Message boards :
Number crunching :
Behaviour of MT Threaded Application
Message board moderation
Author | Message |
---|---|
Conan Send message Joined: 5 Sep 11 Posts: 46 Credit: 7,102,856 RAC: 3,093 |
On my Linux computers running Fedora 16 (64 Bit), I have noticed that while the Resources use shows from 89% to 99% and all cores in use, the Process use is only 4% per core on my 4 core machine and 6% per core on my 6 core machine. This then jumps to 96% to 100% and 1 core in use (with all other cores still held but not being used) when the WU is finishing up the calculation. This then goes to 84% to 100% and all available cores again when the gnfs sieve part of the WU kicks in. So for the first stage all cores are grabbed but the computer is doing very little but in the second stage the computer is fully utilised. I would assume that the Windows WU does a similar thing. Just some observations. Conan |
Conan Send message Joined: 5 Sep 11 Posts: 46 Credit: 7,102,856 RAC: 3,093 |
I still can't get a WU to finish on my AMD Phenom X6 Linux 64 bit computer. I am now getting File transfer errors and 'ecm' command errors. Boinc messages shows File Size : 103866468.000000 Limit : 5000000.000000. See 639963 and 640058 So my file size does not match what it should be and this is failing the WU. Is there a cure for this ? Conan |
yoyo_rkn Volunteer moderator Project administrator Project developer Project tester Volunteer developer Volunteer tester Project scientist Send message Joined: 22 Aug 11 Posts: 734 Credit: 17,574,526 RAC: 510 |
Seems that your Boinc hasn't downloaded the ecm binary, which is here http://yafu.dyndns.org/yafu/download/ecm-6.2.1_x86_64-pc-linux-gnu and exactly 536768 bytes. Can you start a WU, stop it and send me the content of the slot to yoyo (a) mailueberfall . de? Seems that there is no ecm binary there. yoyo |
yoyo_rkn Volunteer moderator Project administrator Project developer Project tester Volunteer developer Volunteer tester Project scientist Send message Joined: 22 Aug 11 Posts: 734 Credit: 17,574,526 RAC: 510 |
I found a possible issue regarding the "ecm command not found" and created a new linux 64 version. yoyo |
Conan Send message Joined: 5 Sep 11 Posts: 46 Credit: 7,102,856 RAC: 3,093 |
Right, thanks YOYO. I have detached and reattached to the project just to make sure that I have downloaded the latest version. The "ecm" binary is in the project directory on both computers, I did not check to see if it was there before. Have downloaded another WU on both machines so will now run them and see how they go. Conan |
Conan Send message Joined: 5 Sep 11 Posts: 46 Credit: 7,102,856 RAC: 3,093 |
Thanks YOYO the application is working now and I have validated and successful results. Reading the Boinc messages it reads that the app found "negative CPU", what does that mean? Conan |
Speedy51 Send message Joined: 25 Jan 12 Posts: 23 Credit: 1,529,974 RAC: 0 |
I'm aware this ap is very new. Are there plans to have the progress % complete while the task is processing? |
yoyo_rkn Volunteer moderator Project administrator Project developer Project tester Volunteer developer Volunteer tester Project scientist Send message Joined: 22 Aug 11 Posts: 734 Credit: 17,574,526 RAC: 510 |
Don't know what this really means. I saw it also. yoyo |
yoyo_rkn Volunteer moderator Project administrator Project developer Project tester Volunteer developer Volunteer tester Project scientist Send message Joined: 22 Aug 11 Posts: 734 Credit: 17,574,526 RAC: 510 |
I'm aware this ap is very new. Are there plans to have the progress % complete while the task is processing? There is currently now way to show a progress bar. yoyo |
Speedy51 Send message Joined: 25 Jan 12 Posts: 23 Credit: 1,529,974 RAC: 0 |
I'm aware this ap is very new. Are there plans to have the progress % complete while the task is processing? Thanks for your prompt answer |
Krzychu P. Send message Joined: 23 Sep 11 Posts: 3 Credit: 52,276 RAC: 0 |
I'll wirte it here. Could you take a look at this WU: yafu_C104_1328030114_495_0 Something went wrong at the end of computing, cause it finished with error. In the meantime it checkpointed in the range of 30 to 90 minutes. What should I do, to finish such WUs correctly? (now I have similar long-running WU) |
Krzychu P. Send message Joined: 23 Sep 11 Posts: 3 Credit: 52,276 RAC: 0 |
I don't know where my last post is, so I'll write again. I'll wirte it here. Once again the same situation as above. WU finished with (Stderr output): linear algebra completed 261019 of 261780 dimensions (99.7%, ETA 0h 0m) app exit status: 0xff 14:45:09 (752): called boinc_finish On the WU's page I found: Client state Compute error Exit status 195 (0xc3) EXIT_CHILD_FAILED Could you fix it? I've reseted project in my boinc manager, but it didn't help. |
Krzychu P. Send message Joined: 23 Sep 11 Posts: 3 Credit: 52,276 RAC: 0 |
|
yoyo_rkn Volunteer moderator Project administrator Project developer Project tester Volunteer developer Volunteer tester Project scientist Send message Joined: 22 Aug 11 Posts: 734 Credit: 17,574,526 RAC: 510 |
Do you have still the Boinc log if this? The result claims that the maximum disc usage was exeeded. The maximum is set to 2GB. I wonder if this was really exeeded. yoyo |
frankhagen Send message Joined: 2 Sep 11 Posts: 6 Credit: 2,081,426 RAC: 0 |
you really think anything inside that: <result_name>yafu_C106_1330635007_173_0</result_name> <checkpoint_cpu_time>2205.109000</checkpoint_cpu_time> <checkpoint_elapsed_time>7130.078125</checkpoint_elapsed_time> <fraction_done>0.000000</fraction_done> as it should be? talking about a job running in MT4 mode now for 140 minutes... |
yoyo_rkn Volunteer moderator Project administrator Project developer Project tester Volunteer developer Volunteer tester Project scientist Send message Joined: 22 Aug 11 Posts: 734 Credit: 17,574,526 RAC: 510 |
No, I mean in the Boinc stdout log. There should be something stated, that the maximal disc was exeeded. |
frankhagen Send message Joined: 2 Sep 11 Posts: 6 Credit: 2,081,426 RAC: 0 |
No, I mean in the Boinc stdout log. There should be something stated, that the maximal disc was exeeded. i know! i'm not at that point yet - 460 MB currently, 03:38:50 done, boinc not showing any sign of a checkpoint and still sitting there at 0%. :( that chain of different apps being run is freaky! and the wrapper does definitely not pick up the correct runtime. back to the lab.. |
Viking69 Send message Joined: 16 Feb 12 Posts: 12 Credit: 107,382 RAC: 0 |
I'm aware this ap is very new. Are there plans to have the progress % complete while the task is processing? Yoyo, is there a reason as to why this is not possable? |
yoyo_rkn Volunteer moderator Project administrator Project developer Project tester Volunteer developer Volunteer tester Project scientist Send message Joined: 22 Aug 11 Posts: 734 Credit: 17,574,526 RAC: 510 |
The cpu time is NOT used to calculate the credit in creditnew. But anyway, I update the wrapper in version 130.02 and now it seems a bitter better with the cpu time. But as said, you gain nothing by a correct cpu time, thos is NOT used to calculate the credits. yoyo |