Prime95 version 29.3

Prime95 · 2017-09-08, 00:26

Prime95 version 29.3 build 1 is available.

From whatsnew.txt:

Code:

1)  A new error check for LL testing has been implemented.  This error check,
    called a Jacobi error check, has a 50% chance of detecting hardware error(s)
    since the last time a Jacobi error check was performed.  This error check
    takes roughly 30 seconds and is scheduled to run twice a day.  The program
    now saves two additional intermediate files that have passed the Jacobi
    error check.  This test requires use of the GMP (GNU multi-precision) library.
2)  The GCD step in P-1 and ECM factoring is faster.

This version is not heavily tested - but I don't expect many problems. Thanks to all that helped debug version 29.1 and 29.2. I'm eager to make this the official version as the new error check code will lead to fewer incorrect results reported to the server.

Get your bug reports in soon!

Download links:
Windows 64-bit: ftp://mersenne.org/gimps/p95v293.win64.zip
Linux 64-bit: ftp://mersenne.org/gimps/p95v293.linux64.tar.gz
Mac OS X: ftp://mersenne.org/gimps/p95v293.MacOSX.zip
Windows 32-bit: ftp://mersenne.org/gimps/p95v293.win32.zip
Linux 32-bit: ftp://mersenne.org/gimps/p95v293.linux32.tar.gz
FreeBSD11 64-bit: ftp://mersenne.org/gimps/p95v293.FreeBSD11-64.tar.gz
Source: ftp://mersenne.org/gimps/p95v293.source.zip

Prime95 · 2017-09-08, 00:27

1) The 1M FFT size is not selected when it should be. Only affects computers that do not support AVX and later extensions (i.e. SSE2 computers). Fixed in 29.4.
2) FFTs smaller than 6K do not run on Ryzen CPUs. Fixed in 29.4.
3) The .bu3 and .bu4 save files (the ones known to pass the Jacobi test) are not considered for use. Fixed in 29.4 build 3.

Dubslow · 2017-09-08, 01:43

I motion that this "build" nonsense be dropped and replaced with simply a third segment of the version number, such that this release be merely called 29.3.1 (or maybe it should have been 29.3.0?), and we can just say that some version of 29.3 is targeted for release and distribution.

storm5510 · 2017-09-08, 05:36

Quote:

Originally Posted by Prime95

Download links:
Windows 64-bit: ftp://mersenne.org/gimps/p95v293.win64.zip
Linux 64-bit: ftp://mersenne.org/gimps/p95v293.linux64.tar.gz
Mac OS X: ftp://mersenne.org/gimps/p95v293.MacOSX.zip
Windows 32-bit: ftp://mersenne.org/gimps/p95v293.win32.zip
Linux 32-bit: ftp://mersenne.org/gimps/p95v293.linux32.tar.gz
FreeBSD11 64-bit: ftp://mersenne.org/gimps/p95v293.FreeBSD11-64.tar.gz
Source: ftp://mersenne.org/gimps/p95v293.source.zip

For some reason, MS Edge seems unable to handle an FTP address. No problems with Firefox.

rudi_m · 2017-09-08, 09:41

Quote:

Originally Posted by Dubslow

I motion that this "build" nonsense be dropped and replaced with simply a third segment of the version number, such that this release be merely called 29.3.1 (or maybe it should have been 29.3.0?), and we can just say that some version of 29.3 is targeted for release and distribution.

I don't care how the version is pronounced. 29.3.1 or 29.3 build 1, anyways. But IMHO the build number should be part of the download binary archives. It's confusing for users to provide re-used file names with different content.

GP2 · 2017-09-08, 11:33

Quote:

Originally Posted by storm5510

For some reason, MS Edge seems unable to handle an FTP address. No problems with Firefox.

You can still get it via HTTP using, for example:

Code:

https://www.mersenne.org/ftp_root/gimps/p95v293.linux64.tar.gz

https://www.mersenne.org/ftp_root/gi...linux64.tar.gz

pepi37 · 2017-09-08, 11:51

Quote:

The GCD step in P-1 and ECM factoring is faster.

I can confirm this:
In my test case GCD stage 1 drop from 13.19 seconds to 1.127 seconds
GCD stage 2 from 11.57 seconds drop to 1.12 seconds.
All factor I found in previous version is found also in this new one :)
Great work!

Dubslow · 2017-09-08, 14:43

Quote:

Originally Posted by rudi_m

I don't care how the version is pronounced. 29.3.1 or 29.3 build 1, anyways. But IMHO the build number should be part of the download binary archives. It's confusing for users to provide re-used file names with different content.

That's part of why I suggest the condensed labeling, because it is both shorter and also strongly hints that "29.3" is an incomplete and ambiguous version number (and thus not suitable for file names).

