mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > PrimeNet

Reply
 
Thread Tools
Old 2017-10-25, 15:49   #12
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

2·3,701 Posts
Default

Quote:
Originally Posted by rudi_m View Post

I would have expected that the default BenchCores should be the same like CoresPerTest.

I can fix that.
Prime95 is offline   Reply With Quote
Old 2017-10-25, 16:26   #13
Mark Rose
 
Mark Rose's Avatar
 
"/X\(‘-‘)/X\"
Jan 2013

29·101 Posts
Default

I had the no idea about the BenchCores and BenchWorkers being set apart from CoresPerTest and WorkerThreads.

I'll go around and update that on my machines.
Mark Rose is offline   Reply With Quote
Old 2017-10-25, 18:21   #14
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

1CEA16 Posts
Default

Quote:
Originally Posted by Mark Rose View Post
I had the no idea about the BenchCores and BenchWorkers being set apart from CoresPerTest and WorkerThreads.

I'll go around and update that on my machines.
The default in 29.3 for auto-bench is to use all cores. In 29.4 it will be the sum of each workers cores-per-test.
Prime95 is offline   Reply With Quote
Old 2017-10-27, 09:53   #15
rudi_m
 
rudi_m's Avatar
 
Jul 2005

2·7·13 Posts
Default

Quote:
Originally Posted by Prime95 View Post
The default in 29.3 for auto-bench is to use all cores. In 29.4 it will be the sum of each workers cores-per-test.
Thanks!

By the way I've noticed another possible issue with the auto-benchmark. If mprime is stopped while the benchmark is running, then it writes incomplete results to gwnum.txt. Example:

Code:
[Main thread Oct 27 05:28] Benchmarking multiple workers to tune FFT selection.
[Work thread Oct 27 05:28] Stopping primality test of M41709119 at iteration 736439 [1.76%]
[Work thread Oct 27 05:28] Worker stopped while running needed benchmarks.
[Main thread Oct 27 05:28] Timing 2240K FFT, 3 cores, 1 worker.  Average times: 10.34 ms.  Total throughput: 96.75 iter/sec.
[Main thread Oct 27 05:28] Timing 2240K FFT, 3 cores, 1 worker.  Average times: 10.25 ms.  Total throughput: 97.56 iter/sec.
[Main thread Oct 27 05:28] Timing 2240K FFT, 3 cores, 1 worker.  Average times: 11.09 ms.  Total throughput: 90.13 iter/sec.
[Main thread Oct 27 05:28] Timing 2240K FFT, 3 cores, 1 worker.  Average times: 10.86 ms.  Total throughput: 92.12 iter/sec.
[Main thread Oct 27 05:29] Timing 2240K FFT, 3 cores, 1 worker.  Average times: 10.71 ms.  Total throughput: 93.33 iter/sec.
[Main thread Oct 27 05:29] Timing 2240K FFT, 3 cores, 1 worker.  Average times: 14.75 ms.  Total throughput: 67.81 iter/sec.
[Main thread Oct 27 05:29] Timing 2304K FFT, 3 cores, 1 worker.  Average times: 11.60 ms.  Total throughput: 86.22 iter/sec.
[Main thread Oct 27 05:29] Timing 2304K FFT, 3 cores, 1 worker.  [Oct 27 05:30] Stopping all worker threads.
[Work thread Oct 27 05:30] Benchmarks complete, restarting worker.
[Work thread Oct 27 05:30] Worker stopped.
Don't know if this is a problem. But another unlucky thing is that on this machine the benchmark almost never completes because it always want to run at 5:30 but I stop it everyday at 5:30, See

Code:
$ crontab -l
30 21 * * 1-5 cd "$HOME/MPrime/$(hostname)" && nohup ./mprime -d >> mprime.log 2>&1 &
30 05 * * 1-5 killall mprime

$ grep "Benchmarking multiple workers"  mprime.log
[Main thread Sep 14 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Sep 16 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Sep 17 02:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Sep 17 23:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Sep 20 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Sep 23 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Sep 24 02:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Sep 24 23:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Sep 26 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Sep 29 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Sep 30 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 1 02:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 1 23:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 3 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 4 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 7 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 9 05:29] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 10 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 11 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 13 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 14 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 15 02:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 15 23:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 20 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 21 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 22 02:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 22 23:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 24 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 25 05:30] Benchmarking multiple workers to tune FFT selection.
[Main thread Oct 27 05:28] Benchmarking multiple workers to tune FFT selection.

$ grep "^BenchData"  gwnum.txt  | cut -d, -f 6 | sort | uniq -c
     21 2017-09-16
     42 2017-09-17
     21 2017-09-23
     60 2017-09-24
     39 2017-09-30
     78 2017-10-01
     39 2017-10-07
      3 2017-10-09
      9 2017-10-14
     30 2017-10-15
     21 2017-10-21
     42 2017-10-22
      7 2017-10-27
rudi_m is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
LL Credit Calculator swl551 Software 0 2012-12-04 17:49
What does the gHz credit actually mean? mack Information & Answers 5 2009-12-17 10:41
How much credit for LL? hj47 PrimeNet 26 2009-01-23 22:14
My v4 credit appears in: uigrad PrimeNet 14 2008-12-31 02:37
v4 Credit on v5 precius1 Information & Answers 3 2008-11-03 22:23

All times are UTC. The time now is 00:06.

Mon Apr 12 00:06:23 UTC 2021 up 3 days, 18:47, 1 user, load averages: 1.44, 1.48, 1.53

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.