Page 5 of 9 FirstFirst 123456789 LastLast
Results 41 to 50 of 88

Thread: 8. BOINC Pentathlon

  1. #41
    Join Date
    Sep 2010
    Location
    Leiden, the Netherlands
    Posts
    4,372
    Quote Originally Posted by SuicideCabbage View Post
    I ran into a new issue, the 5830s are a 1GB card, in BOINC they are reporting 512MB thus Einstein is refusing them. Is this an error or is it simply reporting what is available to be used for OpenCL? If it's an error, how can I fix it? if it's not, is there any way I can force it to run via config files?
    Can you show the first 15-20 lines of the BOINC log file?


  2. #42
    Mon 08 May 2017 10:53:18 PM EDT | | Starting BOINC client version 7.2.42 for x86_64-pc-linux-gnu
    Mon 08 May 2017 10:53:18 PM EDT | | log flags: file_xfer, sched_ops, task
    Mon 08 May 2017 10:53:18 PM EDT | | Libraries: libcurl/7.35.0 OpenSSL/1.0.1f zlib/1.2.8 libidn/1.28 librtmp/2.3
    Mon 08 May 2017 10:53:18 PM EDT | | Data directory: /var/lib/boinc-client
    Mon 08 May 2017 10:53:18 PM EDT | | OpenCL: AMD/ATI GPU 0: Cypress (driver version 1800.11, device version OpenCL 1.2 AMD-APP (1800.11), 512MB, 512MB available, 896 GFLOPS peak)
    Mon 08 May 2017 10:53:18 PM EDT | | OpenCL: AMD/ATI GPU 1: Cypress (driver version 1800.11, device version OpenCL 1.2 AMD-APP (1800.11), 512MB, 512MB available, 896 GFLOPS peak)
    Mon 08 May 2017 10:53:18 PM EDT | | OpenCL CPU: AMD Phenom(tm) II X4 955 Processor (OpenCL driver vendor: Advanced Micro Devices, Inc., driver version 1800.11 (sse2), device version OpenCL 1.2 AMD-APP (1800.11))
    Mon 08 May 2017 10:53:18 PM EDT | | Creating new client state file
    Mon 08 May 2017 10:53:18 PM EDT | | Host name: Aimee-MSI-870
    Mon 08 May 2017 10:53:18 PM EDT | | Processor: 4 AuthenticAMD AMD Phenom(tm) II X4 955 Processor [Family 16 Model 4 Stepping 3]
    Mon 08 May 2017 10:53:18 PM EDT | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc rep_good nopl nonstop_tsc extd_apicid pni monitor cx16 popcnt lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt nodeid_msr hw_pstate npt lbrv svm_lock nrip_save
    Mon 08 May 2017 10:53:18 PM EDT | | OS: Linux: 3.13.0-32-generic
    Mon 08 May 2017 10:53:18 PM EDT | | Memory: 7.80 GB physical, 0 bytes virtual
    Mon 08 May 2017 10:53:18 PM EDT | | Disk: 82.40 GB total, 74.46 GB free
    Mon 08 May 2017 10:53:18 PM EDT | | Local time is UTC -4 hours
    Mon 08 May 2017 10:53:18 PM EDT | | Config: GUI RPCs allowed from:
    Mon 08 May 2017 10:54:36 PM EDT | | Reading preferences override file
    Mon 08 May 2017 10:54:36 PM EDT | | Preferences:
    Mon 08 May 2017 10:54:36 PM EDT | | max memory usage when active: 6386.34MB
    Mon 08 May 2017 10:54:36 PM EDT | | max memory usage when idle: 7184.64MB
    Mon 08 May 2017 10:54:36 PM EDT | | max disk usage: 74.16GB
    Mon 08 May 2017 10:54:36 PM EDT | | (to change preferences, visit a project web site or select Preferences in the Manager)
    Mon 08 May 2017 10:54:36 PM EDT | | Resuming GPU computation
    Mon 08 May 2017 10:54:46 PM EDT | | Fetching configuration file from https://einsteinathome.org/get_project_config.php
    Mon 08 May 2017 10:55:08 PM EDT | | Running CPU benchmarks
    Mon 08 May 2017 10:55:08 PM EDT | | Suspending computation - CPU benchmarks in progress
    Mon 08 May 2017 10:55:40 PM EDT | | Benchmark results:
    Mon 08 May 2017 10:55:40 PM EDT | | Number of CPUs: 4
    Mon 08 May 2017 10:55:40 PM EDT | | 3642 floating point MIPS (Whetstone) per CPU
    Mon 08 May 2017 10:55:40 PM EDT | | 16009 integer MIPS (Dhrystone) per CPU
    Mon 08 May 2017 10:55:40 PM EDT | | Resuming computation

