I am currently running a Prime95 (please help)

Brandon_140

Distinguished
Nov 30, 2016
89
1
18,635
I am currently running a prime95 torture test, CPU-Z says the core speed is correct about 4.3ghz and my cpu's average temperature is 51 degrees celceius. Worker 4 has stopped working about 10 minutes iin said that a fatal hardware error occured. 1 error 0 warnings. then 17 minutes in the same happened to worker 6. Does this mean that my overclock settings are no good? The its not going over 63 degrees which is the cpu max temp. Does this mean i should up my voltage abit and try again? the current settings are 21.5x cpu core multiplyer, 1.4v cpu, and 1.2375 NB. worker logs for not running workers:
worker 4:
[Feb 22 01:55] Worker starting
[Feb 22 01:55] Setting affinity to run worker on logical CPU #4
[Feb 22 01:55] Beginning a continuous self-test on your computer.
[Feb 22 01:55] Please read stress.txt. Choose Test/Stop to end this test.
[Feb 22 01:55] Test 1, 720000 Lucas-Lehmer iterations of M580673 using AMD K10 type-1 FFT length 28K, Pass1=112, Pass2=256.
[Feb 22 02:00] Test 2, 720000 Lucas-Lehmer iterations of M573441 using AMD K10 type-1 FFT length 28K, Pass1=112, Pass2=256.
[Feb 22 02:01] FATAL ERROR: Resulting sum was 7214653820511398, expected: 7214748778085717
[Feb 22 02:01] Hardware failure detected, consult stress.txt file.
[Feb 22 02:01] Torture Test completed 1 tests in 5 minutes - 1 errors, 0 warnings.
[Feb 22 02:01] Worker stopped.
worker 6:
[Feb 22 01:55] Worker starting
[Feb 22 01:55] Setting affinity to run worker on logical CPU #6
[Feb 22 01:55] Beginning a continuous self-test on your computer.
[Feb 22 01:55] Please read stress.txt. Choose Test/Stop to end this test.
[Feb 22 01:55] Test 1, 720000 Lucas-Lehmer iterations of M580673 using AMD K10 type-1 FFT length 28K, Pass1=112, Pass2=256.
[Feb 22 02:00] Test 2, 720000 Lucas-Lehmer iterations of M573441 using AMD K10 type-1 FFT length 28K, Pass1=112, Pass2=256.
[Feb 22 02:05] Self-test 28K passed!
[Feb 22 02:05] Test 1, 3200000 Lucas-Lehmer iterations of M172031 using AMD K10 type-1 FFT length 8K, Pass1=32, Pass2=256.
[Feb 22 02:10] Self-test 8K passed!
[Feb 22 02:10] Test 1, 480000 Lucas-Lehmer iterations of M827279 using AMD K10 type-1 FFT length 40K, Pass1=32, Pass2=1280.
[Feb 22 02:13] FATAL ERROR: Rounding was 0.5, expected less than 0.4
[Feb 22 02:13] Hardware failure detected, consult stress.txt file.
[Feb 22 02:13] Torture Test completed 3 tests in 17 minutes - 1 errors, 0 warnings.
[Feb 22 02:13] Worker stopped.
 
Solution
^ everything bgunner says is true , the pallas knocks close to 20c off my vrm temps .

You should probably dial back to 4.2ghz & get the voltage as low as possible under load.
The vrm's will cope with that with no cooling , only 4 phase but better quality than you'd expect.

I cannot believe you do not get any core speed throttling under prime though , because the board has insanely aggressive power management which needs to be disabled.

You can try settings , worst case they don't work - I did forget to tell you to disable turbocore aswell (a necessity)
You shouldn't touch any power saviing features in bios , it can be done via software at low level which just works better & doesn't affect idle temps/voltages...
IF a worker stopped it means that that core is unstable. This could be a lack of voltage, you have reached the maximum overclock of the CPU or maybe the ram is unstable. This last one comes into play if you have adjusted the Base clock.

EDIT: WHen asking for advice on PC parts please include the full system specs. What can be deduced = you have an AMD CPU. I know this because of the voltage. If you throw 1.4V at an Intel CPU it will fry. Other than this we have no clue what you are using for hardware which can make any advice you get wrong.

Also AMD's FX CPU's like the CPU/NB voltage to be around 1.35V where as there Phenom II's liked 1.25V. This is actually the memory controller voltage. The north bridge usually sits between 1.2 and 1.25V which is different than the CPU/NB voltage.
 


The processor is a AMD FX 6300, the MB is a ASUS M5A78L-M/USB3 and cooler is a Coolermaster Hyper 212 LED (I have 2 120mm fans hooked up to it) and i have run a memtest before and found out there was no errors with my ram. so what should my next step be? should i up the voltage by a couple and try it again? is there better overclock settings you could recoomend?
 


