prime stable?

DJ_Hughesy

DJ_Hughesy

DJ_Hughesy

Associate
Joined
8 Jan 2011
Posts
373
Location
Barrow in Furness
jus wonderin what the average time for a test is in prime95 before youd call id stable? some say few hours some day 24..

i need a realistic time (cant leave pc on 24 hours)

how long should i test for?
 
As above - I leave Prime running overnight and do about 20 cycles of IBT. You may want to run memtest for a few hours too if it's totally new system.

There's no danger leaving prime on overnight - it will either pass or stop working if a core fails. If things get too hot it should reboot (but you could test for possible heat issues with IBT or running prime for an hour.).
 
You've passed but the rather large variation in GFlops suggests instability.

IBT is a good quick benchmark for stability if you tweak the overclock but for the final test, you need to do prime95 for hours, really.

10 runs of IBT was never enough for me, try 20-25, also at maximum stress level.

2148 MB Ram is really low, I remember forcing around 3GB to get a proper test.
 
If you are not on the bleeding edge of overclock, I would say most of the above is excessive. Look at your temps after an hour or two of testing IBT and / or prime. Your room is probably quite warm now. Give it another hour and if no significant changes, you are probably stable. You will not usually be stressing your computer for three hours to that extent doing anything normal, even crysis. ;)
 
aparently very high uses moer ram that maximum? :S lol retest with 20 runs using 4gig :)

^^ woops cant.. stupid igb... wish my gpu didnt die lol
well max is best i can do...so again
----------------------------
IntelBurnTest v2.51
Created by AgentGOD
----------------------------

Processor: AMD Phenom(tm) II X4 955 Processor
Clock Speed: 4.02 GHz
Active Physical Cores: 4
Total System Memory: 3581 MB

Stress Level: Maximum (2148 MB)
Testing started on 16/05/2011 16:49:25
Time (s) Speed (GFlops) Result
[16:50:55] 57.072 49.8565 3.153118e-002
[16:52:24] 56.207 50.6234 3.153118e-002
[16:53:53] 57.606 49.3937 3.153118e-002
[16:55:21] 56.099 50.7209 3.153118e-002
[16:56:49] 56.250 50.5847 3.153118e-002
[16:58:17] 56.414 50.4376 3.153118e-002
[16:59:46] 57.119 49.8152 3.153118e-002
[17:01:15] 57.335 49.6272 3.153118e-002
[17:02:44] 56.328 50.5146 3.153118e-002
[17:04:12] 56.349 50.4957 3.153118e-002
[17:05:42] 57.961 49.0910 3.153118e-002
[17:07:10] 56.396 50.4539 3.153118e-002
[17:08:39] 57.213 49.7330 3.153118e-002
[17:10:08] 56.473 50.3851 3.153118e-002
[17:11:36] 56.414 50.4380 3.153118e-002
[17:13:04] 56.392 50.4571 3.153118e-002
[17:14:33] 57.095 49.8357 3.153118e-002
[17:16:01] 56.348 50.4968 3.153118e-002
[17:17:31] 57.928 49.1194 3.153118e-002
[17:18:59] 56.458 50.3982 3.153118e-002
Testing ended on 16/05/2011 17:19:00
Test Result: Success.
----------------------------

reran tests with same settings just 20 runs this time.. temps same as earlier
gunna prime for 7-8 hours...

oh n @ nkata yes ino im not "on the bleeding edge" but best to learn how to test stability before attempting that isnt it? :D
 
Last edited:
DJ_Hughesy said:
oh n @ nkata yes ino im not "on the bleeding edge" but best to learn how to test stability before attempting that isnt it? :D
Click to expand...

Im sorry DJ, I was not picking on your overclock, what I was trying to say was extended testing is not always necessary. If you just use a PC for normal tasks and gaming, you are not likely to hit the extremes of heat found in IBT or prime and a couple of hours maximum will usually cause failure if you are near the bleeding edge.
Old skool overclocking of stepping up to the failure to boot or go into windows and then rowing back a notch or two will give a perfectly stable PC 99% of the time with no testing at all. Running prime for a couple or three hours will just confirm that.

