Prime95 version 29.2

[mackerel]

Originally Posted by Prime95

The option is only available (meaningful) if number of worker windows (in Test/Workers Windows) is set to more than one.

I only run Prime95 as a stress test or benchmark. With the move towards ever more cores/threads, stress test ends up popping up a load of windows, each of which too small to display anything useful. As mentioned previously I found I could work around it by editing prime.txt but it would be nice just to allow selection of this without limitation. At worst, if not meaningful it will do nothing right?

[Prime95]

Originally Posted by mackerel

I only run Prime95 as a stress test or benchmark. With the move towards ever more cores/threads, stress test ends up popping up a load of windows, each of which too small to display anything useful. As mentioned previously I found I could work around it by editing prime.txt but it would be nice just to allow selection of this without limitation. At worst, if not meaningful it will do nothing right?

Good point. I'll change it.

[thyw]

Is it intended for prime95 to write those gwnum.txt timings into the results.txt ?
On manual testing and reporting it could get in the way.
Also on my AMD Kaveri cpu (a10-7300) it detects that it has 2 MB L2, while every other program/website reports it has 2 x 2 MB L2.

[Prime95]

Originally Posted by thyw

Is it intended for prime95 to write those gwnum.txt timings into the results.txt ?
On manual testing and reporting it could get in the way.
Also on my AMD Kaveri cpu (a10-7300) it detects that it has 2 MB L2, while every other program/website reports it has 2 x 2 MB L2.

Manual results processor will ignore all those benchmark lines. I thought the user might find some of the data interesting.

Don't fret about the mis-identified L2 cache.

[S485122]

AFAIK Prime95 identifies L1 and L2 cache by core and L3 by CPU.

A question : there are new versions of Prime95 29.2 for Windows out, same file and program version but a bit bigger.What is the difference ? It must not affect the core of the program because the other 29.2 versions have not changed. (The previous 29.2 was dated 2017-06-23, the current one 2017-08-27.)

Jacob

[Prime95]

Originally Posted by S485122

A question : there are new versions of Prime95 29.2 for Windows out, same file and program version but a bit bigger.What is the difference ? It must not affect the core of the program because the other 29.2 versions have not changed. (The previous 29.2 was dated 2017-06-23, the current one 2017-08-27.)

see 3 and 4 in post #2 of this thread

[kladner]

I moved to 292-B4 this evening. I ran some quick torture tests, and then set it to the throughput bench with Benchmark all FFT implementations checked. Somewhere in the mid-to upper 3000s it crashed with the following error message:

Problem signature:
  Problem Event Name:    APPCRASH
  Application Name:    prime95.exe
  Application Version:    29.2.1.0
  Application Timestamp:    594c5cad
  Fault Module Name:    ntdll.dll
  Fault Module Version:    6.1.7601.23864
  Fault Module Timestamp:    595fa942
  Exception Code:    c0000005
  Exception Offset:    00000000000239bc
  OS Version:    6.1.7601.2.1.0.256.48
  Locale ID:    1033
  Additional Information 1:    7428
  Additional Information 2:    7428a25e70a5bdacbffc298dd7718274
  Additional Information 3:    abff
  Additional Information 4:    abff51d694d05bdb8440c6b914fa2ceb

This is an i7-6700k running at 4.3 GHz, which it has been happy at in the past. 16 GB RAM, Asus Z170 board

[Prime95]

can you email me the results.txt file?

[kladner]

Originally Posted by Prime95

can you email me the results.txt file?

Will do, via gmail.
I think this may have been an error on my part, involving saved BIOS settings.
I have subsequently completed the benchmark without error.

[vsuite]

Originally Posted by Prime95

1) The Windows version does not use new benchmark data in FFT selection until prime95 is restarted. Fixed in 29.2 build 3.
2) Throughput benchmarks using less than all cores improperly used all cores. Fixed in 29.2 build 4.
3) Using benchmark data on AMD Bulldozer with SUM(INPUTS) checking enabled can lead to errors initializing FFTs. Fixed in 29.2 build 5.
4) Auto benchmark runs even if user is not running any workers. Auto benchmark runs on battery power even if "run on battery power" option is not selected. Fixed in 29.2 build 5.

Has any functionality or operation changed between 29.2 build 4 and 5 for i3, i5 and i7 desktops already doing LL, DC, P-1 and TF please?

[Prime95]

Originally Posted by vsuite

Has any functionality or operation changed between 29.2 build 4 and 5 for i3, i5 and i7 desktops already doing LL, DC, P-1 and TF please?

No. Only two obscure bugs were fixed (see post #2 in this thread).

29.3 is coming soon.