Need a Little Help With MB CUDA Settings in Linux

Message boards : Number crunching : Need a Little Help With MB CUDA Settings in Linux
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Fawkesguy
Volunteer tester
Avatar

Send message
Joined: 8 Jan 01
Posts: 108
Credit: 188,578,766
RAC: 0
United States
Message 1728802 - Posted: 25 Sep 2015, 14:05:11 UTC

Hi everyone,

I need a little help figuring out command switches for adjusting MB CUDA settings. In Windows, I would use mbcuda.cfg and set something like this:

processpriority = abovenormal
pfblockspersm = 16
pfperiodsperlaunch = 400

On one of my Linux hosts, I tried command switches in app_info.xml, but it didn't work. Stderr is showing:

bad arg: -abo
bad arg: -pfb
bad arg: 16
bad arg: -pfp
bad arg: 400

Example WU: http://setiathome.berkeley.edu/result.php?resultid=4405418321

Please take a look at how I tried setting this in my app_info.xml below. Maybe the switches are in the wrong place? Also, is there a list of switches I can reference? I wanted to set priority to "high", but only found "-abo" which I assume is "above normal", from searching the forums.

<app>
<name>setiathome_v7</name>
</app>

<file_info>
<name>setiathome_x41zc_x86_64-pc-linux-gnu_cuda60</name>
<executable/>
</file_info>

<file_info>
<name>libcudart.so.6.0</name>
<executable/>
</file_info>

<file_info>
<name>libcufft.so.6.0</name>
<executable/>
</file_info>

<app_version>
<app_name>setiathome_v7</app_name>
<version_num>704</version_num>
<plan_class>cuda60</plan_class>
<avg_ncpus>0.15</avg_ncpus>
<max_ncpus>0.30</max_ncpus>
<cmdline>-abo -pfb 16 -pfp 400</cmdline>

<coproc>
<type>CUDA</type>
<count>0.5</count>
</coproc>

<file_ref>
<file_name>setiathome_x41zc_x86_64-pc-linux-gnu_cuda60</file_name>
<main_program/>
</file_ref>

<file_ref>
<file_name>libcudart.so.6.0</file_name>
</file_ref>

<file_ref>
<file_name>libcufft.so.6.0</file_name>
</file_ref>
</app_version>
ID: 1728802 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 1728813 - Posted: 25 Sep 2015, 15:01:02 UTC - in response to Message 1728802.  
Last modified: 25 Sep 2015, 15:03:39 UTC

Hi FawkesGuy,
For Linux builds the graphic driver model is considerably different, enough that those settings don't make sense, so are not used (and the mbcuda.cfg in fact is never read there or on Mac.

About the only possible setting that would apply at this time, is some priority setting capability, but that is also too different to use the mechanism, and automated re-nicing should be used if desired.

As we move to a more unified and automated development system, some of these things will either change entirely in versions to come, be removed and replaced with better more general mechanisms, and/or the (Windows) problems those settings address (not that well) avoided by engineering a completely different mechanism of action.
"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 1728813 · Report as offensive
Profile Fawkesguy
Volunteer tester
Avatar

Send message
Joined: 8 Jan 01
Posts: 108
Credit: 188,578,766
RAC: 0
United States
Message 1728819 - Posted: 25 Sep 2015, 16:09:47 UTC

Thank you Jason. :-)

I'm very new to crunching on Linux, so there is a big learning curve for me.
ID: 1728819 · Report as offensive
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 5204
Credit: 840,779,836
RAC: 2,768
United States
Message 1728823 - Posted: 25 Sep 2015, 16:24:59 UTC - in response to Message 1728819.  

For setting the priority the easiest way is to just use;
<no_priority_change>1</no_priority_change>
in cc_config.xml. That applies to all tasks, but it's better than nothing and does work.
ID: 1728823 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 1728836 - Posted: 25 Sep 2015, 16:52:50 UTC - in response to Message 1728823.  

