Prime95 problem with Intel Tiger Lake processor?
 

I have purchased a HP Pavilion 15-eg0901nc laptop with Intel Core i5 1135G7 Tiger Lake processor, 16 GB RAM, Windows 10 (20H2), no overclock.

Everything works OK, except for Prime95 stress test, which ends up rather quickly. After one minute of running Small FFTs (8 threads), all 8 workers stop with the same error:
[QUOTE][Wed Apr 28 12:08:48 2021]
FATAL ERROR: Final result was 00000000, expected: 7B7030D4.
Hardware failure detected running 120K FFT size, consult stress.txt file.[/QUOTE]
Other runs ended up similarly, with different expected values, but always the same in all workers, and always ending up with the result 00000000, e.g.
[QUOTE]FATAL ERROR: Final result was 00000000, expected: F6376191.
Hardware failure detected running 256K FFT size, consult stress.txt file.[/QUOTE]
[QUOTE]FATAL ERROR: Final result was 00000000, expected: BA409794.
Hardware failure detected running 280K FFT size, consult stress.txt file.[/QUOTE]
I tried disabling AVX-512 with no effect.

I tried AIDA64, IntelBurnTest, HP HW diagnostics and MemTest86, all finished successfully.

Could this be an issue with Prime95 with the newish Intel processor, or is this really a HW issue, which manifests only with Prime95 and no other workflows/stress tests?

[QUOTE=jakubklimek;577095]Could this be an issue with Prime95 with the newish Intel processor, or is this really a HW issue, which manifests only with Prime95 and no other workflows/stress tests?[/QUOTE]It [i]could[/i] be an issue with P95, but probably not.

I would suspect either your RAM or your CPU or both is/are being clocked too fast. Slow it down.

Which exact version of Prime95 have you used? The current one from the website? Some version from this forum? An older one?

[QUOTE=kruoli;577098]Which exact version of Prime95 have you used? The current one from the website? Some version from this forum? An older one?[/QUOTE]

It was v30.6b3 from [url]https://www.techpowerup.com/download/prime95/[/url]

Indeed, when I retested with v30.3b6 from this forum, it worked ok with no errors.

Thanks @kruoli ! And, therefore, may this serve as a bug report for v30.6b3 I guess.

Hi,

I have the exact same issue (same error, memtest86 OK over night) but with a different CPU/OS : Xeon Gold 6240 (x2) on CentOS 8

Fails repeatedly regardless of Prime95 parameters (any FFT size, duration, max memory, etc...) with version : p95v306b3

[C]FATAL ERROR: Final result was 00000000, expected: 7B7030D4.
Hardware failure detected running 120K FFT size, consult stress.txt file.
Torture Test completed 1 tests in 4 minutes - 1 errors, 0 warnings.
[/C]

If you use the version from the official website ([URL="http://www.mersenne.org/download/"]here[/URL]), the error will not occur.

@George:
Do we have anything we could use as a lever against TechPowerUp displaying our beta/forum versions as if they are our latest stable release?

[QUOTE=kruoli;577111]If you use the version from the official website ([URL="http://www.mersenne.org/download/"]here[/URL]), the error will not occur.
[/QUOTE]

Honnestly I don't remember where I downloaded it but I can't from your link, the page does not show due to a TLS error

[C]Secure Connection Failed
An error occurred during a connection to www.mersenne.org. PR_END_OF_FILE_ERROR
[/C]
Version "p95v305b2.linux64.tar.gz" from the forum works like a charm though :smile:

Will be fixed in 30.6 build 4. I should have it uploaded later today. Thanks for reporting the problem.