BOINC 7.0.40 update heads up

[Yeroon]

If you want to run HCC GPU and other CPU projects then you'll need to add some code to the app_config file for that particular CPU project. Also, if you leave the max_concurrent to 3 then only 3 of your CPU threads will be used not matter what the other settings may be.. Setting max_concurrent to 8, leaving gpu_usage at .33 and setting cpu_usage to 1 should get you 3 GPU tasks running and leave the other 5 threads for whatever else you want to run. Like Snow Crash said you may have to shut down and restart boinc twice for the app_config to start. I had that happen to me a couple of times. First time it did I thought something was wrong with the config file.

This has not been my experience. I have only the gpu app config info and its specced to use two cpu cores for the gpu. Boinc set 100% of cpu cores will then fill the remaining two cores with whatever cpu projects i have enabled.

[nanoprobe]

This has not been my experience. I have only the gpu app config info and its specced to use two cpu cores for the gpu. Boinc set 100% of cpu cores will then fill the remaining two cores with whatever cpu projects i have enabled.

That works also if you're using all the remaining cores for a specific CPU project. I thought he was trying to run a predetermined amount of CPU cores ( say 2 of the remaining 5) on a given WCG project. That would require extra coding in the app_config file.

[nanoprobe]

At the moment I have this in an app_config file and I have 4 GPU WU?s and 4 CPU WU?s (from different projects)

Would I be correct in assuming they are not WCG CPU projects?

[Rob_B]

No. I only run WCG.
There are 4 GPU tasks running and it is downloading and running CPU tasks for the 4 cores that are not used for the GPU.

[STEvil]

If you want to run HCC GPU and other CPU projects then you'll need to add some code to the app_config file for that particular CPU project. Also, if you leave the max_concurrent to 3 then only 3 of your CPU threads will be used not matter what the other settings may be.. Setting max_concurrent to 8, leaving gpu_usage at .33 and setting cpu_usage to 1 should get you 3 GPU tasks running and leave the other 5 threads for whatever else you want to run. Like Snow Crash said you may have to shut down and restart boinc twice for the app_config to start. I had that happen to me a couple of times. First time it did I thought something was wrong with the config file.

So the commented portions of this are not correct?

<app_config>
<app>
<name>hcc1</name>
<max_concurrent>3</max_concurrent> This is the total number of threads you want to run on BOTH cards added together
<gpu_versions>
<gpu_usage>.33</gpu_usage> This is 1 divided by the number of threads you want to run on ONE card
<cpu_usage>2.67</cpu_usage> This is the number of cores(threads) you have on your CPU divided by the max_concurrent above
</gpu_versions>
</app>
</app_config>

[nanoprobe]

@ STEvil I'm not sure what you're trying to do. If you're trying to run 3 tasks on each card the max_concurrent should read 6. (3 for each card)
Your gpu_usage setting of .33 would be correct. (1 divided by the number of tasks for each card ( 3)
Your cpu_usage would be 1.33 (The total number of cpu cores (8) divided by the max_concurrent (6)

@Rob_B Interesting. When I played with the max_concurrent setting it would not use more cores than I had in the setting when I tried to run HCC GPU tasks other CPU tasks at the same time. I'll have to go back and revisit that.

[Rob_B]

@Rob_B Interesting. When I played with the max_concurrent setting it would not use more cores than I had in the setting when I tried to run HCC GPU tasks other CPU tasks at the same time. I'll have to go back and revisit that.

I have played a little more.
Now I have this in the app_config file:

<app_config>
<app>
<name>hcc1</name>
<max_concurrent>4</max_concurrent>
<gpu_versions>
<gpu_usage>.25</gpu_usage>
<cpu_usage>1</cpu_usage>
</gpu_versions>
</app>
<app>
<name>cep2</name>
<max_concurrent>1</max_concurrent>
</app>
</app_config>

I let the rig download plenty of CEP2 wu's and only after that wu's from other projects. Now it is doing 4 GPU wu's, 1 CEP2 wu and 3 wu's of other WCG projects.
Now I'm curious if it will start a new CEP2 wu as soon is the first CEP2 wu is done. That is what I would like to see. Running a 4 day cache and limiting the number of CEP2 wu is in the device profile only gives you one CEP2 wu every 4 days. Allowing more in the device profile and limit it to running just one wu at the time in app_config file gives you a few finished wu per day.

