Prime95 version 25.9

[Prime95]

Version 25.9 is ready for testing. If all goes well with the changes to save files, this will become the official release.

Download links:
Windows: ftp://mersenne.org/gimps/p95v259.zip
Windows 64-bit: ftp://mersenne.org/gimps/p64v259.zip
Windows service: ftp://mersenne.org/gimps/winnt259.zip
Windows 64-bit service: ftp://mersenne.org/gimps/win64nt259.zip
Linux: ftp://mersenne.org/gimps/mprime259.tar.gz
Linux 64-bit: ftp://mersenne.org/gimps/mprime259-linux64.tar.gz
Mac OS X: ftp://mersenne.org/gimps/mprime259-MacOSX.tar.gz
Source: ftp://mersenne.org/gimps/source259.tar.gz


This release contains both performance improvements and bug fixes.

From the included whatsnew.txt file:

1) Performance increases for the 64-bit version by taking advantage of the eight extra SSE2 registers. The 32-bit version may also be a bit faster by taking advantage of the Core 2 architecture. The FFT code was originally optimized for the Pentium 4 where instructions like "movapd reg,reg" take 6 clocks and should be avoided, whereas on Core 2 the same instruction uses just 1 clock cycle and improves scheduling. The 32-bit version may be slightly faster or slower on the Pentium 4, but not enough to worry about. Preliminary benchmarks indicate:

32-bit Pentium 4: between 2% slower and 1% faster
32-bit Core 2: between 2% and 6% faster
64-bit Pentium 4: between 2% and 4% faster
64-bit Core 2: between 10% and 13% faster
64-bit Core i7: between 7% and 12% faster
64-bit Phenom: between 10% and 13% faster
AMD64: between 1% slower and 1% faster

2) Intermediate save files were overhauled. Trial factoring, P-1, and ECM now support multiple save files. The backup save file uses a .bu extension. While a new save file is being written a .write extension is used and renamed when successfully written (thus, briefly there are 3 save files). NOTE #1: LL save files no longer use an initial letter of q and r. The program will automatically use these obsolete files if necessary and delete them when no longer needed. NOTE #2: ECM and P-1 save files on k*2^n+1 no longer use file names beginning with d and l. The -An command line argument no longer appends an extension to save file names. NOTE #3: The client now supports keeping three save files.

Bug fixes are described here: http://www.mersenneforum.org/showpos...44&postcount=2

[Prime95]

Known bugs and fixes:

1) PFactor save files begin with the letter p -- it should be m like all other P-1 work. Fixed in next release.

2) Linux and Mac OS X load averaging, -c and -s command line arguments do not work properly. All fixed in next release.

3) Corrupt save files can cause memory corruption and/or a program crash. Fixed in next release.

4) PRP of 99*1^50+2 fails. Fixed in next release.

5) When PRP first starts it incorrectly outputs a "time per iteration" message. Fixed in next release.

6) Some SeventeenOrBust users would like hourly checkins with the server. The next release will allow setting DAYS_BETWEEN_CHECKINS as low as 0.04 in prime.txt. The GUI will only allow integer values.

7) Writing save files could be a little bit faster. Fixed in the next release.

8) A quad core i7 with hyperthreading disabled in the BIOS is incorrectly identified as a dual core with hyperthreading. Fix coded for the next release but not tested.

9) Some memory allocation errors cause a crash - especially during P-1 and ECM GCD. Improved in next release.

10) P-1 start 2 bits before the TF limit -- this isn't optimal. Changed to 1 bit in next release.

11) If a worktodo.txt file contains more than one ECM lines working on the same exponent then the percent complete will be incorrectly copied from the first occurrence to the subsequent occurrences. Fixed in 25.10.

12) The number of ECM curves to run on a worktodo.txt file would sometimes be output incorrectly using 64-bit executables. Fixed in 25.10.

13) The Linux 32-bit version links to libcurl.so.3 rather than the newer libcurl.so.4. Fixed in 25.10.

14) Many Linux and FreeBSD do not have libcurl installed. In 25.11 libcurl will be staticly linked into mprime.

15) If there is an error reading a save file and that save file cannot be deleted, then an infinite loop will result. Fixed in 25.11 by limiting read attempts to 99 tries.

16) If two workers are ECMing the same number to different B1 limits, then there can be some confusion reading save files on restarting. No fix yet. If two or more workers are ECMing the same number with the same B1, then 25.11 will make a better effort of keeping the save files associated with the same workers when you stop/start.

17) The server only accepts 50 benchmark timings. When a quad core processor does a benchmark, the various multithread timings fill up our 50 limit and the TF data does not get sent. Fixed in 25.11. Benchmark data will only be sent for 1, 2, and 4 threads. Also data for 768K FFT will not be sent.

[Uncwilly]

Should those of us that are using build 2 upgrade?

[db597]

This is great news! After quite a number of years since it's release we're finally seeing the benefits of 64bits on LL. This just might provide enough incentive for me to finally leave XP 32bit!

[tmorrow]

Originally Posted by Uncwilly

Should those of us that are using build 2 upgrade?

The zip pack contains build4 rather than build2 of prime95.exe and has updates to the text files, mainly about the 64 bit improvements using the 8 extra SSE2 registers - release notes etc. It's your call on whether to upgrade now or wait until the official release. I'm not sure what the delta is between build2 and build4.

[ET_]

Does mprime259-MacOSX64 work fine on Linux machines?

Luigi

[Prime95]

Originally Posted by ET_

Does mprime259-MacOSX64 work fine on Linux machines?

No. I'll work on the Linux builds today.

[ET_]

Originally Posted by Prime95

No. I'll work on the Linux builds today.

Thank you George

Luigi

[Prime95]

Originally Posted by Uncwilly

Should those of us that are using build 2 upgrade?

Check out the bug fix list and decide for yourself if you want to. I would for the better save file support.

[CADavis]

I exited prime95 and copied the exe 25.9b4 over 25.9b3, vista 64-bit, and upon starting got this:

[Main thread Mar 17 16:07] Mersenne number primality test program version 25.9
[Main thread Mar 17 16:07:55] Starting workers.
[Main thread Mar 17 16:08:15] Use count locking problem in reading worktodo file.

Anything to worry about?

[petrw1]

Originally Posted by CADavis

[Main thread Mar 17 16:08:15] Use count locking problem in reading worktodo file.

Anything to worry about?

I got this same error a couple months ago after a 25.9 upgrade and asked the same question on this forum. I got no response ... however, there were also no fallouts from the error either.