    And from Einstein...
    2017-05-09 06:48:42.9887 [PID=17218] Request: [USER#xxxxx] [HOST#12528613] [IP xxx.xxx.xxx.210] client 7.2.42
    2017-05-09 06:48:43.0144 [PID=17218] [debug] have_master:1 have_working: 1 have_db: 1
    2017-05-09 06:48:43.0144 [PID=17218] [debug] using working prefs
    2017-05-09 06:48:43.0144 [PID=17218] [debug] have db 1; dbmod 1494043228.000000; global mod 1494043228.000000
    2017-05-09 06:48:43.0144 [PID=17218] [send] effective_ncpus 4 max_jobs_on_host_cpu 999999 max_jobs_on_host 999999
    2017-05-09 06:48:43.0144 [PID=17218] [send] effective_ngpus 2 max_jobs_on_host_gpu 999999
    2017-05-09 06:48:43.0144 [PID=17218] [send] Not using matchmaker scheduling; Not using EDF sim
    2017-05-09 06:48:43.0144 [PID=17218] [send] CPU: reqclient(): unrecognized: <final_peak_working_set_size>267931648</final_peak_working_set_size>
    2017-05-09 06:48:43.0145 [PID=17218] [send] ATI: req 103680.00 sec, 2.00 instances; est delay 0.00
    2017-05-09 06:48:43.0145 [PID=17218] [send] work_req_seconds: 0.00 secs
    2017-05-09 06:48:43.0145 [PID=17218] [send] available disk 73.89 GB, work_buf_min 8640
    2017-05-09 06:48:43.0145 [PID=17218] [send] active_frac 0.999966 on_frac 1.000000 DCF 1.000000
    2017-05-09 06:48:43.0154 [PID=17218] [mixed] sending locality work first (0.5384)
    2017-05-09 06:48:43.0154 [PID=17218] [mixed] sending non-locality work second
    2017-05-09 06:48:43.0393 [PID=17218] [version] no app version available: APP#19 (einsteinbinary_BRP4) PLATFORM#7 (x86_64-pc-linux-gnu) min_version 0
    2017-05-09 06:48:43.0393 [PID=17218] [version] no app version available: APP#19 (einsteinbinary_BRP4) PLATFORM#1 (i686-pc-linux-gnu) min_version 0
    2017-05-09 06:48:43.0396 [PID=17218] [version] Checking plan class 'FGRPSSE'
    2017-05-09 06:48:43.0432 [PID=17218] [version] reading plan classes from file '/BOINC/projects/EinsteinAtHome/plan_class_spec.xml'
    2017-05-09 06:48:43.0432 [PID=17218] [version] plan class ok
    2017-05-09 06:48:43.0432 [PID=17218] [version] Don't need CPU jobs, skipping version 105 for hsgamma_FGRPB1 (FGRPSSE)
    2017-05-09 06:48:43.0432 [PID=17218] [version] Checking plan class 'FGRPSSE'
    2017-05-09 06:48:43.0432 [PID=17218] [version] plan class ok
    2017-05-09 06:48:43.0432 [PID=17218] [version] Don't need CPU jobs, skipping version 105 for hsgamma_FGRPB1 (FGRPSSE)
    2017-05-09 06:48:43.0432 [PID=17218] [version] no app version available: APP#32 (hsgamma_FGRPB1) PLATFORM#7 (x86_64-pc-linux-gnu) min_version 0
    2017-05-09 06:48:43.0432 [PID=17218] [version] no app version available: APP#32 (hsgamma_FGRPB1) PLATFORM#1 (i686-pc-linux-gnu) min_version 0
    2017-05-09 06:48:43.0455 [PID=17218] [version] Checking plan class 'FGRPopencl-ati'
    2017-05-09 06:48:43.0455 [PID=17218] [version] parsed project prefs setting 'gpu_util_fgrp': 1.000000
    2017-05-09 06:48:43.0455 [PID=17218] [version] OpenCL GPU RAM required min: 803209216.000000, supplied: 536870912
    2017-05-09 06:48:43.0455 [PID=17218] [version] Checking plan class 'FGRPopencl-nvidia'
    2017-05-09 06:48:43.0455 [PID=17218] [version] parsed project prefs setting 'gpu_util_fgrp': 1.000000
    2017-05-09 06:48:43.0455 [PID=17218] [version] No CUDA devices found
    2017-05-09 06:48:43.0455 [PID=17218] [version] Checking plan class 'FGRPopencl1K-ati'
    2017-05-09 06:48:43.0455 [PID=17218] [version] parsed project prefs setting 'gpu_util_fgrp': 1.000000
    2017-05-09 06:48:43.0455 [PID=17218] [version] OpenCL GPU RAM required min: 1048576000.000000, supplied: 536870912
    2017-05-09 06:48:43.0455 [PID=17218] [version] Checking plan class 'FGRPopencl1K-nvidia'
    2017-05-09 06:48:43.0455 [PID=17218] [version] parsed project prefs setting 'gpu_util_fgrp': 1.000000
    2017-05-09 06:48:43.0455 [PID=17218] [version] No CUDA devices found
    2017-05-09 06:48:43.0455 [PID=17218] [version] no app version available: APP#40 (hsgamma_FGRPB1G) PLATFORM#7 (x86_64-pc-linux-gnu) min_version 0
    2017-05-09 06:48:43.0455 [PID=17218] [version] no app version available: APP#40 (hsgamma_FGRPB1G) PLATFORM#1 (i686-pc-linux-gnu) min_version 0
    2017-05-09 06:48:43.0522 [PID=17218] [debug] [HOST#12528613] MSG(high) No work sent
    2017-05-09 06:48:43.0523 [PID=17218] [debug] [HOST#12528613] MSG(high) Binary Radio Pulsar Search (Arecibo) is not available for your type of computer.
    2017-05-09 06:48:43.0523 [PID=17218] [debug] [HOST#12528613] MSG(high) see scheduler log messages on https://einsteinathome.org/host/12528613/log
    2017-05-09 06:48:43.0523 [PID=17218] Sending reply to [HOST#12528613]: 0 results, delay req 60.00
    2017-05-09 06:48:43.0524 [PID=17218] Scheduler ran 0.068 seconds