Ooh, I can use that tip too, thanks for that :) App side will definitely have to work out something that can be at least somewhat consistant across platforms. probably won't happen short term, but will probably factor into next major design change.
"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 1728836 · Report as offensive
Profile Fawkesguy
Volunteer tester
Avatar

Send message
Joined: 8 Jan 01
Posts: 108
Credit: 188,578,766
RAC: 0
United States
Message 1728893 - Posted: 25 Sep 2015, 20:22:54 UTC

Thanks TBar, I'll give that a shot.
ID: 1728893 · Report as offensive
Profile Fawkesguy
Volunteer tester
Avatar

Send message
Joined: 8 Jan 01
Posts: 108
Credit: 188,578,766
RAC: 0
United States
Message 1728896 - Posted: 25 Sep 2015, 20:36:04 UTC

I looked at the explanation for that command on the BOINC wiki (https://boinc.berkeley.edu/wiki/Client_configuration) and I want to make sure I understand its usage.

<no_priority_change>0|1</no_priority_change>
If 1, don't change priority of applications (run them at same priority as client).
NB: This option can, if activated, impact system responsiveness for the user. Default, all CPU science apps run at lowest (idle) priority Nice 15.


So if I set it to 1, it runs at a lower priority than if I set it to 0?
ID: 1728896 · Report as offensive
Profile Fawkesguy
Volunteer tester
Avatar

Send message
Joined: 8 Jan 01
Posts: 108
Credit: 188,578,766
RAC: 0
United States
Message 1728905 - Posted: 25 Sep 2015, 21:21:52 UTC

Hmmm, this is odd. BOINC for some reason can't see my cc_config.xml. It's definitely there. I even tried deleting it and creating a fresh one. Fails both on reboot and "Read config files" Any idea what might cause this?


