCI-E speeds, I assume they would have noticed that. There must be something else I am not doing that is stopping it.

Does anyone have any other suggestions?

Thanks again for help so far.
Click to expand...

Any luck?
Big.Waynes right, it’s a pretty common problem

I still feel it’s an issue with the PCI-E bus.

Try upping the bus speed from 100mhz. try 105, 110, 115 and maybe a few in between to see if it lets you boot in with 16x

Are you running the latest bios? that to has helped people with this issue.
 
What are these boards like with all 4 ram slots populated?

Looking to run another pair of ballistix at 900Mhz on my P5K Premium and keep my current clock on my quad, recon this is feasible :confused:

Cheers :)
 
Sorry got half way through this thread then lost the will to live.

What is the best / latest bios for the standard P5K I will be running a Q6600
 
Big.Wayne said:
asusp5kchotspotsdv0.jpg


The issue you are describing has been around for a year or so and has effected most of the P965/P35 chipset users at some point. It happened to me once or twice when running higher than 400MHz-FSB.

I have a hunch that the cause is *HEAT* related, something to do with either the northbridge or southbridge getting to hot and fuzzing out. You would think it was more likely something to do with the northbridge as that handles graphics PCI-E x16 data but a lot of people who have this problem you are experiencing have pumped their SouthBridge voltages.

Have a check of the two chipset heatsinks and see how hot they are running, even with the voltages on manual/minimum they can get a bit *Toasty* and will probably need some additional airflow, if you are running the NB/SB volts on auto or have them manually pumped up high and the case doesn't have much airflow then I bet the NB/SB heatsinks are *SCORCHING* hot!

The Northbridge heatsink should be a little less hot because it has a heatpipe and it's near to the CPU cooler but the poor old little southbridge has just a weeny little copper plated heatsink.

Anyway thats my Theory and I'm not quite sure how an overheating Southbridge can make the mobo drop the PCI-E from x16 to x1 but since keeping both the NB & SB well cooled I have never had that problem since?

Good Luck! :)
Click to expand...


Thanks Big.wayne and Hesky82 for the input. I have a couple of interesting results to mention after last night.

I had left my system at FSB 385 to ensure I was still getting the x16. I also had the PCI-E set at 100. I recalled that a recomendation from an earlier post had said to try the PCI-E at 100/101. I ignored the 101 bit thinking that it was more a metter of making sure it was at least 100. However, i decided to retry the FSB at 400 and also set the PCI-E to 101. Hey presto, it worked at x16!!. I then set it back down to 100 again and it still stayed at x16!! So I upped the FSB all the way to 460, set NB to 1.4v and I still got x16. I was then thinking that this was random and I was just lucky at that stage. I decided to take advantage of it working to run the 3dmark06 tests and got 14200 for stock GPU settings and 15574 for GPU overclocked to 794 and 1074. Not bad I thought. Later in the eveining after a reboot or 2 the PCI-E again reverted to x1 so I immediately set it back up to 101 and it switched to x16. I have now left it at 101 and so far it is holding up ok.

I do not really know whats going on here but I can envisage a case where the system needs to know that the PCI-E is running at at least 100MHz for it to declare x16 available but with all the various multipliers/dividers and other system noise and heat related timing variances it maybe only detects its speed as a fraction below this value and hence goes into x1 mode. By setting the speed to 101, this ensures it is always seen as at least 100 and so works at x16. This may be a totally crap theory but my tests seem to indicate that moving to 101 has allowed it to work consistently. I will keep on monitoring this to see how it goes.
I also ran orthos all night again with no errors

I take the point about heat on NB/SB and will check this out some more.

Just to sumarise my settings are now:

FSB 460
Mult x7
CPU 3.22GHz
PCI_E 101MHz
Vcore 1.4v
DRAM 1:1 920MHz
DRAM volts 2.2v
Timings 5.0-5-5-16
Transaction booster disabled
NB Auto

CORETEMP's about 67 on full oad
ASUS Probe Temps 49 on full load
GPU temp with overclock 55 full load

Thanks again, comments welcome

Phil
 
Last edited:
Minor Problem on Boot Up - XP Pro or Vista Premium

This is not a major issue but as the M/C is a new build I would like to sort it out.