GP2 · 2017-09-09, 00:49

I'm looking at the internal structure of the save files, and sometimes they differ from one another by only one iteration, for example:

Code:

p81419599: [Sep 09 00:12] LL (mprime): 922856 / 81419599 [1.13%]
p81419599.bu: [Sep 09 00:12] LL (mprime): 922855 / 81419599 [1.13%]
p81419599.bu2: [Sep 08 23:41] LL (mprime): 786059 / 81419599 [0.97%]

Code:

p82062787: [Sep 09 00:31] LL (mprime): 23495979 / 82062787 [28.63%]
p82062787.bu: [Sep 09 00:21] LL (mprime): 23449576 / 82062787 [28.58%]
p82062787.bu2: [Sep 09 00:12] LL (mprime): 23405855 / 82062787 [28.52%]
p82062787.bu3: [Sep 09 00:12] LL (mprime): 23405854 / 82062787 [28.52%]
p82062787.bu4: [Sep 08 12:07] LL (mprime): 20072393 / 82062787 [24.46%]

But other cases don't appear to do this, for example:

Code:

p26S2579: [Sep 09 00:33] LL (mprime): 3549381 / 78262579 [4.54%]
p26S2579.bu: [Sep 09 00:23] LL (mprime): 3503255 / 78262579 [4.48%]
p26S2579.bu2: [Sep 09 00:13] LL (mprime): 3457149 / 78262579 [4.42%]
p26S2579.bu3: [Sep 08 23:43] LL (mprime): 3321337 / 78262579 [4.24%]

Code:

p81471113: [Sep 09 00:41] LL (mprime): 27713599 / 81471113 [34.02%]
p81471113.bu: [Sep 09 00:31] LL (mprime): 27667700 / 81471113 [33.96%]
p81471113.bu2: [Sep 09 00:21] LL (mprime): 27621799 / 81471113 [33.90%]
p81471113.bu3: [Sep 09 00:12] LL (mprime): 27578749 / 81471113 [33.85%]
p81471113.bu4: [Sep 08 12:08] LL (mprime): 24285228 / 81471113 [29.81%]

Code:

p28S7887: [Sep 09 00:41] LL (mprime): 66655155 / 78287887 [85.14%]
p28S7887.bu: [Sep 09 00:31] LL (mprime): 66611262 / 78287887 [85.09%]
p28S7887.bu2: [Sep 09 00:21] LL (mprime): 66567374 / 78287887 [85.03%]
p28S7887.bu3: [Sep 09 00:12] LL (mprime): 66526121 / 78287887 [84.98%]
p28S7887.bu4: [Sep 08 12:09] LL (mprime): 63372973 / 78287887 [80.95%]

Is this the expected behavior?

Prime95 · 2017-09-09, 01:51

Quote:

Originally Posted by GP2

Is this the expected behavior?

No.

Madpoo · 2017-09-09, 02:06

Quote:

Originally Posted by GP2

I'm looking at the internal structure of the save files, and sometimes they differ from one another by only one iteration, for example:

Code:

p81419599: [Sep 09 00:12] LL (mprime): 922856 / 81419599 [1.13%]
p81419599.bu: [Sep 09 00:12] LL (mprime): 922855 / 81419599 [1.13%]
p81419599.bu2: [Sep 08 23:41] LL (mprime): 786059 / 81419599 [0.97%]

...
Is this the expected behavior?

The timestamps of the main and .bu are identical... did mprime stop right after making that first backup, making it write out a new main save file that was almost the same iteration?

Or anything else that might have stopped mprime and kicked it off again (some task you have running that monitors/restarts stuff?)

2017-09-08, 00:26	#1
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 5×11×137 Posts	Prime95 version 29.3 Prime95 version 29.3 build 1 is available. From whatsnew.txt: Code: 1) A new error check for LL testing has been implemented. This error check, called a Jacobi error check, has a 50% chance of detecting hardware error(s) since the last time a Jacobi error check was performed. This error check takes roughly 30 seconds and is scheduled to run twice a day. The program now saves two additional intermediate files that have passed the Jacobi error check. This test requires use of the GMP (GNU multi-precision) library. 2) The GCD step in P-1 and ECM factoring is faster. This version is not heavily tested - but I don't expect many problems. Thanks to all that helped debug version 29.1 and 29.2. I'm eager to make this the official version as the new error check code will lead to fewer incorrect results reported to the server. Get your bug reports in soon! Download links: Windows 64-bit: ftp://mersenne.org/gimps/p95v293.win64.zip Linux 64-bit: ftp://mersenne.org/gimps/p95v293.linux64.tar.gz Mac OS X: ftp://mersenne.org/gimps/p95v293.MacOSX.zip Windows 32-bit: ftp://mersenne.org/gimps/p95v293.win32.zip Linux 32-bit: ftp://mersenne.org/gimps/p95v293.linux32.tar.gz FreeBSD11 64-bit: ftp://mersenne.org/gimps/p95v293.FreeBSD11-64.tar.gz Source: ftp://mersenne.org/gimps/p95v293.source.zip

2017-09-08, 00:27	#2
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 5·11·137 Posts	1) The 1M FFT size is not selected when it should be. Only affects computers that do not support AVX and later extensions (i.e. SSE2 computers). Fixed in 29.4. 2) FFTs smaller than 6K do not run on Ryzen CPUs. Fixed in 29.4. 3) The .bu3 and .bu4 save files (the ones known to pass the Jacobi test) are not considered for use. Fixed in 29.4 build 3. Last fiddled with by Prime95 on 2017-11-02 at 15:43

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Prime95 version 27.3	Prime95	Software	148	2012-03-18 19:24
Prime95 version 26.3	Prime95	Software	76	2010-12-11 00:11
Prime95 version 25.5	Prime95	PrimeNet	369	2008-02-26 05:21
Prime95 version 25.4	Prime95	PrimeNet	143	2007-09-24 21:01
When the next prime95 version ?	pacionet	Software	74	2006-12-07 20:30

2017-09-08, 01:43	#3
Dubslow Basketry That Evening! "Bunslow the Bold" Jun 2011 40<A<43 -89<O<-88 3·29·83 Posts	I motion that this "build" nonsense be dropped and replaced with simply a third segment of the version number, such that this release be merely called 29.3.1 (or maybe it should have been 29.3.0?), and we can just say that some version of 29.3 is targeted for release and distribution.