Anyway you seem to have a nice 4Ghz overclock.

cheers, andy.
 
sorry think i might of sounded a bit offensive there and didnt mean to :)

what i ment was, its nice to learn how to test stability propperly incase at somepoint i wish to go to the bleeding edge as it seems fun pushing things further than there ment to

although i did leave prime95 running for around 6hours and came back to see these temps... surely heat cant of hit that level shown in max temp, or even max cpu volts as its set to manual voltages or is there a setting im supposed to turn off? lol



Uploaded with ImageShack.us
 
DJ_Hughesy said:
sorry think i might of sounded a bit offensive there and didnt mean to :)

what i ment was, its nice to learn how to test stability propperly incase at somepoint i wish to go to the bleeding edge as it seems fun pushing things further than there ment to

although i did leave prime95 running for around 6hours and came back to see these temps... surely heat cant of hit that level shown in max temp, or even max cpu volts as its set to manual voltages or is there a setting im supposed to turn off? lol
Click to expand...

Very strange, an AMD CPU would throttle, turn off or fail before 87C therefore the temperature should not be attainable. The core temps at 48C look realistic although the phenom does not record true core temps. My 1055T has a CPU temp about +10C above nominal reported core temp.

Likewise 3.06V vcore, should not happen. 1.45V max.

This is one reason I do not like unobserved prime testing.
Were all the cores still working when you turned it off?
Was the 87C and 3.06V the last reading before you stopped testing or was it at more normal readings?
 
na i started the test n went round a freinds, cam back 5+1/2 to 6 hours later n yes prime still running perfectly fine (not failed)

i stoped n closer prime then screenshotted... temps you see above are temps it was at when i returned...

considering i have my volts set to manual, surely it shouldnt jump up the vcore at all let alone to 3.0+ :S


id noticed this before while doin other stuff (bad company 2) after a while itl report max temp/vcore as something stupidly high but if i watch it ill never see the current readout hit those readings so its more than likely doing it for a second at a time or something?
 
DJ_Hughesy said:
its more than likely doing it for a second at a time or something?
Click to expand...

Milliseconds rather than a second, but it is a worrying spike. I have the same motherboard and have not seen this before. Also the voltage may spike like that but the temperature would not, it would take several seconds to heat up.

Could be an error in the sensors or software, can you log the results through AIDA64 against time?
 
nkata said:
Milliseconds rather than a second, but it is a worrying spike. I have the same motherboard and have not seen this before. Also the voltage may spike like that but the temperature would not, it would take several seconds to heat up.

Could be an error in the sensors or software, can you log the results through AIDA64 against time?
Click to expand...

i guess not, i dont think i have aida64 logging to a file... :o
 
What HT and NB are you using, I usually limit to standard clocks ie 2000Mhz. As you are overclocking on the multiplier I expect yours is at the default too.

CPU 20 x 200
Memory 8 x 200
HT 10 x 200
NB 10 x 200

I would try a lower multiplier 19 or 19.5 and rerun the tests.

Will it run at 4Ghz with voltages set at auto?
 
nkata said:
What HT and NB are you using, I usually limit to standard clocks ie 2000Mhz. As you are overclocking on the multiplier I expect yours is at the default too.

CPU 20 x 200
Memory 8 x 200
HT 10 x 200
NB 10 x 200

I would try a lower multiplier 19 or 19.5 and rerun the tests.

Will it run at 4Ghz with voltages set at auto?
Click to expand...

yea i jus upped multiplier... n would you trust it at 4ghs with auto volts?...
 
I have had mine on auto at 3.5GHz, I undervolt it by 0.075V in bios at 3.5GHz so am on manual currently.

My typical volts under load with prime is 1.28V confirmed by HWmonitor and CPUID. Should be up to 1.375V max from bios.

This motherboard is quite good with stable power and should keep it within the max vcore. having said that you are getting some peculiar results.
 
nkata said:
I have had mine on auto at 3.5GHz, I undervolt it by 0.075V in bios at 3.5GHz so am on manual currently.

My typical volts under load with prime is 1.28V confirmed by HWmonitor and CPUID. Should be up to 1.375V max from bios.

This motherboard is quite good with stable power and should keep it within the max vcore. having said that you are getting some peculiar results.
Click to expand...

yea seems like something isnt quiet right :(
any tests i could do to see what the problem is?
 
You must log in or register to reply here.
Back
Top Bottom