The processor is a AMD FX 6300, the MB is a ASUS M5A78L-M/USB3 and cooler is a Coolermaster Hyper 212 LED (I have 2 120mm fans hooked up to it) and i have run a memtest before and found out there was no errors with my ram. so what should my next step be? should i up the voltage by a couple and try it again? is there better overclock settings you could recoomend?
 
Wow it just amazes me that this motherboard keeps getting used for overclocking. This motherboard is not meant to overclock because it has no VRM cooling. This causes the motherboard to throttle the core speeds due to the VRM overheating. The only cure for this is having a heatsink on the VRM. Adding aftermarket cooling here just does not do the trick.

4.3GHz (21.5 * 200) will need around 1.43V or so to be stable. This is not set in stone due to the differences in chips but a good place to start.
 


What about NB voltage should i just leave that at 1.2375, and what about hyperlink transport speed is 2400 good? i am really novice with this so bear with me xD and my MB temp only went up 1 degree during the stress test if that means anything
 
I'm running 4.3ghz on the same board at 1.36v .
Stable as a rock , I do have a raijintek pallas on there though which cools vrm's very very well.

Brandon - under a prime test what's your CPU voltage showing as in cpu-z??
Board has severe v droop under load , it needs to have llc adjusted .
I wouldn't want to be running 1.4v on this board at all really & especially not with no vrm cooling.
Take that as a warning , as bgunner states this is not an overclocking board by any means , although its better than the specs would suggest it has limits.
 


Okay so what kind of VRM cooling could i purchase? and could you share with me your overclock settings? i am using overdrive... and it was at 1.38volts during the prime test

 
You don't want to touch any bios settings apart from CPU voltage , multiplier & llc.
Just leave everything else well along.

You should have both amd overdrive & CPU z up & running while stress testing.
Overdrive for clock speeds & thermal margin , CPUz for the load voltage.

My overclock settings ?? Bear in mind I know I have a decent chip here which hit 4.8ghz on a different board.

CPU voltage - 1.3625
Llc - 33%
Multiplier 21.5


You need to run overdrive aswell , you may find cores throttling under load due to the boards power management.

Vrm cooling ?? Nothing better than direct airflow , stick on vrm sinks are fairly worthless - the heat transferral through thermal pads is not that great.
 
MB temp is nothing really since it is usually located near the bottom of the motherboard.

The North Bridge voltage (NB voltage) should be roughly where you say it is. This should not need to be changed.

Make sure the CPU/NB voltage is between 1.35V and 1.42V. After the overclocking session is done you can try to lower this to as close to 1.35 as it will allow to lower the temps. This is the memory controller located in the CPU. AMD defaults this to around 1.45V which is much higher than needed in most cases.

the HT link or HyperTransport Link is the link between the CPU and north bridge. The speed of this will depend on the motherboard. According to Asus it supposed to support up to 2600 MHz. If you can get it up to this it would be best but 2400 MHz will do if you can't.
 


What is Llc? where do i find that in amd overdrive? and i do have 5 case fans installed, i have a deepcool case that came with 5 pre installed fans. i have decent ventilation. when never gets any warm air being blown out the case.

 




Bare in mind, Brandon, that Mdmatt has a CPU cooler that blows downward on to the VRM actively cooling them where as you have a tower cooler that can actually starve the VRM of air flow. His settings may not work for you plus every piece of hardware will OC differently and could need different settings. You can use his settings as a starting point but be ready to change them.

Overdrive should not be used to overclock. The reason for this is it adds in another place for instability. Adjust your clocks from inside of the BIOS. This will remove the program from causing instability. Just use overdrive for the thermal margin ans clock speeds as madmatt suggested

As for VRM cooling, as I already mentioned there is not a good solution for that motherboard. It is not designed for overclocking.
 


So should i go to my bios and turn off all power saving features and just change my cpu voltage to 1.36, llc to 33% and core multiplier to 21.5 and leave everything else auto? I know it may not work with my system but its worth a shot. xD
 
^ everything bgunner says is true , the pallas knocks close to 20c off my vrm temps .

You should probably dial back to 4.2ghz & get the voltage as low as possible under load.
The vrm's will cope with that with no cooling , only 4 phase but better quality than you'd expect.

I cannot believe you do not get any core speed throttling under prime though , because the board has insanely aggressive power management which needs to be disabled.

You can try settings , worst case they don't work - I did forget to tell you to disable turbocore aswell (a necessity)
You shouldn't touch any power saviing features in bios , it can be done via software at low level which just works better & doesn't affect idle temps/voltages.

https://drive.google.com/file/d/0B26I8uSTEOTseFpuS24wWGtIOHM/view?usp=drivesdk

Download & unzip that , double clicking the off.bat file should disble clock throttling at heavy load - it's a singular instance so has to be rerun on every cold boot.
 
Solution


Also turn off Turbo Core and enter CPU configuration and turn off APM. These will help dramatically in overclocking.