[nanoprobe]

@Rob_B Do you have all the other projects selected to use your 3 available cores? I would expect your next CEP2 task would start when the one that is currently running finishes. Great idea. I haven't run any CEP2 since the windows version came out so I may try your setup to run some more.

[STEvil]

@ STEvil I'm not sure what you're trying to do. If you're trying to run 3 tasks on each card the max_concurrent should read 6. (3 for each card)
Your gpu_usage setting of .33 would be correct. (1 divided by the number of tasks for each card ( 3)
Your cpu_usage would be 1.33 (The total number of cpu cores (8) divided by the max_concurrent (6)

@Rob_B Interesting. When I played with the max_concurrent setting it would not use more cores than I had in the setting when I tried to run HCC GPU tasks other CPU tasks at the same time. I'll have to go back and revisit that.

I only have 1 card.

edit

Just did a brand new fresh install (cleaned out EVERYTHING) and still only 1 thread...

[Rob_B]

@Rob_B Do you have all the other projects selected to use your 3 available cores? I would expect your next CEP2 task would start when the one that is currently running finishes. Great idea. I haven't run any CEP2 since the windows version came out so I may try your setup to run some more.

I have run this with no projects selected, all projects selected and only CEP2 selected. But always with "If there is no work available for my computer for the projects I have selected above, please send me work from another project." checked.

Adding the CEP2 info in the app_config file does not give piority over other projects. When a task finishes, an other project can take that slot and no CEP2 wu will be running.
Would have been nice if it worked like I hoped it would.

[Snow Crash]

@STEvil - Couple of thoughts ... have you updated your WCG profile that particular machine is using to say
Help Conquer Cancer = Yes
Run on GPU = Yes
Run CPU = No

Please drop the following app_config.xml file into the WCG directory of BOINC (mine happens to be on Z:\)
Z:\Program Data\projects\www.worldcommunitygrid.org

<app_config>
 <app>
 <name>hcc1</name>
 <gpu_versions>
  <max_concurrent>3</max_concurrent>
  <gpu_usage>0.333</gpu_usage>
  <cpu_usage>1.0000</cpu_usage>
 </gpu_versions>
 </app>
</app_config>

Fully stop BOINC
- If you are not given an option to "stop running tasks", go to Tools --> Options --> General tab, check the Enable Manager Exit box)
- Make sure the app_config is in the correct folder
- Restart BOINC
- Give i a few seconds and if it is still runnning only 1 WU, fully stop and restart BOINC again (yes, sometimes this has to be done 2 times)
- If it is still not running 3 then please post the messages from the Event Log from when it starts up to when it starts the first WU.

[nanoprobe]

I only have 1 card.

edit

Just did a brand new fresh install (cleaned out EVERYTHING) and still only 1 thread...

Post #55 of yours above mentioned something about both cards so I thought you were trying to run 2.

This is the total number of threads you want to run on BOTH cards added together

The app_config file below should run 3 GPU tasks using 3 of your CPU cores. If it's not working then the only thing that I can think of is you may not be putting the config file in the right folder. It must go into the WCG project folder, not the BOINC folder or anywhere else. Make sure the file is named app_config.xml

<app_config>
<app>
<name>hcc1</name>
<max_concurrent>8</max_concurrent>
<gpu_versions>
<gpu_usage>.33</gpu_usage>
<cpu_usage>1.0</cpu_usage>
</gpu_versions>
</app>
</app_config>

[STEvil]

So apparently it doesnt go in C:\Users\All Users\BOINC anymore like google said.. rofl :p

edit

Nanoprobe - that was a copy/paste of OldChap's info on page 2.

