Unable to run more than one task

Message boards : Number crunching : Unable to run more than one task
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
WPrion

Send message
Joined: 10 Apr 13
Posts: 16
Credit: 796,282
RAC: 0
United States
Message 1822617 - Posted: 8 Oct 2016, 2:52:01 UTC

I've not run SETI for a while, but now that I'm back I'm only able to get ONE setiathome_v8 task to run at a time on one 980TI. That doesn't seem right! I do run multiple tasks fine on other projects (e.g. Einstein, GPUGrid)

I've upadted my app_config.xml to include _v8, and it loads without error:

<app_config>
<app>
<name>setiathome_v8</name>
<max_concurrent>4</max_concurrent>
<gpu_versions>
<gpu_usage>0.25</gpu_usage>
<cpu_usage>0.25</cpu_usage>
</gpu_versions>
</app>
</app_config>

I've omitted the sections for astropulse_v7 and setiathome_v7. Do I still need those?

Help appreciated.

Win
ID: 1822617 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13722
Credit: 208,696,464
RAC: 304
Australia
Message 1822632 - Posted: 8 Oct 2016, 3:50:36 UTC - in response to Message 1822617.  
Last modified: 8 Oct 2016, 4:00:55 UTC

Running 4 WUs at a time would be pushing things a bit, not to mention you'll be doing less work than if just running 2WUs at a time. 2-3 seems to be the sweet spot for higher end cards.

If you really want to run 4WUs at a time, try dropping the <cpu_usage>0.25</cpu_usage> value down to 0.20, Read config files again & things should start working then.


EDIT-also depending on what your settings are for sharing time between all your different projects will affect what runs, and how many instances. I suspect your system only having 2 cores/4 threads will also have an affect on work allocation as well.
Grant
Darwin NT
ID: 1822632 · Report as offensive
Profile Brent Norman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester

Send message
Joined: 1 Dec 99
Posts: 2786
Credit: 685,657,289
RAC: 835
Canada
Message 1822650 - Posted: 8 Oct 2016, 4:24:54 UTC - in response to Message 1822617.  

I've omitted the sections for astropulse_v7 and setiathome_v7. Do I still need those?


You can leave the astropulse_v7 section with 2 task running, and remove the setiathome_v7 section.

astropulse_v7 tasks come in spurts about once a week.
ID: 1822650 · Report as offensive
WPrion

Send message
Joined: 10 Apr 13
Posts: 16
Credit: 796,282
RAC: 0
United States
Message 1822697 - Posted: 8 Oct 2016, 11:44:37 UTC - in response to Message 1822632.  

You're probably correct about 4 being too much, but I can't get past ONE concurrent task, even when SETI is the only task enabled.
ID: 1822697 · Report as offensive
WPrion

Send message
Joined: 10 Apr 13
Posts: 16
Credit: 796,282
RAC: 0
United States
Message 1822698 - Posted: 8 Oct 2016, 11:44:59 UTC - in response to Message 1822650.  

Thanks!
ID: 1822698 · Report as offensive
WPrion

Send message
Joined: 10 Apr 13
Posts: 16
Credit: 796,282
RAC: 0
United States
Message 1822702 - Posted: 8 Oct 2016, 12:11:51 UTC - in response to Message 1822617.  

Based on a PM I received maybe this is the problem: I only want to run GPU tasks; no CPU-only tasks, so I have CPU deselected in the project settings.

With CPU selected I can run more than one CPU-only tasks, but I want to run more than one GPU task with NO CPU-only tasks and am still unable to do so.

It was suggested to delete the max_concurrent line, but that did not help.

Still wondering...

Thanks,

Win
ID: 1822702 · 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 1822765 - Posted: 8 Oct 2016, 18:30:12 UTC - in response to Message 1822617.  
Last modified: 8 Oct 2016, 18:30:51 UTC

I've not run SETI for a while, but now that I'm back I'm only able to get ONE setiathome_v8 task to run at a time on one 980TI. That doesn't seem right! I do run multiple tasks fine on other projects (e.g. Einstein, GPUGrid)

I've upadted my app_config.xml to include _v8, and it loads without error:

<app_config>
<app>
<name>setiathome_v8</name>
<max_concurrent>4</max_concurrent>
<gpu_versions>
<gpu_usage>0.25</gpu_usage>
<cpu_usage>0.25</cpu_usage>
</gpu_versions>
</app>
</app_config>

I've omitted the sections for astropulse_v7 and setiathome_v7. Do I still need those?

Help appreciated.

Win



2 suggestions.

First install lunatics v45 beta 4 and select the SoG for the GPU. Don't use the CPU for any work so you can support your GPU. You can do 3 work units per GPU on that 980TI

Second, try this app_config.xml

If you find too much screen lag, let us know and we can fine tune it for you.


<app_config>
<app_version>
<app_name>setiathome_v8</app_name>
<plan_class>opencl_nvidia_SoG</plan_class>
<avg_ncpus>1</avg_ncpus>
<ngpus>0.33</ngpus>
<cmdline>-sbs 512 -period_iterations_num 10 -hp -high_prec_timer -high_perf </cmdline>
</app_version>
<app_version>
<app_name>astropulse_v7</app_name>
<plan_class>opencl_nvidia_100</plan_class>
<avg_ncpus>1</avg_ncpus>
<ngpus>0.33</ngpus>
<cmdline>-unroll 28 -oclFFT_plan 256 16 256 -ffa_block 12288 -ffa_block_fetch 6144 -tune 1 64 4 1 -tune 2 64 4 1 -hp </cmdline>
</app_version>
<project_max_concurrent>3</project_max_concurrent>
</app_config>

Zalster
ID: 1822765 · Report as offensive
WPrion

Send message
Joined: 10 Apr 13
Posts: 16
Credit: 796,282
RAC: 0
United States
Message 1822986 - Posted: 9 Oct 2016, 16:08:03 UTC - in response to Message 1822765.  

Thanks for the tips here plus some I've rcvd in PMs.

I haven't touched Lunatics for two years so I upgraded to the latest non-beta version. That seems to be enough (I'm not fond of betas) as I am now running the 3 concurrent tasks my app_config is set for. GPU utilization is around 93-94%, and each associated CPU task is around 7-9.5%. Does that sound right?

One oddity. The event log showed error msgs from setiathome_v7 statements in the new app_info file. Is _v7 completely history now (shows how long I've been gone)? Anyway, I commented those lines out and get no errors when reloading config files.

Thanks again,

Win
ID: 1822986 · Report as offensive
AMDave
Volunteer tester

Send message
Joined: 9 Mar 01
Posts: 234
Credit: 11,671,730
RAC: 0
United States
Message 1822988 - Posted: 9 Oct 2016, 16:25:09 UTC - in response to Message 1822986.  

One oddity. The event log showed error msgs from setiathome_v7 statements in the new app_info file. Is _v7 completely history now (shows how long I've been gone)? Anyway, I commented those lines out and get no errors when reloading config files.

Yes,  SETI@home Version 8 has been released.
ID: 1822988 · 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 1823032 - Posted: 9 Oct 2016, 19:34:48 UTC - in response to Message 1822986.  

Just a FYI, SoG are faster at processing Breakthrough Listen work units than the cudas. You can experiment with them and see what you prefer.
ID: 1823032 · Report as offensive
iwajabitw
Avatar

Send message
Joined: 18 Mar 00
Posts: 19
Credit: 9,285,966
RAC: 0
United States
Message 1825543 - Posted: 20 Oct 2016, 4:01:46 UTC - in response to Message 1822765.  

I've not run SETI for a while, but now that I'm back I'm only able to get ONE setiathome_v8 task to run at a time on one 980TI. That doesn't seem right! I do run multiple tasks fine on other projects (e.g. Einstein, GPUGrid)

I've upadted my app_config.xml to include _v8, and it loads without error:

<app_config>
<app>
<name>setiathome_v8</name>
<max_concurrent>4</max_concurrent>
<gpu_versions>
<gpu_usage>0.25</gpu_usage>
<cpu_usage>0.25</cpu_usage>
</gpu_versions>
</app>
</app_config>

I've omitted the sections for astropulse_v7 and setiathome_v7. Do I still need those?

Help appreciated.

Win



2 suggestions.

First install lunatics v45 beta 4 and select the SoG for the GPU. Don't use the CPU for any work so you can support your GPU. You can do 3 work units per GPU on that 980TI

Second, try this app_config.xml

If you find too much screen lag, let us know and we can fine tune it for you.


<app_config>
<app_version>
<app_name>setiathome_v8</app_name>
<plan_class>opencl_nvidia_SoG</plan_class>
<avg_ncpus>1</avg_ncpus>
<ngpus>0.33</ngpus>
<cmdline>-sbs 512 -period_iterations_num 10 -hp -high_prec_timer -high_perf </cmdline>
</app_version>
<app_version>
<app_name>astropulse_v7</app_name>
<plan_class>opencl_nvidia_100</plan_class>
<avg_ncpus>1</avg_ncpus>
<ngpus>0.33</ngpus>
<cmdline>-unroll 28 -oclFFT_plan 256 16 256 -ffa_block 12288 -ffa_block_fetch 6144 -tune 1 64 4 1 -tune 2 64 4 1 -hp </cmdline>
</app_version>
<project_max_concurrent>3</project_max_concurrent>
</app_config>

Zalster


I have the MSI GTX980 OC and just changed my app_config to the one above. So far so good will report back. Tomorrow I should be adding in another MSI GTX 980 OC in SLI.
ID: 1825543 · Report as offensive
iwajabitw
Avatar

Send message
Joined: 18 Mar 00
Posts: 19
Credit: 9,285,966
RAC: 0
United States
Message 1825546 - Posted: 20 Oct 2016, 4:51:52 UTC - in response to Message 1825543.  

Curios though its not showing 3 GPU tasks running as the app_config says it should. Instead it says .04 CPU +1 Nvidia GPU. Its crunching a guppy, so maybe that's it. Strange though.
ID: 1825546 · 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 1825548 - Posted: 20 Oct 2016, 4:58:34 UTC - in response to Message 1825546.  

You will not see a change in the percentage of the CPU until a new work units download with the new app_config.xml

Did you remember to tell BOINC to read the config files. Or you could exit boinc and then restart it and make sure it reads the config file from the event log
ID: 1825548 · Report as offensive
iwajabitw
Avatar

Send message
Joined: 18 Mar 00
Posts: 19
Credit: 9,285,966
RAC: 0
United States
Message 1825552 - Posted: 20 Oct 2016, 5:13:41 UTC - in response to Message 1825548.  

Yes, I always shut boinc down before editing the app file, so that it reads it new on the start up.
ID: 1825552 · 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 1825559 - Posted: 20 Oct 2016, 5:31:30 UTC - in response to Message 1825552.  

hmm, what does the start up log say? You see an error message?
ID: 1825559 · Report as offensive
iwajabitw
Avatar

Send message
Joined: 18 Mar 00
Posts: 19
Credit: 9,285,966
RAC: 0
United States
Message 1825652 - Posted: 20 Oct 2016, 16:08:34 UTC - in response to Message 1825559.  

Woke up to the same as when I went to bed. Only 2cpu tasks going and the .o4cpu +1Nvidia Gpu running a guppi. So I went back to lunatics.44 and my original app_confign this morning. When I get home I plan on getting the second card setup in SLI before trying the beta SOG again. Figure i need to have what I know word before messing too much with anything. I will say though those guppi unit's finished faster than ever, about 15min, but only one. The only error I saw was the server backing my requests off. My specs are--

Asrock x99 Killer 3.1
i7 5820K@3.3ghz
2x MSI Armor 2x GTX 980 OC
16GB DDR2800@2133-XMP turned off

I usually have 84% CPU going for 10 tasks leaving one core free, and .33 in the config for 3 gpu tasks, guppis taking about 53min to complete. Onmy phone so sorry about the grammar.
ID: 1825652 · Report as offensive
Profile TimeLord04
Volunteer tester
Avatar

Send message
Joined: 9 Mar 06
Posts: 21140
Credit: 33,933,039
RAC: 23
United States
Message 1825673 - Posted: 20 Oct 2016, 18:14:38 UTC

BOINC does NOT make use of SLI. If you are just planning on SLI for BOINC, don't waste your money. If you are using SLI for gaming - that's a different matter, and more power to you. 😃

As to the rest, I'll let Zalster continue helping you. 😃


TL
TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos
ID: 1825673 · Report as offensive
iwajabitw
Avatar

Send message
Joined: 18 Mar 00
Posts: 19
Credit: 9,285,966
RAC: 0
United States
Message 1825677 - Posted: 20 Oct 2016, 18:24:02 UTC - in response to Message 1825673.  
Last modified: 20 Oct 2016, 18:25:51 UTC

Yeah, SLI is for Doom. As well as getting the second card for $250 shipped. But will disable it to crunch 2 gpu's the rest of the time.
ID: 1825677 · Report as offensive
Profile Brent Norman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester

Send message
Joined: 1 Dec 99
Posts: 2786
Credit: 685,657,289
RAC: 835
Canada
Message 1825687 - Posted: 20 Oct 2016, 19:04:41 UTC - in response to Message 1825652.  
Last modified: 20 Oct 2016, 19:08:31 UTC

I have a feeling that from copy/pasting that you have acquired some non-standard characters in your app_config.xml that look like a space - but they are NOT. Causing BOINC to reject the file completely.

Delete any spaces/tabs/returns and renter them.


Either that or you have mismatched tags <gpu_usage>0.3</gpu_usage> the tags must match exactly or the file will not be read correctly.

EDIT: And ohhh, app_config.xml is a TXT document not XML, use notepad to edit it.

Brent
ID: 1825687 · Report as offensive
Mike Davis
Volunteer tester

Send message
Joined: 17 May 99
Posts: 240
Credit: 5,402,361
RAC: 0
Isle of Man
Message 1825696 - Posted: 20 Oct 2016, 19:45:09 UTC
Last modified: 20 Oct 2016, 19:47:31 UTC

Its been a while since ive made an app_config, so i have dug out a older one i had around, but something like this should work to get you up and running, might need to change the plan class to whatever apps you are running.

<app_config>
<app_version>
<app_name>setiathome_v8</app_name>
<plan_class>cuda50</plan_class>
<avg_ncpus>0.02</avg_ncpus>
<ngpus>0.25</ngpus>
</app_version>
<app_version>
<app_name>astropulse_v7</app_name>
<plan_class>opencl_nvidia_100</plan_class>
<avg_ncpus>0.2</avg_ncpus>
<ngpus>0.25</ngpus>
</app_version>
<project_max_concurrent>4</project_max_concurrent>
</app_config>
ID: 1825696 · Report as offensive
1 · 2 · Next

Message boards : Number crunching : Unable to run more than one task


 
©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.