Detailed client software thread...

[Prime95]

Version 22.8 is the latest beta. If all goes well it will become an official release. You can download it at:

Windows: ftp://mersenne.org/gimps/p95v228.zip
Linux: ftp://mersenne.org/gimps/mprime228.tgz
ftp://mersenne.org/gimps/sprime228.tgz
NT service: ftp://mersenne.org/gimps/winnt228.zip

New features (see whatsnew.txt for new features in versions 22.1 to 22.7):

1) Soft FFT crossovers have been implemented. If you test an exponent that is within 0.2% of the old hard FFT crossover point, then 1000 test iterations are run to determine if the smaller or larger FFT is appropriate for the exponent. Although not recommended, you can adjust prime95's behavior to be more aggressive using the SoftCrossover and SoftCrossoverAdjust features discussed in omnibet undoc.txt.
2) To better stress main memory, the torture test will now use up to the amount of memory specified in the Options/CPU dialog box.
3) Iterations with roundoff checking are a little faster for non-SSE2 CPUs.

[ebx]

Will there be a FreeBSD version any time soon? Thanks.

[Prime95]

Version 22.9 is the latest beta. If all goes well it will become an official release. You can download it at:

Windows: ftp://mersenne.org/gimps/p95v229.zip
Linux: ftp://mersenne.org/gimps/mprime229.tgz
or: ftp://mersenne.org/gimps/sprime229.tgz
NT service: ftp://mersenne.org/gimps/winnt229.zip

It fixes a hide icon bug and a proxy server password bug. Some setup bugs were also fixed, but set still needs more testing, so please use it and let me know what doesn't work. Two new features are PauseWhileRunning and Stop/Continue from the tray menu. Both new features were suggested in these forums!

[Prime95]

Version 22.10 is the latest beta. If no problems are reported it will become the official release. You can download it at:

Windows: ftp://mersenne.org/gimps/p95v2210.zip
Linux: ftp://mersenne.org/gimps/mprime2210.tgz
or: ftp://mersenne.org/gimps/sprime2210.tgz
NT service: ftp://mersenne.org/gimps/winnt2210.zip

It fixes several running as a service problems (multi-processor systems, NT4, no admin privileges)

Also, double-checking will now trial factor one-bit less than a first-time LL test. This is because finding a factor will save one LL test instead of two. P-1 factoring already took this into account.

[garo]

And it allows you to prevent Prime95 from unreserving work. See undoc.txt for details. The UnreserveDays feature is real nice.

[Prime95]

Version 22.10 is no longer beta. You can download it from http://www.mersenne.org/freesoft.htm

A FreeBSD port is included.

No changes have been made since the 22.10 beta a few weeks ago.

[Prime95]

A semi-serious bug in version 22.10 is fixed. The bug causes factors to be missed in the final stage of P-1 and ECM factoring. While this isn't the end of the world, it will cause you to run an unnecessary LL test.

Version 22.11 can be downloaded at http://www.mersenne.org/freesoft.htm

[Prime95]

Another semi-serious P-1 bug is fixed. From whatsnew.txt:

A bug was fixed that caused some factors to be missed in stage 2 of P-1 when the available memory did not let the program allocate 12 temporary variables. If testing a number in the 16 millions using an FFT size of 768K, then each temporary takes 768K * 8 bytes or 6MB. If your memory setting was less than 72MB (6MB * 12 temporaries) then you were affected by the bug. Actual the program allocates some fixed tables so anything less than about 75MB triggered the bug.

Version 22.12 can be downloaded at http://www.mersenne.org/freesoft.htm

A long overdue, more thorough QA of P-1 and ECM factoring is under way

Hello. I'm new here. I'm sorry if this is the wrong forum for this, but I have tried the "Torture Test" and my system fails on the first iteration, but it runs the "Benchmarks" correctly (and faster than the Athlon XP 1700+ Benchmark given on the website by 1-3%). This MB uses the NForce2 chipset which does a lot of new, cool (and fast) things with memory, but I am wondering if this is causing Prime95 to have problems? I would have thought it was a memory error, but I downloaded Docmem2.0 and ran 44 iterations overnight w/o a single failure.

Any help/comments would be appreciated. If I can get my system up and running with Prime95, I'd be happy to let it chew on primes when I'm not using it during the day.

Thanks.

BlazingDragon

[Prime95]

There is no error checking on the benchmarks. If you fail the first iteration of the torture test it is probably a memory problem, too aggressive overclock, or inadequate power supply problem. It is not a cooling issue.

Try using less aggressive memory settings (CAS latency and/or front side bus speed). Before joining any distributed computing project you should make sure you can pass at least an hour or two of torture testing.

[Xyzzy]

From the Mersenne mailing list...

I've uploaded a fix for an ECM bug. If you run ECM on a P4 you should upgrade to 22.13. All other users can continue to use 22.12.

V22.13 is at ftp://mersenne.org/gimps/p95v2213.zip and ftp://mersenne.org/gimps/p95v2213.exe

The ECM bug only occurred for exponents below 172,000 and near an FFT crossover point. It did not affect any Fermat factoring. Most Cunnningham factoring was unaffected because the FFT crossovers are at 743 and 1473. If you want to determine if some of your past ECM runs on a P4 were affected, then in v22.12 turn on round-off checking, and run an ECM curve on the exponent with B1=100. If the round-off error is 0.5 then your past ECM runs were affected by the bug.

The good news is v22.13 does have a P-1 and ECM QA suite implemented. There is a pretty good chance that the last serious bug in this area of the program has been found. I've run a small QA file successfully. Over time the QA volunteers will undoubtedly improve the QA suite.