Fri 25 Sep 2015 05:00:32 PM EDT | | cc_config.xml not found - using defaults
Fri 25 Sep 2015 05:00:32 PM EDT | | Starting BOINC client version 7.2.42 for x86_64-pc-linux-gnu
Fri 25 Sep 2015 05:00:32 PM EDT | | log flags: file_xfer, sched_ops, task
Fri 25 Sep 2015 05:00:32 PM EDT | | Libraries: libcurl/7.35.0 OpenSSL/1.0.1f zlib/1.2.8 libidn/1.28 librtmp/2.3
Fri 25 Sep 2015 05:00:32 PM EDT | | Data directory: /home/********/BOINC
Fri 25 Sep 2015 05:00:32 PM EDT | | CUDA: NVIDIA GPU 0: GeForce GTX 980 (driver version unknown, CUDA version 7.5, compute capability 5.2, 4096MB, 4004MB available, 8399 GFLOPS peak)
Fri 25 Sep 2015 05:00:32 PM EDT | | CUDA: NVIDIA GPU 1: GeForce GTX 980 (driver version unknown, CUDA version 7.5, compute capability 5.2, 4087MB, 3863MB available, 8399 GFLOPS peak)
Fri 25 Sep 2015 05:00:32 PM EDT | | OpenCL: NVIDIA GPU 0: GeForce GTX 980 (driver version 355.11, device version OpenCL 1.2 CUDA, 4096MB, 4004MB available, 8399 GFLOPS peak)
Fri 25 Sep 2015 05:00:32 PM EDT | | OpenCL: NVIDIA GPU 1: GeForce GTX 980 (driver version 355.11, device version OpenCL 1.2 CUDA, 4087MB, 3863MB available, 8399 GFLOPS peak)
Fri 25 Sep 2015 05:00:32 PM EDT | SETI@home | Found app_info.xml; using anonymous platform
Fri 25 Sep 2015 05:00:32 PM EDT | | Host name: *********
Fri 25 Sep 2015 05:00:32 PM EDT | | Processor: 6 GenuineIntel Intel(R) Core(TM) i7-4930K CPU @ 3.40GHz [Family 6 Model 62 Stepping 4]
Fri 25 Sep 2015 05:00:32 PM EDT | | 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 ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf eagerfpu pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm ida arat epb xsaveopt pln pts dtherm tpr_shadow vnmi flexpriority ept vpid fsgsbase smep erms
Fri 25 Sep 2015 05:00:32 PM EDT | | OS: Linux: 3.16.0-38-generic
Fri 25 Sep 2015 05:00:32 PM EDT | | Memory: 15.61 GB physical, 1.33 GB virtual
Fri 25 Sep 2015 05:00:32 PM EDT | | Disk: 202.47 GB total, 185.51 GB free
Fri 25 Sep 2015 05:00:32 PM EDT | | Local time is UTC -4 hours
Fri 25 Sep 2015 05:00:32 PM EDT | SETI@home | Found app_config.xml
Fri 25 Sep 2015 05:00:32 PM EDT | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 7771865; resource share 100
Fri 25 Sep 2015 05:00:32 PM EDT | SETI@home | General prefs: from SETI@home (last modified 19-Jul-2015 22:33:20)
Fri 25 Sep 2015 05:00:32 PM EDT | SETI@home | Host location: none
Fri 25 Sep 2015 05:00:32 PM EDT | SETI@home | General prefs: using your defaults
Fri 25 Sep 2015 05:00:32 PM EDT | | Preferences:
Fri 25 Sep 2015 05:00:32 PM EDT | | max memory usage when active: 7992.40MB
Fri 25 Sep 2015 05:00:32 PM EDT | | max memory usage when idle: 14386.31MB
Fri 25 Sep 2015 05:00:32 PM EDT | | max disk usage: 0.00GB
Fri 25 Sep 2015 05:00:32 PM EDT | | don't compute while active
Fri 25 Sep 2015 05:00:32 PM EDT | | don't use GPU while active
Fri 25 Sep 2015 05:00:32 PM EDT | | (to change preferences, visit a project web site or select Preferences in the Manager)
Fri 25 Sep 2015 05:00:32 PM EDT | | Not using a proxy
Fri 25 Sep 2015 05:00:32 PM EDT | | Suspending computation - computer is in use
Fri 25 Sep 2015 05:02:32 PM EDT | SETI@home | update requested by user
Fri 25 Sep 2015 05:02:33 PM EDT | SETI@home | Sending scheduler request: Requested by user.
Fri 25 Sep 2015 05:02:33 PM EDT | SETI@home | Reporting 1 completed tasks
Fri 25 Sep 2015 05:02:33 PM EDT | SETI@home | Requesting new tasks for NVIDIA
Fri 25 Sep 2015 05:02:35 PM EDT | SETI@home | Scheduler request completed: got 1 new tasks
Fri 25 Sep 2015 05:02:37 PM EDT | SETI@home | Started download of 22my11ab.24355.301260.438086664197.12.237
Fri 25 Sep 2015 05:02:39 PM EDT | SETI@home | Finished download of 22my11ab.24355.301260.438086664197.12.237
Fri 25 Sep 2015 05:09:26 PM EDT | | Re-reading cc_config.xml
Fri 25 Sep 2015 05:09:26 PM EDT | | cc_config.xml not found - using defaults
Fri 25 Sep 2015 05:09:26 PM EDT | | log flags: file_xfer, sched_ops, task
Fri 25 Sep 2015 05:09:26 PM EDT | SETI@home | Found app_config.xml
Fri 25 Sep 2015 05:14:55 PM EDT | | Re-reading cc_config.xml
Fri 25 Sep 2015 05:14:55 PM EDT | | cc_config.xml not found - using defaults
Fri 25 Sep 2015 05:14:55 PM EDT | | log flags: file_xfer, sched_ops, task
Fri 25 Sep 2015 05:14:55 PM EDT | SETI@home | Found app_config.xml
ID: 1728905 · Report as offensive
Profile Zalster Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 27 May 99
Posts: 5517
Credit: 528,817,460
RAC: 242
United States
Message 1728908 - Posted: 25 Sep 2015, 21:42:07 UTC - in response to Message 1728905.  
Last modified: 25 Sep 2015, 21:42:29 UTC

did you place the cc_config.xml in the BOINC folder and not the Setiathome folder?
ID: 1728908 · Report as offensive
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 5204
Credit: 840,779,836
RAC: 2,768
United States
Message 1728909 - Posted: 25 Sep 2015, 21:43:11 UTC - in response to Message 1728905.  

If the file is not the correct format for xml BOINC will ignore it and say it's not there. The easiest way to create a BOINC friendly xml is to duplicate a BOINC generated xml and just replace the contents and rename it cc_config.xml. Once you get it running All the tasks will run at the Same priority as BOINC, which is nice 0 instead of the lower priority. Hence the warning about sluggishness. Look at the before and after in Top.
ID: 1728909 · Report as offensive
Profile Fawkesguy
Volunteer tester
Avatar

Send message
Joined: 8 Jan 01
Posts: 108
Credit: 188,578,766
RAC: 0
United States
Message 1728910 - Posted: 25 Sep 2015, 21:44:42 UTC - in response to Message 1728908.  

