BOINC using more GPU's WU that I have set
23 Mar 2017 15:00:19 UTC
Topic 206341
(moderation:
Hello, friends.
I've been using Boinc for almost a year and tried several different settings so far.
Now I'm using a Lenovo Yoga 510 as my main system. It has 2 gpus (intel and an AMD).
But something went wrong. After some weeks, Boinc started to use 2 WU for each GPU(4 GPU taks at once).
I've set BRP/FGRP/GW to 1 and it's still running 2 WU. I tried resetting the project and it didn't work.
I could have uninstalled Boinc and installed it again, but I think I'll reach its daily quota if I do it today.
I need some help.
Andre Cosenza wrote: I've
)
Did you try this one:
Boinc Manager - settings - store at least ... days
Increase the number of days a little bit, say by 0,5
That will force Boinc to download a few new workunits. After that, your settings BRP/FGRP/GW to 1 will apply for both old and new WU. Keep in mind that settings (BRP/FGRP/GW to 1) are tied to the application.
Then set back the store at least ... days number.
Hope that helps in this case.
Thank you for answering
)
Thank you for answering me.
I did this now, but didn't work.
Looking at the events I saw something.
23/03/2017 13:41:48 | | Starting BOINC client version 7.6.33 for windows_x86_64
23/03/2017 13:41:48 | | log flags: file_xfer, sched_ops, task
23/03/2017 13:41:48 | | Libraries: libcurl/7.47.1 OpenSSL/1.0.2g zlib/1.2.8
23/03/2017 13:41:48 | | Data directory: C:\ProgramData\BOINC
23/03/2017 13:41:48 | | Running under account Lenovo
23/03/2017 13:41:49 | | OpenCL: AMD/ATI GPU 0: AMD Radeon(TM) R5 M430 (driver version 2117.14 (VM), device version OpenCL 1.2 AMD-APP (2117.14), 2048MB, 2048MB available, 659 GFLOPS peak)
23/03/2017 13:41:49 | | OpenCL: AMD/ATI GPU 1: AMD Radeon(TM) R5 M430 (driver version 2117.14 (VM), device version OpenCL 1.2 AMD-APP (2117.14), 2048MB, 2048MB available, 659 GFLOPS peak)
23/03/2017 13:41:49 | | OpenCL: Intel GPU 0: Intel(R) HD Graphics 520 (driver version 21.20.16.4550, device version OpenCL 2.0, 3222MB, 3222MB available, 202 GFLOPS peak)
23/03/2017 13:41:49 | | OpenCL: Intel GPU 1: Intel(R) HD Graphics 520 (driver version 21.20.16.4550, device version OpenCL 2.0, 3222MB, 3222MB available, 202 GFLOPS peak)
23/03/2017 13:41:49 | | OpenCL CPU: Intel(R) Core(TM) i7-6500U CPU @ 2.50GHz (OpenCL driver vendor: Intel(R) Corporation, driver version 6.8.0.392, device version OpenCL 2.0 (Build 392))
23/03/2017 13:41:49 | | OpenCL CPU: Intel(R) Core(TM) i7-6500U CPU @ 2.50GHz (OpenCL driver vendor: Intel(R) Corporation, driver version 6.8.0.392, device version OpenCL 2.0 (Build 392))
23/03/2017 13:41:49 | | Host name: DESKTOP-NVB1ONA
23/03/2017 13:41:49 | | Processor: 4 GenuineIntel Intel(R) Core(TM) i7-6500U CPU @ 2.50GHz [Family 6 Model 78 Stepping 3]
23/03/2017 13:41:49 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 fma cx16 sse4_1 sse4_2 movebe popcnt aes f16c rdrandsyscall nx lm avx avx2 vmx tm2 pbe fsgsbase bmi1 smep bmi2
23/03/2017 13:41:49 | | OS: Microsoft Windows 10: Professional x64 Edition, (10.00.14393.00)
23/03/2017 13:41:49 | | Memory: 7.89 GB physical, 15.89 GB virtual
23/03/2017 13:41:49 | | Disk: 194.44 GB total, 150.91 GB free
23/03/2017 13:41:49 | | Local time is UTC -3 hours
23/03/2017 13:41:49 | | VirtualBox version: 5.0.18
23/03/2017 13:41:49 | Einstein@Home | URL http://einstein.phys.uwm.edu/; Computer ID 12496347; resource share 100
23/03/2017 13:41:49 | Einstein@Home | General prefs: from Einstein@Home (last modified 25-Jul-2016 17:15:52)
23/03/2017 13:41:49 | Einstein@Home | Host location: none
