15th WCG Birthday Challenge

Terminator

Member
Dear members of BOINC team USA ,

For Nov 16th to Nov 22nd 2019 SETI.Germany is arranging a challenge at World Community Grid on the occasion of the grids 15th birthday:




We would like to invite BOINC team USA to participate in the challenge and hope that your team and many members of your team will be joining and participting this event.
!!Be aware, only your WCG-Team-Captain can join your team to the challenge!!
https://www.worldcommunitygrid.org/team/challenge/viewTeamChallenge.do?challengeId=9700

More information and statistics regarding the challenge can be found here.http://www.seti-germany.de/wcg/1_en_Welcome.html


Your participation is very much appreciated.

Best regards
Terminator
SETI.Germany
 

supdood

Well-Known Member
USA team member
Looking forward to it! Always appreciate the effort that SETI.Germany puts into their challenge websites--love the excitement of hourly stats updates. Thanks, Terminator!
 

Vester

Well-Known Member
USA team member
I would join you, but my four cores are busy supporting 6 GPUs running Milkyway. Good luck. I joined United Devices in January of 2002 on the anthrax project.
 

Vester

Well-Known Member
USA team member
Billy, I don't want to have a bottleneck by using the CPU that would possibly decrease GPU performance, but I will try it for a while.

Edit: It runs two WCG jobs while running 18 Milkyway work units on the 6 GPUs. While waiting for more GPU work, it runs 4 cores on WCG. My Intel(R) Core(TM) i5-7600 CPU is not a multi-threaded processor (could only use single-threaded mode on my ASUS B250 Mining Rig anyhow, so I made this choice to save money). The CPU is running at 4 GHz by settings in the BIOS. The CPU is not loaded 100% when running only two cores for WCG while running GPU work as well, so Milkyway's GPU work should not be affected. Thanks for your encouragement, Billy.

Second Edit: WCG and Milkyway do not play well together. Does anyone know how to make both of them run 100% of the time instead of alternating projects? I found WCG running on four cores and no Milkyway work units running on the GPUs although Milkyway had work to do. I am unwilling to decrease the performance of the GPUs.
 
Last edited:

Nick Name

Administrator
USA team member
Second Edit: WCG and Milkyway do not play well together. Does anyone know how to make both of them run 100% of the time instead of alternating projects? I found WCG running on four cores and no Milkyway work units running on the GPUs although Milkyway had work to do. I am unwilling to decrease the performance of the GPUs.
How many WCG jobs do you want to run? You should probably limit them to no more than three to keep the 4th core free to feed those GPUs.
This goes in the WCG project folder, which is located in your BOINC data folder.

Code:
<app_config>

<project_max_concurrent>3</project_max_concurrent>

</app_config>
Save this with Notepad or any plain text editor. Make sure it's named app_config.xml. Make sure you save it as type All Files, or else it will save as a text file and that won't work.


One of the quirks with the BOINC scheduler is it will overcommit the CPU when you are also doing GPU work. E.g. for your four cores, it will use all four cores for WCG and then also try to run MilkyWay. Usually AMD cards don't require a lot of CPU support, relatively speaking, but this would definitely cause some slowdown. You will also run into problems like you've seen where only project is running when you want both to run. The best solution is to set up separate clients, one for CPU projects and one for GPU. It takes a little work but will save you a lot of aggravation in the long run.

*Edited for a typo in the app_config
 
Last edited:

Billy Ewell 1931

Member
USA team member
Vester: You are a dedicated cruncher and I fully support your individualized settings of project "Preferences". I am concerned that any advice I give you on the Second Edit, particularly, would only confuse the issue. I am surprised that the experts out there have not already posted to give you the solutions to the preferences matter you have addressed.
 

Vester

Well-Known Member
USA team member
I have an app_config.xml in the milkyway project folder. The app_config.xml for WCG did not change anything. I tried several other ways, too, but I suspect that it wants the specific project(s) to be named.

Here's what I have done. I am running WCG on all four cores. That reduced the number of GPU work units to one (1) when running 0.33 GPU and 0.22 CPU settings in app_config.xml. I decided to reduce the CPU value and after several tries I got down to 0.05 CPU. It is now running four CPU cores on WCG and the six GPUs are running 3 Milkyway work units each. I will have to wait and see if daily output at Milkyway stays around 2,000,000 per day.
 

Nick Name

Administrator
USA team member
I have an app_config.xml in the milkyway project folder. The app_config.xml for WCG did not change anything. I tried several other ways, too, but I suspect that it wants the specific project(s) to be named.
<project_max_concurrent> isn't project or app specific, it should work right away. I see I have a typo in the above, I'll fix that now. :eek:

I forgot to mention that you need to either select Read Config files in the manager (Options -> Read config files), or quit and restart BOINC to apply the app_config. Usually just selecting Read is enough, but more than once I've had to restart BOINC to get it working. Either way, check your log under Tools (or CTRL + Shift + E) to make sure it was picked up. Look for this:

331 World Community Grid 11/12/2019 12:15:00 AM Found app_config.xml

If you don't see "Found app_config.xml", then something is wrong. If you have it in the right folder then it's probably in text format instead of xml.

Some of us are getting an early start ;), but work returned before the event's starting time doesn't count. If you'd rather wait until then to start crunching WCG, that would be fine.
 

Vester

Well-Known Member
USA team member
I do not see any degradation of Milkyway GPU work production while running four WCG tasks on my 4 CPU cores. I am going to finish WCG tasks in queue and allow new tasks on Saturday.
 

Nick Name

Administrator
USA team member
I do not see any degradation of Milkyway GPU work production while running four WCG tasks on my 4 CPU cores. I am going to finish WCG tasks in queue and allow new tasks on Saturday.
That's great to hear. I also am not seeing any impact on #1 with basically 100% CPU load.
 

supdood

Well-Known Member
USA team member
Looks like it's going to be a good one! With a little over a day to go until the start, there are 84 teams registered.
 
Top