  3. #43
    Join Date
    Sep 2010
    Location
    Leiden, the Netherlands
    Posts
    4,372
    Quote Originally Posted by Tweeder View Post
    We seem to be making up some ground on the City Run and should be able to catch the next two spots before that event ends. I tried to get a head start on the Einstein project, but this ended up being me waiting for any of them to finish:
    Attachment 360

    I have never done any Einstein work before, but they seem to be taking much longer than advertised to finish and take up some memory. I have not started trying to bank the LHC work units yet. I think that will be my next move after city run closes depending on if it looks like we need some work on the marathon project or not.
    How long is long here? On the IGP of my Athlon 5350 it takes 5 hours, the HD4000 of my i7-3770 takes a bit over one hour (much smaller WU's, less points).
    How do you crunch Einstein WUs, by CPU or GPU? If by GPU, what GPU are you using?

    Edit: I peeked via our group page on Einstein. You use a nVidia GeForce GT610 and on another box an Intel HD5000.
    You may want to change some settings here: BOINC runs always and, while in use, BOINC may run and use the GPU. It looks like Einstein isn't getting enough resources. You may want to look in the taskmanager to see what's going on. Have you installed the lastest Intel OpenCL drivers for the i5 box? Can you expand the 4GB RAM size from that NUC-like box to at least 8GB?
    Last edited by Dirk Broer; 05-09-2017 at 12:44 PM.


