boinc

[kronickiller]

is there anyway to dump all power to one task

boinc runs like 9 tasks at once for me

guessing 1 per a core and 1 for the gpu

[tobben]

Well, the boinc client is a vast forest of options and customizability. But not as far

as i am aware, you are probably better off running it as intended, unless the ETA

is horribly long? 

 

I did come over this: "--fetch_minimal_work"

Description: Fetch only 1 job per device (CPU, GPU). Used with --exit_when_idle, the client will process one job per device, then exit.

 

That is a flag of sorts that you add to the cc_config.xml file in the boinc data directory.

Not sure if it will give the desired effect or literally just do 1 task and give you the

utilization corresponding to one task on one thread.

 

To see more about client configuration in boinc through the configuration file you can

have a look here http://boinc.berkeley.edu/wiki/client_configuration

 

@alpenwasser might be able to help.

[alpenwasser]

is there anyway to dump all power to one task

boinc runs like 9 tasks at once for me

guessing 1 per a core and 1 for the gpu

Do you mean you wish to run only one application at once, but dedicate all available

resources to it? I've never really tried that, nor looked for it, but this thread might

be of some interest: http://boinc.berkeley.edu/dev/forum_thread.php?id=8870

Honestly though, there probably isn't all that much point to it. Unless you're losing

bonus credits on your projects because of slow turnover it's probably not worth it. I

haven't looked into which projects offer bonuses, and if so, under what sort of conditions,

except for GPUGrid, which offers the biggest bonuses if you return the WU within 24 hours

(or at least that was the way it worked last I checked a few months ago), which so far I'm

able to keep even when running two applications in parallel on my GPU (I reckon this will

change as units get harder to crunch to deal with more powerful GPUs, but for now it still

works).

 

 

Well, the boinc client is a vast forest of options and customizability. But not as far

as i am aware, you are probably better off running it as intended, unless the ETA

is horribly long?

Vast forest indeed, and agreed.