I am currently running dual boot OSs with Vista Premium and XP Pro. Probably once every 7 or 8 boots the system hangs on the Windows boot screen (which appears very faint) for about 45secs before continuing to boot normally. This occurs on Vista or XP Pro and from when the M/C is cold or has been running. Once into the respective OS the M/C runs like a dream.

The system is:
Mobo: Asus P5KC (BIOS Updated 28th Jan 08)
Processor: Intel Core 2 Quad Q6600 (Running at 9 x 333, CPU Voltage 1.3125)
Temperatures on full load typically below 52 degrees Celcius
HDD: Samsung 500Gb (Sata)
DVD Drive: Samsung
Memory: 4 x 1Gb Corsair XMS2 DDR2 PC6400 (800Mhz) 5-5-5-12 1.9v
(All memory specs manually input into BIOS)
Power Supply: Corsair HX 520w
Graphics Card: Sapphire HD3850 PCIE ATI 256Mb
 
theoldtannery said:
Thanks Big.wayne and Hesky82 for the input. I have a couple of interesting results to mention after last night.

I had left my system at FSB 385 to ensure I was still getting the x16. I also had the PCI-E set at 100. I recalled that a recomendation from an earlier post had said to try the PCI-E at 100/101. I ignored the 101 bit thinking that it was more a metter of making sure it was at least 100. However, i decided to retry the FSB at 400 and also set the PCI-E to 101. Hey presto, it worked at x16!!. I then set it back down to 100 again and it still stayed at x16!! So I upped the FSB all the way to 460, set NB to 1.4v and I still got x16. I was then thinking that this was random and I was just lucky at that stage. I decided to take advantage of it working to run the 3dmark06 tests and got 14200 for stock GPU settings and 15574 for GPU overclocked to 794 and 1074. Not bad I thought. Later in the eveining after a reboot or 2 the PCI-E again reverted to x1 so I immediately set it back up to 101 and it switched to x16. I have now left it at 101 and so far it is holding up ok.

I do not really know whats going on here but I can envisage a case where the system needs to know that the PCI-E is running at at least 100MHz for it to declare x16 available but with all the various multipliers/dividers and other system noise and heat related timing variances it maybe only detects its speed as a fraction below this value and hence goes into x1 mode. By setting the speed to 101, this ensures it is always seen as at least 100 and so works at x16. This may be a totally crap theory but my tests seem to indicate that moving to 101 has allowed it to work consistently. I will keep on monitoring this to see how it goes.
I also ran orthos all night again with no errors

I take the point about heat on NB/SB and will check this out some more.

Just to sumarise my settings are now:

FSB 460
Mult x7
CPU 3.22GHz
PCI_E 101MHz
Vcore 1.4v
DRAM 1:1 920MHz
DRAM volts 2.2v
Timings 5.0-5-5-16
Transaction booster disabled
NB Auto

CORETEMP's about 67 on full oad
ASUS Probe Temps 49 on full load
GPU temp with overclock 55 full load

Thanks again, comments welcome

Phil
Click to expand...

Sweet, I had a feeling it was an issue with the PCI-E bus frequency.
Glad you have sorted it :)
 
yup you will find that as you increase FSB past 375ish you will have to start raising PCI-E frequency to stop it sometimes running at *1 rather than *16.. I have found that 105 works all the way up to around 525mhz
 
A vanilla P5K is anything but basic. Got an e2180 to 3GHz easily with a P5K. The only difference between the vanilla board and the more expensive versions are bells and wistles like RAID and firewire, and heatpipe cooling that will only come into its own at 450FSB+
 
You must log in or register to reply here.
Back
Top Bottom