  4. #44
    Join Date
    Sep 2010
    Location
    Leiden, the Netherlands
    Posts
    4,372
    Quote Originally Posted by SuicideCabbage View Post
    Mon 08 May 2017 10:53:18 PM EDT | | OpenCL: AMD/ATI GPU 0: Cypress (driver version 1800.11, device version OpenCL 1.2 AMD-APP (1800.11), 512MB, 512MB available, 896 GFLOPS peak)
    Mon 08 May 2017 10:53:18 PM EDT | | OpenCL: AMD/ATI GPU 1: Cypress (driver version 1800.11, device version OpenCL 1.2 AMD-APP (1800.11), 512MB, 512MB available, 896 GFLOPS peak)
    And the Catalyst Control Center reports 1 GB for both cards? Note that video RAM can't be shared. A dual GPU, e.g. a HD 7990 with 6 GB of RAM will show as two Tahiti GPUs with 3 GB each.

    Do you have the possibility to install an older driver, preferably one that is still CAL-capable?
    look at the GFLOPS:
    06-Feb-2013 20:47:52 [---] CAL: ATI GPU 0: ATI Radeon HD 5800 series (Cypress) (CAL version 1.4.1741, 1024MB, 986MB available, 3584 GFLOPS peak)
    06-Feb-2013 20:47:52 [---] OpenCL: ATI GPU 0: ATI Radeon HD 5800 series (Cypress) (driver version 1016.4, device version OpenCL 1.2 AMD-APP (1016.4), 1024MB, 986MB available, 3584 GFLOPS peak)
    Last edited by Dirk Broer; 05-10-2017 at 06:47 AM.


  5. #45
    Join Date
    Apr 2017
    Location
    Texas
    Posts
    18
    Quote Originally Posted by Dirk Broer View Post
    ...lastest Intel OpenCL drivers for the i5 box? ...
    I was running CPU and GPU on both, but backed off to just GPU so that I could get a mix of projects on the Linux box.

    I am having troubles finding the latest drivers because it is on Mac OS X El Capatain and I cant seem to find an easy way to check if I have the most updated. Any pointers there would be great because right now it looks like the GPU is being found, but I am getting a message that einstein does not support that type of computer.

    2017-05-10 01:02:31.8320 [PID=27365] [debug] [HOST#12526340] MSG(high) Binary Radio Pulsar Search (Arecibo) is not available for your type of computer.
    2017-05-10 01:02:31.8320 [PID=27365] [debug] [HOST#12526340] MSG(high) see scheduler log messages on https://einsteinathome.org/host/12526340/log

  6. #46
    Join Date
    Sep 2010
    Location
    Leiden, the Netherlands
    Posts
    4,372
    Apart from Binary Radio Pulsar Search (Arecibo), all Einstein@Home applications have an OS X variant, all Mac OS 10.5 or later, running on Intel 64 Bit.
    Binary Radio Pulsar Search (Arecibo) will only run, OS X-wise, on PPC architecture (e.g. PowerMac G3, G4 and G5)


  7. #47
    Join Date
    Feb 2017
    Location
    Yorkshire, England, Europe
    Posts
    45
    Quote Originally Posted by SuicideCabbage View Post
    ... the 5830s are a 1GB card, in BOINC they are reporting 512MB ...
    I had the same problem with my 2Gb HD6670 card, being recognised as a 512Mb one. It was the default Linux drivers, I installed FGLRX and switched to that to cure it.
    What version of Ubuntu are you using, 16 by any chance? If so, you will find FGLRX isn't supported any more, you'll need to use 14 instead. I'm using Mint 17.3 on my Linux box with the HD6670 in it, because that is based on Ubuntu 14 and works. The latest version of Mint (18) is based on Ubuntu 16 and will have the same problem too. Do a Google/Bing/whatever search for FGLRX and Ubuntu 16 - there are a lot of unhappy graphics card users!

  8. #48
    Join Date
    Feb 2017
    Location
    Yorkshire, England, Europe
    Posts
    45
    Quote Originally Posted by Tweeder View Post
    ...seem to be taking much longer than advertised to finish ...
    I don't know what they're advertised at, but on both my AMD machines my CPU tasks are taking around 12 to 14 hours to complete, and my GPU rattles them off at just under 3 hours each.