23/03/2017 13:41:49 | Einstein@Home | General prefs: using your defaults
23/03/2017 13:41:49 | | Reading preferences override file
23/03/2017 13:41:49 | | Preferences:
23/03/2017 13:41:49 | | max memory usage when active: 4037.88MB
23/03/2017 13:41:49 | | max memory usage when idle: 7268.18MB
23/03/2017 13:41:49 | | max disk usage: 97.22GB
23/03/2017 13:41:49 | | max CPUs used: 3
23/03/2017 13:41:49 | | (to change preferences, visit a project web site or select Preferences in the Manager)
For some reason, the app thinks I have four physical GPU.
Andre Cosenza wrote: For
)
.... and two CPUs, it seems? Strange! I haven't got a solution but I wonder what you need a virtual machine for as it isn't necessary for Einstein@home but for some Cern apps only.
Andre Cosenza wrote:Hello,
)
Go onto the website under Preferences, Project and you will see this:
mikey wrote:I think Einstein
)
There's nothing funky about Einstein compared to other projects with regards to "normal" preferences/settings, it works the same way as any other project. The only thing different is that Einstein provides an additional setting under "other settings" for controlling "GPU utilization factor" for the different GPU apps available.
This setting only take affect when downloading new tasks after the setting has been modified and then applies to all tasks currently in the cache, even though Boinc might still show the previous setting.
There is no need to finish all existing tasks before the new setting takes effect and there is absolutely no need to abort all tasks in the cache.
Holmis wrote:mikey wrote:I
)
Hmm I did not know that it applied to ALL tasks even those already in the queue...THANKS Holmis for that info!!
The settings from the website
)
The settings from the website only pertain to new WUs downloaded AFTER the change. Only updating an app_config file will update the settings for all WUs in queue.
"The settings from the
)
The web site should state that, currently it doesn't.
mmonnin wrote:The settings
)
Assuming in discussing "settings form the website" are are discussing the Einstein project preference items titled "GPU utilization factor" this is only partially correct, and it is not the interesting part.
The "message" from web site to your PC is carried to the PC at the time of download of new GPU work after the setting change, and so far as actual processing of work is concerned it takes effect immediately.
For example, if your previous setting was 0.5, and thus you were running two of that type of GPU tasks, and the new setting is 0.33, immediately on download BOINC on the PC will start an additional GPU task, so now three are running.
Confusingly enough, there is some sort of "branding" label, which is displayed by boincmgr in the status column as, for example, "Ready to Start (1 CPU + 0.5 NVIDIA GPUSs)" or in the Use column of BOINCTasks as "1 CPU + 0.5 NVIDIA GPUs". This text string does NOT get updated, which has the effect that you can see such labelling on in-process work clearly not coordinated with the actual processing.
But the assertion that one must update an app_config in order to change the multiplicity under which already downloaded work is processed is false.
I uninstalled and installed
)
I uninstalled and installed Boinc again. I deleted the program data folder and downloaded new tasks again.
But boinc still thinks I have 2 AMD gpu, 2 Intel GPU and 2 intel CPU.
It wouldn't be a problem if I had 8 threads or more, but my laptop only have 4, thus I can run only 1 cpu task, for the GPUs takes 3 threads to run(1+1 for AMD and 0.5+0.5 for intel).