MW@H WU "stuck" at 100%?

[VirtualLarry]

It's the same type of WU that estimates 37 seconds. I've got a whole bunch of them. I woke up, came out to my PC, and Waterfox had stopped responding, and one of these WUs was sitting on 100%, GPU load was 0%, and it was just spinning its wheels for 4 hours.

I currently have it suspended. Should I let it continue? Or should I abort it?

 

[VirtualLarry]

I guess it doesn't matter. When I resumed the WU, it restarted from the beginning, and finished in 37s.

 

[Orange Kid]

Glad we could help 😉

 

[TennesseeTony]

Milkyway seems to do that a lot. I have 4 PC's with R9-280X's and as many times a day as I can I go around and see who is stuck. The CPU tasks do it too occasionally.

I have minimized the wasted time by configuring each vid card to run 3 threads at a time, with 1/3 of a cpu assigned per GPU task.

If you've not already done something similar, show hidden files, go to:

Program Data/Boinc/projects/milkyway and create a plain text document, then save it in that directory named: app_config.xml

The contents of app_config.xml should be as follows (feel free to change some numbers to suit you):

<app_config>
<app>
<name>milkyway</name>
<gpu_versions>
<gpu_usage>0.333</gpu_usage>
<cpu_usage>0.333</cpu_usage>
</gpu_versions>
</app>
<app>
<name>milkyway_separation__modified_fit</name>
<gpu_versions>
<gpu_usage>0.333</gpu_usage>
<cpu_usage>0.333</cpu_usage>
</gpu_versions>
</app>
</app_config>



After updating it should run 3 task per GPU now.
<gpu_usage> 1.0= 1 task, 0.5= 2 tasks, 0.333=3 tasks, 0.25 would be 4 tasks