  9. #49
    KarmannGaz, I was running 14.04.5 because I knew about the issue with 16 before I deployed. However from research during the initial issues, I found someone make mention that support was lost after 14.04.1 and sure enough, when I deployed that everything was happy on the 6970 and BOINC at least saw the cards on the 5830. Under 14.04.5 the drivers appeared to install, and function, but I guess not fully as OpenCL wouldn't work, at least I couldn't get it to work. As far as the 5830s go, I purged and reinstalled the drivers last night and BOINC recognizes them with correct memory and also as CAL devices, however everything is tossing computational errors almost instantly. The only difference between the 6970 and the 5830 machine as far as operating system is the desktop environment. Both run MSI boards, but the processor as well as chipset on the 5830 machine is newer.

    For now I am just going to hold as I can't afford to lose anymore computational time, but I will attempt to get back to 14.04.5 after the pentathlon is over. I have found a few things broken under .1 that worked without issue in .5, and I also miss the MATE environment. If nothing else I will install .1 as a headless setup then install MATE so I get a cleaner install rather than putting MATE on over another environment. Xfce is nice, and I could get use to it if I wasn't already fond of MATE, Unity, the default Ubuntu environment, is a train wreck. The utilities that ship with MATE are much more what I am looking for, even simple things like media players and file managers have greater functionality in the way I desire under MATE. I know I could get all that on Xfce, but one and done seems cleaner. That may just be old-world thinking though (Windows) where installing and uninstalling even simple stuff chips away at the feeble armor of the OS.

    EDIT: I don't run Einstein on desktop CPU, haven't for years, my phone takes about 24h to complete a unit on it's CPU. The 6970 GPU is knocking out a unit every hour with app_config set to run 0.5 CPU 1 GPU. If memory serves, it was about an hour and 40-50m per unit on the 5830 using the same setup.
    Last edited by SuicideCabbage; 05-11-2017 at 03:31 AM.

  10. #50
    Join Date
    Sep 2010
    Location
    Leiden, the Netherlands
    Posts
    4,372
    You computational errors:
    Using OpenCL platform provided by: Advanced Micro Devices, Inc.
    Using OpenCL device "Cypress" by: Advanced Micro Devices, Inc.
    Max allocation limit: 536870912
    Global mem size: 536870912
    low_memory_configuration not supported
    initialize_ocl returned error [2013]
    OCL context null
    OCL queue null
    Error generating generic FFT context object [5]
    00:09:30 (3123): [CRITICAL]: ERROR: MAIN() returned with error '5'

    In your Boinc Program Data directory there is a file called coproc_info.xml, for my HD 4770 under Win10 it gives:
    Code:
        <coprocs>
    <coproc_ati>
       <count>0</count>
       <name>ATI Radeon HD 4700/4800 (RV740/RV770)</name>
       <available_ram>503316480.000000</available_ram>
       <have_cal>1</have_cal>
       <have_opencl>0</have_opencl>
       <peak_flops>1920000000000.000000</peak_flops>
       <CALVersion>1.4.1734</CALVersion>
       <target>5</target>
       <localRAM>512</localRAM>
       <uncachedRemoteRAM>2047</uncachedRemoteRAM>
       <cachedRemoteRAM>2047</cachedRemoteRAM>
       <engineClock>750</engineClock>
       <memoryClock>800</memoryClock>
       <wavefrontSize>64</wavefrontSize>
       <numberOfSIMD>8</numberOfSIMD>
       <doublePrecision>1</doublePrecision>
       <pitch_alignment>256</pitch_alignment>
       <surface_alignment>4096</surface_alignment>
       <maxResource1DWidth>8192</maxResource1DWidth>
       <maxResource2DWidth>8192</maxResource2DWidth>
       <maxResource2DHeight>8192</maxResource2DHeight>
        <atirt_detected/>
    </coproc_ati>
    <warning>No NVIDIA library found</warning>
    <warning>No OpenCL library found</warning>
        </coprocs>
    Perhaps you can edit your coproc_info.xml to give your HD 5830 more RAM available

    Your HD 6970 has no listed tasks with Einstein (last contact April 23th), but you do not need an app_config.xml with Einstein to crunch two WUs at at time, you do it in the projects preferences
    account - preferences - project - other settings
    Set the settings of choice to 0.50 instead of the default 1.00 for two WUs at a time or to 0.33 for three WUs at a time (on one GPU)
    Last edited by Dirk Broer; 05-11-2017 at 11:24 AM.


Page 5 of 9 FirstFirst 123456789 LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •