Posts by mmonnin

1) Message boards : Number crunching : Strange credit calculation for Spectra of Latin squares (Message 2023) Posted 28 days ago by mmonnin Post: In your 25 years have you come up with something better? Yes it can suck. But its used as its the easiest option when run times are not consistent.
2) Message boards : Number crunching : 2 Hour Deadline (Message 2013) Posted 26 May 2025 by mmonnin Post: Any chance of making multi-core tasks? I've got 24 cores on four of my machines. I could get through these much faster. The best multi-core task - is the task splitted to other tasks. :) Unless it uses a lot of cache causing it to fill/swap and slowing all tasks down.
3) Message boards : Number crunching : Strange credit calculation for Spectra of Latin squares (Message 2006) Posted 23 May 2025 by mmonnin Post: Hi! When granted credit is proportional to amount of calculations, for each computer granted credit = run_time / k. But k are different for different computers. :) The runtime resets if the task is paused, PC shut down etc
4) Message boards : Number crunching : 2 Hour Deadline (Message 1973) Posted 30 Apr 2025 by mmonnin Post: I don't really give a F if they are short. A 2 hour deadline means BOINC Mgr will interrupt ALL other task to run it. Some projects do not have any checkpoints and hours of work could have been wasted for 2 min of Rakesearch work. No deadline should be less than several days.
5) Message boards : Number crunching : 2 Hour Deadline (Message 1961) Posted 10 Apr 2025 by mmonnin Post: Are you nuts? Sent 10 Apr 2025, 8:11:31 UTC Deadline 10 Apr 2025, 10:11:31 UTC Spectra of Latin squares Every one of the tasks I downloaded is a resend as they couldn't be completed in time.
6) Message boards : Number crunching : can't get tasks for my Win7 machine as it says (Message 1713) Posted 31 Oct 2022 by mmonnin Post: My PCs ran out so I assumed there was no work. I didn't check server status, did you?
7) Message boards : News : Future of the RakeSearch project (Message 1029) Posted 6 Jun 2019 by mmonnin Post: For search in rank 10, we made a new application. And of course, we implemented the most powerful optimizations. (Due to changes of structure, the need for some is gone). But we hope that this application can be additionally optimized - it's another very interesting problem! For rank 9 a third-party client was released which utilizes SSE/AVX etc. variants and is significantly faster compared to the original client. Are you telling me that the current rank 10 app release (which requires run times of up to 4 hrs for some of the tasks I completed - compared to 20 min. for rank 9 tasks on the same machine) already employs the same optimizations incl. an autodetection module to select the appropriate SSE/AVX code? Michael. Why would you assume a separate data set will complete in the same time? Might as well say I want my PrimeGrid Seventeen or Bust task to complete in the same time as Sophie Germain. Daniel and hoarfrost has already said multiple times that much of the optimizations are already in the rank 10 app and to not expect the huge reduction with any AVX/SSE optimization.
8) Message boards : Number crunching : Optimized RakeSearch app for rank 9 (computations finished) (Message 1026) Posted 5 Jun 2019 by mmonnin Post: Anything you can do would be very much appreciated. Some of the new units I've returned have taken over 2 hours to complete.... Run time is irrelevant for credit.
9) Message boards : Number crunching : Optimized RakeSearch app for rank 9 (computations finished) (Message 1021) Posted 5 Jun 2019 by mmonnin Post: Anything you can do would be very much appreciated. Some of the new units I've returned have taken over 2 hours to complete.... Run time is irrelevant for credit.
10) Message boards : News : Future of the RakeSearch project (Message 980) Posted 30 May 2019 by mmonnin Post: Edit: Evidently the refreshes of a post submission worked but I could never see them. :(
11) Message boards : News : Future of the RakeSearch project (Message 977) Posted 29 May 2019 by mmonnin Post: Over 300 v10 tasks completed w/o error. There is no more work atm for the new app.
