Prime95 beta version 28.3

Prime95 · 2014-01-13, 01:18

Prime95 version 28.3 is available. This version is for you intrepid beta testers. I've run the torture test and done some QA, but I haven't run any double-checks yet. I don't expect any problems, and I am now using it on my computers.

This version includes nearly all the Haswell optimizations I plan on implementing. A few of these changes may also benefit Sandy/Ivy Bridge machines. On my Sandy Bridge I'm seeing a 1.5% improvement.

Download links:
Windows 64-bit: ftp://mersenne.org/gimps/p95v283.win64.zip
Linux 64-bit: ftp://mersenne.org/gimps/p95v283.linux64.tar.gz
Mac OS X: ftp://mersenne.org/gimps/p95v283.MacOSX.zip
Windows 32-bit: ftp://mersenne.org/gimps/p95v283.win32.zip
All versions (folder): ftp://mersenne.org/gimps/

Prime95 · 2014-01-13, 01:21

Placeholder for reporting bugs fixed

TheMawn · 2014-01-13, 01:39

Anything specific you'd like us to do? I'll just queue up two or three DC's for each worker otherwise.

kracker · 2014-01-13, 01:42

I have three IB's and a SB laptop I'll be testing

. Are there any improvements from 28.2(1) for Haswell?

sdbardwick · 2014-01-13, 01:49

Quote:

Originally Posted by kracker

Are there any improvements from 28.2(1) for Haswell?

My Haswell went from 4.059 ms/iteration to 4.019 ms/it, so yes

(4 threads, 1 exponent)
Edit: Although the range of variation in time per iteration appears to have increased.

2014-01-13, 09:27

I don't suppose this needs testing from Ivy Bridge stress testing only users?

bayanne · 2014-01-13, 14:46

I'll test this on my Mac Pro ...

petrw1 · 2014-01-13, 15:56

Instead of one setting for # of iterations between screen updates...could it be either:
- # of minutes or hours between updates
Or
- allow a diff # of iterations for eack worker.

I find with mixed work types the time to complete X iterations is VERY different.
I have 1 core doing ECM on 2377 and another doing DC.
If I set iterations to 100000 for example DC updates about hourly and ECM every second. At 500000000 ECM every 10 minutes....DC never until it completes.

Thx

blip · 2014-01-14, 12:51

Linux link seems to be broken...

chalsall · 2014-01-14, 18:03

Quote:

Originally Posted by blip

Linux link seems to be broken...

Indeed.

Try ftp://mersenne.org/gimps/p95v283.liux64.tar.gz instead (what's an "n" between friends)....

chalsall · 2014-01-15, 17:31

George.

I don't know under what environment you complied the Linux version in, but under CentOS 6.4 I get the following:

Code:

[chalsall@backup t]$ ./mprime -v
./mprime: /lib64/libc.so.6: version `GLIBC_2.15' not found (required by ./mprime)
./mprime: /lib64/libc.so.6: version `GLIBC_2.14' not found (required by ./mprime)

Any chance we could get a "statically linked" version?

2014-01-13, 01:18	#1
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 2×53×71 Posts	Prime95 beta version 28.3 Prime95 version 28.3 is available. This version is for you intrepid beta testers. I've run the torture test and done some QA, but I haven't run any double-checks yet. I don't expect any problems, and I am now using it on my computers. This version includes nearly all the Haswell optimizations I plan on implementing. A few of these changes may also benefit Sandy/Ivy Bridge machines. On my Sandy Bridge I'm seeing a 1.5% improvement. Download links: Windows 64-bit: ftp://mersenne.org/gimps/p95v283.win64.zip Linux 64-bit: ftp://mersenne.org/gimps/p95v283.linux64.tar.gz Mac OS X: ftp://mersenne.org/gimps/p95v283.MacOSX.zip Windows 32-bit: ftp://mersenne.org/gimps/p95v283.win32.zip All versions (folder): ftp://mersenne.org/gimps/ Last fiddled with by Prime95 on 2014-01-14 at 19:46 Reason: (a working link for linux version)

2014-01-13, 15:56	#8
petrw1 1976 Toyota Corona years forever! "Wayne" Nov 2006 Saskatchewan, Canada 2²·7·167 Posts	is it too late for requests? Instead of one setting for # of iterations between screen updates...could it be either: - # of minutes or hours between updates Or - allow a diff # of iterations for eack worker. I find with mixed work types the time to complete X iterations is VERY different. I have 1 core doing ECM on 2377 and another doing DC. If I set iterations to 100000 for example DC updates about hourly and ECM every second. At 500000000 ECM every 10 minutes....DC never until it completes. Thx

2014-01-15, 17:31	#11
chalsall If I May "Chris Halsall" Sep 2002 Barbados 37·263 Posts	George. I don't know under what environment you complied the Linux version in, but under CentOS 6.4 I get the following: Code: [chalsall@backup t]$ ./mprime -v ./mprime: /lib64/libc.so.6: version `GLIBC_2.15' not found (required by ./mprime) ./mprime: /lib64/libc.so.6: version `GLIBC_2.14' not found (required by ./mprime) Any chance we could get a "statically linked" version?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
LLR beta Version 3.8.13 (deprecated)	Jean Penné	Software	111	2015-01-26 21:41
Prime95 beta version 28.4	Prime95	Software	20	2014-03-02 02:51
Prime95 version 27.1 early preview, not-even-close-to-beta release	Prime95	Software	126	2012-02-09 16:17
Beta version 24.12 available	Prime95	Software	33	2005-06-14 13:19
Beta version of PRP	Prime95	PSearch	15	2004-09-17 19:21

2014-01-13, 01:21	#2
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 2×53×71 Posts	Placeholder for reporting bugs fixed

2014-01-13, 01:39	#3
TheMawn May 2013 East. Always East. 11010111111₂ Posts	Anything specific you'd like us to do? I'll just queue up two or three DC's for each worker otherwise.

2014-01-13, 01:42	#4
kracker "Mr. Meeseeks" Jan 2012 California, USA 4170₈ Posts	I have three IB's and a SB laptop I'll be testing . Are there any improvements from 28.2(1) for Haswell?

2014-01-13, 09:27	#6
Octopuss 2³×251 Posts	I don't suppose this needs testing from Ivy Bridge stress testing only users?

2014-01-13, 14:46	#7
bayanne "Tony Gott" Aug 2002 Yell, Shetland, UK 2²·83 Posts	I'll test this on my Mac Pro ...

2014-01-14, 12:51	#9
blip Jan 2014 2·73 Posts	Linux link seems to be broken...