Currently running this which gives me 6 cpu threads and 3 GPU threads. Seems to be about 4:10-4:20 per GPU WU where running 7 CPU threads and 3 GPU threads results in about 5 minutes per GPU thread. It also seems to desynchronize the GPU threads slightly which is what I was aiming for

<app_config>
<app>
<name>hcc1</name>
<gpu_versions>
<max_concurrent>3</max_concurrent>
<gpu_usage>0.333</gpu_usage>
<cpu_usage>0.667</cpu_usage>
</gpu_versions>
</app>
</app_config>

[nanoprobe]

Glad you got it working. Now that you have it in the right folder you can play with the settings to find what works best for you.

[bearcatrp]

I see folks changing the cpu usage to less than one in app_config file.. Doesn't that increase crunching time not giving it a full thread/core, especially if your maxing out all threads crunching? Maybe I am reading to much into this. I have mine a 1, which seems to work well. I do have empty thread to spare though.

[STEvil]

The GPU does not require a full CPU thread all the time. If I were running nothing but GPU tasks it wouldnt be an issue, but I am running CPU and GPU tasks (actually ran out of GPU tasks at the moment.. hmm).

[jeanguy2]

STevil,

This has happened in the pass, looks like the WCG server has run out of ATI HCC WUs. My cache is dry too... We will have to run other WUs until the WCG techs address the problem..

[Yeroon]

Trried running two wu on llano with 7.0.42, didn't help so deleted app_config. Won't stop running two wu's. Restarted boinc/computer, reinstalled 7.0.42, uninstalled/reinstalled 7.0.28, it just keeps running two gpu units. Any suggestions? Currently getting it to empty task buffer and will try to clean it all out later.
Yeroon

[Snow Crash]

sounds like app_config is still in your WCG project directory ...
... \Program Data\projects\www.worldcommunitygrid.org
or you have an app_info file in that same directory?

[Yeroon]

Never ran an app_file on that comp, too much work when it was already doing 14-18min wu's 1 at a time. With how easy app_config is i figured i'd give it a try, hut it was getting too many comp errors. Got fit fixed by re-installing 7.0.42 and running the app_config at 1+1. New problem, threw in a 6450 to see if it works, don't know how to setup so that both can crunch.
Installed in 3850 llano machine, can enable/disable dual graphics in catalyst. Disabled, crunches on IGP (faster one), enabled its only doing the 6450.
Max concurent is set to 2
Gpu/cpu both set to 1

No biggie if I don't get if figured out, the 6450 won't be in for long anyways, but its in now so if it can do science why not let it?

[Snow Crash]

You may need to add a cc_config.xml for both GPU to work.
Unlike app_config, this goes in the root dir of ...\BOINC\ Program Data

<cc_config> 
   <options> 
      <use_all_gpus>1</use_all_gpus>
   </options> 
   <log_flags> 
   </log_flags> 
</cc_config>

[Yeroon]

You may need to add a cc_config.xml for both GPU to work.
Unlike app_config, this goes in the root dir of ...\BOINC\ Program Data

<cc_config> 
   <options> 
      <use_all_gpus>1</use_all_gpus>
   </options> 
   <log_flags> 
   </log_flags> 
</cc_config>

works now, thanks!
37min Wu are great lol

[Yeroon]

New machine, new problem. As its an older machine with a dual core athlon xp @ 2.8, it can barely keep up to the stock clocked 7770 running 3 threads. How do i tell boinc to not download any cpu tasks for this machine? I have a separate device profile that only allows hcc wu's, but its still sending the hcc cpu tasks. Running 7.0.42 on linux.

[wplachy]

In the profile "Advanced Options" set "Allow research to run on my CPU?" to No

[Yeroon]

In the profile "Advanced Options" set "Allow research to run on my CPU?" to No

Thank you!