It's sitting in setiathome.berkeley.edu, right next to app_info and app_config, which BOINC has no problem finding.
ID: 1728910 · Report as offensive
Profile Fawkesguy
Volunteer tester
Avatar

Send message
Joined: 8 Jan 01
Posts: 108
Credit: 188,578,766
RAC: 0
United States
Message 1728912 - Posted: 25 Sep 2015, 21:52:32 UTC - in response to Message 1728909.  

I did what you suggested with client_state.xml and got the same result. Very strange that is has no issues with app_info and app_config, both of which are fresh files, not copied from elsewhere, same as cc_config


If the file is not the correct format for xml BOINC will ignore it and say it's not there. The easiest way to create a BOINC friendly xml is to duplicate a BOINC generated xml and just replace the contents and rename it cc_config.xml. Once you get it running All the tasks will run at the Same priority as BOINC, which is nice 0 instead of the lower priority. Hence the warning about sluggishness. Look at the before and after in Top.
ID: 1728912 · Report as offensive
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 5204
Credit: 840,779,836
RAC: 2,768
United States
Message 1728919 - Posted: 25 Sep 2015, 22:08:25 UTC - in response to Message 1728910.  

It's sitting in setiathome.berkeley.edu, right next to app_info and app_config, which BOINC has no problem finding.

cc_config.xml should be in the BOINC folder right next to client_state.xml, not in setiathome.berkeley.edu folder...
ID: 1728919 · Report as offensive
Profile Fawkesguy
Volunteer tester
Avatar

Send message
Joined: 8 Jan 01
Posts: 108
Credit: 188,578,766
RAC: 0
United States
Message 1728921 - Posted: 25 Sep 2015, 22:13:20 UTC - in response to Message 1728919.  

Well that's embarrassing...... :-)

It's now where it belongs and all is well.

Thank you!



It's sitting in setiathome.berkeley.edu, right next to app_info and app_config, which BOINC has no problem finding.

cc_config.xml should be in the BOINC folder right next to client_state.xml, not in setiathome.berkeley.edu folder...
ID: 1728921 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 1729215 - Posted: 26 Sep 2015, 18:29:15 UTC - in response to Message 1728896.  

I looked at the explanation for that command on the BOINC wiki (https://boinc.berkeley.edu/wiki/Client_configuration) and I want to make sure I understand its usage.

<no_priority_change>0|1</no_priority_change>
If 1, don't change priority of applications (run them at same priority as client).
NB: This option can, if activated, impact system responsiveness for the user. Default, all CPU science apps run at lowest (idle) priority Nice 15.


So if I set it to 1, it runs at a lower priority than if I set it to 0?

No. By default BOINC runs at Normal priority & the science apps run at a lower priority. Such as Low or Below Normal. Setting this to 1 will cause the science apps to run at the same priority as BOINC.
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 1729215 · Report as offensive
Profile Fawkesguy
Volunteer tester
Avatar

Send message
Joined: 8 Jan 01
Posts: 108
Credit: 188,578,766
RAC: 0
United States
Message 1729578 - Posted: 27 Sep 2015, 17:20:30 UTC - in response to Message 1729215.  

Oh, so I had it backwards. Thank you for clarifying. I'll change that setting to "1".


I looked at the explanation for that command on the BOINC wiki (https://boinc.berkeley.edu/wiki/Client_configuration) and I want to make sure I understand its usage.

<no_priority_change>0|1</no_priority_change>
If 1, don't change priority of applications (run them at same priority as client).
NB: This option can, if activated, impact system responsiveness for the user. Default, all CPU science apps run at lowest (idle) priority Nice 15.


So if I set it to 1, it runs at a lower priority than if I set it to 0?

No. By default BOINC runs at Normal priority & the science apps run at a lower priority. Such as Low or Below Normal. Setting this to 1 will cause the science apps to run at the same priority as BOINC.
ID: 1729578 · Report as offensive

Message boards : Number crunching : Need a Little Help With MB CUDA Settings in Linux


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.