Some things

Hello everyone, i'm into the GIMPS for 4 months now. Untill i did the update (v27.7 build 2)
-and afterwards i was experimenting with the settings (like the number of workers
etc) everything was fine :)

Suddenly these roundoff errors came onto the screen. What should i do?
I already searched much info but this goes beyond my limits.
Torture test has been running for 9 hours now but didn't found anything.

worker#1

[June 4 22:43] Torture Test completed 328 tests in 9 hours, 7 minutes – 0 errors, 0 warnings.
Worker stopped.
Worker starting
Setting affinity to run worker on any logical CPU.
Resuming primality test of M57633361 using AVX FFT length 3M, Pass1=512, Pass2=6K
Iteration …/… [48.35%]
Possible hardware errors have occurred during the test!
14 ROUNDOFF > 0.4 of which 2 were repeatable (not hardware errors).
Confidence in final result is very poor.

Worker #3
//same text but this is different:
30 ROUNDOFF > 0.4 of which 6 were repeatable (not hardware errors).

[Dubslow]

Si vous voulez, vous pouvez utiliser le français; quelques personnes parmi nous peuvent comprendre assez pour vous aider.

Avez-vous vérifié la température de votre CPU? C'est le plus commun raison d'avoir erreurs.

For the rest of you: Have you checked your CPU temperatures? That's the most common cause of errors.

Oh i'm dutch, i just wanted to type in 'une personne' ;)

Temperatures (with Core Temp and CPUID HWMonitor)
around 83-86°C when I was doing the torture test.
Tjmax= 100°C

I have an HP dv6
processor: Intel(R) Core(TM) i7-2630QM @ 2.00GHz
64 bit

I receive the message directly after i started Prime95.

My benchmarks are even faster than http://mersenne.org/report_benchmarks/
-> I guess thats the cause of the new version?

[Dubslow]

Originally Posted by Unregistered

Oh i'm dutch, i just wanted to type in 'une personne' ;)

Temperatures (with Core Temp and CPUID HWMonitor)
around 83-86°C when I was doing the torture test.
Tjmax= 100°C

I have an HP dv6
processor: Intel(R) Core(TM) i7-2630QM @ 2.00GHz
64 bit

I receive the message directly after i started Prime95.

My benchmarks are even faster than http://mersenne.org/report_benchmarks/
-> I guess thats the cause of the new version?

Mijn Opa is Dutch, but I don't speak an ounce of Nederlands (That's not quite true: Een, twei, drie...)
(As a matter of fact we call him Opa, not grandpa :) )

Anywho, those temps are pretty high bordering on error-prone. Make sure your ventilation fans are pointing out into free space so that heat doesn't get trapped. Edit: Yes, for Sandy Bridge CPUs such as yours, v27 will see a ~25% speedup over v26.

[cheesehead]

Originally Posted by Unregistered

Temperatures (with Core Temp and CPUID HWMonitor)
around 83-86°C when I was doing the torture test.
Tjmax= 100°C

Definitely try to get those temperatures down. Even if that heat didn't cause immediate errors, it would shorten the useful life of your circuitry.

(The manufacturer might say it's safe, but how long does their replacement warranty last?)

[moebius]

I suspect, this error occurs only with one specific exponent! The temperature is probably too high, maybe you just use a better CPU cooler like Nocturna or somethimg else.

mooie groet

[davieddy]

Originally Posted by cheesehead

Definitely try to get those temperatures down. Even if that heat didn't cause immediate errors, it would shorten the useful life of your circuitry.

(The manufacturer might say it's safe, but how long does their replacement warranty last?)

How many times do I have to warn you Richard?
Stop encouraging me.
Cool Jerk

D

Sorry for the late answer, i just wanted to let you know that i have exams. I will reply soon.

[kracker]

I usually keep mine at 50C, I never let it go above 65-70C.