12) Questions and Answers : Unix/Linux : Project not reading app_config file (Message 963) Posted 23 May 2019 by mmonnin Post: Settings in app_config can be re-read when using the BOINC read config file function. BOINCTasks specifies app_config in the button. The event log will show this and any possible errors in the syntax. app_info.xml files require a client restart.
13) Questions and Answers : Unix/Linux : Segfault on Linux (Message 813) Posted 11 Mar 2019 by mmonnin Post: Was that Ryzen purchased early? Early Zen CPUs segfault and could be RMA'd.
14) Message boards : Number crunching : Optimized RakeSearch app for rank 9 (computations finished) (Message 729) Posted 5 Feb 2019 by mmonnin Post: Thanks Daniel for your efforts in getting this out there. However I seem to have a problem with your new programme, all I get is <core_client_version>7.4.25</core_client_version> <![CDATA[ <message> process exited with code 193 (0xc1, -63) </message> <stderr_txt> SIGILL: illegal instruction Stack trace (10 frames): [0x411740] [0x4f0b10] [0x40c282] [0x409d6c] [0x40cdc4] [0x4063dd] [0x404f5c] [0x4f21a4] [0x4f2421] [0x4058b6] Exiting... </stderr_txt> Which is trashing a heap of work units, I re-downloaded and re-installed the files but this made no difference. I noticed that the line <platform>xxxx</platform> had been added to the apt_info.xml file so I removed that to see if it made a difference but alas it did not. So I will re-install your older version and see if all starts working again. Thanks Conan Which app was used? The 955 doesn't have AVX so the SSE app would be required. Same exe name so these should be a drop in replacement. Thanks for the great work Daniel.
15) Message boards : Number crunching : Sprint on project RakeSearch from 07/26/2018 22:00 (UTC) to 07/29/2018 21:59 (UTC) (Message 554) Posted 13 Aug 2018 by mmonnin Post: Were these when tasks became validated or either of the two tasks per unit? The long tail would make me assume it was when tasks were validated and thus push a higher completion count towards the 2nd half of the event.
16) Questions and Answers : Windows : Scheduler request failed: Error 413 ? (Message 494) Posted 27 Jul 2018 by mmonnin Post: GenuineIntel Intel(R) Xeon(R) CPU E5-2670 0 @ 2.60GHz [Family 6 Model 45 Stepping 7] (128 processors) Self inflicted?
17) Message boards : News : New badges for total credit! (Message 212) Posted 8 Dec 2017 by mmonnin Post: New badge levels can always be added to get to 1bil.
18) Message boards : News : The last part of November and the first part of December results! (Message 200) Posted 5 Dec 2017 by mmonnin Post: What will be the other badge thresholds be set at? Link above shows there are more point levels to come although the weasel is missing from that link.
19) Message boards : Number crunching : Optimized RakeSearch app for rank 9 (computations finished) (Message 190) Posted 2 Dec 2017 by mmonnin Post: Trying it out now. Posting as a time reference between apps. :)
20) Message boards : Number crunching : Optimized RakeSearch app for rank 9 (computations finished) (Message 188) Posted 2 Dec 2017 by mmonnin Post: Not sure it's working on my 1950x with AVX2 app. 34min into my 1st set of tasks and its only 55% done. My older 3770k with only AVX is 19min in and 66%. It works, but slower than expected. Please try AVX version, it may be faster for you. Recently I read that PEXT instruction from BMI2 set is slow on AMD CPUs, and AVX2 app uses it in most performance-critical part. This can explain why app is so slow on AMD CPU. https://www.reddit.com/r/Amd/comments/60i6er/ryzen_and_bmi2_strange_behavior_and_high_latencies/ Maybe AVX2 app without BMI instructions would be better here. I will take a look on this. Yeah, I guess I meant it wasn't working as well as expected. Seeing big numbers put up by others. Point/CPU Sec went from 0.0454 average to 0.0613 with the AVX app. A good 35% improvement on the 1950x. Great job and thanks for another optimized app for BOINC!

Next 20