Not utilising all my 6 GPUs

Message boards : GPUs : Not utilising all my 6 GPUs
Message board moderation

To post messages, you must log in.

AuthorMessage
Svetoslav Nikolov

Send message
Joined: 25 Feb 17
Posts: 4
Bulgaria
Message 76135 - Posted: 25 Feb 2017, 14:27:37 UTC

Greetings to all,

I have 6 GPUs on my computer. I want to use all of the in the BOINC software but only 3 of them are working and doing jobs. Any ideas how to enable the other 3 to work.

Thank you.
ID: 76135 · Report as offensive
Profile Jord
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 29 Aug 05
Posts: 15480
Netherlands
Message 76136 - Posted: 25 Feb 2017, 14:46:53 UTC - in response to Message 76135.  

/me shakes his crystal ball... nope.
So could you give some basic information please?
ID: 76136 · Report as offensive
Svetoslav Nikolov

Send message
Joined: 25 Feb 17
Posts: 4
Bulgaria
Message 76137 - Posted: 25 Feb 2017, 14:58:19 UTC - in response to Message 76136.  

Using latest version of BOINC
Windows 10
Project: einstain@home with account http://pool.gridcoin.co , team GridCoin

2/25/2017 4:51:24 PM | | cc_config.xml not found - using defaults
2/25/2017 4:51:24 PM | | Starting BOINC client version 7.6.33 for windows_x86_64
2/25/2017 4:51:24 PM | | log flags: file_xfer, sched_ops, task
2/25/2017 4:51:24 PM | | Libraries: libcurl/7.47.1 OpenSSL/1.0.2g zlib/1.2.8
2/25/2017 4:51:24 PM | | Data directory: C:\ProgramData\BOINC
2/25/2017 4:51:24 PM | | Running under account RX470
2/25/2017 4:51:26 PM | | OpenCL: AMD/ATI GPU 0: Radeon (TM) RX 470 Graphics (driver version 2236.10, device version OpenCL 2.0 AMD-APP (2236.10), 4096MB, 4096MB available, 5202 GFLOPS peak)
2/25/2017 4:51:26 PM | | OpenCL: AMD/ATI GPU 1: Radeon (TM) RX 470 Graphics (driver version 2236.10, device version OpenCL 2.0 AMD-APP (2236.10), 4096MB, 4096MB available, 5202 GFLOPS peak)
2/25/2017 4:51:26 PM | | OpenCL: AMD/ATI GPU 2: Radeon (TM) RX 470 Graphics (driver version 2236.10, device version OpenCL 2.0 AMD-APP (2236.10), 4096MB, 4096MB available, 5202 GFLOPS peak)
2/25/2017 4:51:26 PM | | OpenCL: AMD/ATI GPU 3: Radeon (TM) RX 470 Graphics (driver version 2236.10, device version OpenCL 2.0 AMD-APP (2236.10), 4096MB, 4096MB available, 5202 GFLOPS peak)
2/25/2017 4:51:26 PM | | OpenCL: AMD/ATI GPU 4: Radeon (TM) RX 470 Graphics (driver version 2236.10, device version OpenCL 2.0 AMD-APP (2236.10), 4096MB, 4096MB available, 5202 GFLOPS peak)
2/25/2017 4:51:26 PM | | OpenCL: AMD/ATI GPU 5: Radeon (TM) RX 470 Graphics (driver version 2236.10, device version OpenCL 2.0 AMD-APP (2236.10), 4096MB, 4096MB available, 5202 GFLOPS peak)
2/25/2017 4:51:26 PM | | Host name: DESKTOP-8RSSR24
2/25/2017 4:51:26 PM | | Processor: 2 GenuineIntel Intel(R) Celeron(R) CPU G1840 @ 2.80GHz [Family 6 Model 60 Stepping 3]
2/25/2017 4:51:26 PM | | 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 cx16 sse4_1 sse4_2 movebe popcnt rdrandsyscall nx lm vmx tm2 pbe fsgsbase
2/25/2017 4:51:26 PM | | OS: Microsoft Windows 10: Professional x64 Edition, (10.00.14393.00)
2/25/2017 4:51:26 PM | | Memory: 7.93 GB physical, 39.18 GB virtual
2/25/2017 4:51:26 PM | | Disk: 118.75 GB total, 58.92 GB free
2/25/2017 4:51:26 PM | | Local time is UTC +2 hours
ID: 76137 · Report as offensive
Juha
Volunteer developer
Volunteer tester
Help desk expert

Send message
Joined: 20 Nov 12
Posts: 801
Finland
Message 76139 - Posted: 25 Feb 2017, 17:03:04 UTC - in response to Message 76137.  

If Einstein is the only project you run you are probably running out of CPU cores. Currently the only GPU app on Einstein is FGRPB and I believe it is set up to reserve one CPU core per task. So you'd need to have five or six CPU cores to use all GPUs (BOINC appears to overcommit CPU up to one core).

You can set up app_config.xml to reserve less than one CPU core per task though having only two CPU cores available for six GPUs may be less than optimal.
ID: 76139 · Report as offensive
Svetoslav Nikolov

Send message
Joined: 25 Feb 17
Posts: 4
Bulgaria
Message 76140 - Posted: 25 Feb 2017, 17:08:20 UTC - in response to Message 76139.  

Can you recommend projects that are not bind to CPU.

Also, where is the app_config.xml situated.

Thanks.
ID: 76140 · Report as offensive
Juha
Volunteer developer
Volunteer tester
Help desk expert

Send message
Joined: 20 Nov 12
Posts: 801
Finland
Message 76143 - Posted: 25 Feb 2017, 21:10:15 UTC - in response to Message 76140.  

I don't have an AMD GPU and I haven't joined all that many projects so I'd rather not recommend anything. Also, while some applications doesn't seem to require full CPU core they may be slowed down badly if there isn't a free CPU core available.

app_config.xml is placed in the directory of the project you want to configure. In this case projects\einstein.phys.uwm.edu and the projects directory is in BOINC's data directory, usually C:\ProgramData\BOINC . app_config.xml is a file that is not present by default but created by users.

This should reserve 0.1 CPU cores per task:

<app_config>
  <app>
    <name>hsgamma_FGRPB1G</name>
    <gpu_versions>
        <gpu_usage>1</gpu_usage>
        <cpu_usage>0.1</cpu_usage>
    </gpu_versions>
  </app>
</app_config>


You may benefit from running two or three tasks per GPU which can be done by changing <gpu_usage> above. The people at Einstein forums know their applications best so if you have further tuning related question you may want to try their forums.
ID: 76143 · Report as offensive
Svetoslav Nikolov

Send message
Joined: 25 Feb 17
Posts: 4
Bulgaria
Message 76144 - Posted: 25 Feb 2017, 21:33:19 UTC - in response to Message 76143.  

Thank you very much. All GPUs are working now with the following config of the xml file:

<app_config>
<app>
<name>hsgamma_FGRPB1G</name>
<gpu_versions>
<gpu_usage>1</gpu_usage>
<cpu_usage>0.1</cpu_usage>
</gpu_versions>
</app>
</app_config>

[/img]
ID: 76144 · Report as offensive

Message boards : GPUs : Not utilising all my 6 GPUs

Copyright © 2024 University of California.
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.