why I'm dumping eon
Posted: Thu Jan 24, 2013 2:52 am
Just wanted to let you know that I'm dropping your project after it finishes the current work unit because it ignores BOINC run parameters and takes over my computer. Fortunately, I read the forum after starting it.
Yes, I appreciate that you need fast parallel results for an app that models the current state of a simulation (i.e., that past results are irrelevant). I also realize that this requires changing the task's CPU-dispatch priority from "idle". This is completely legitimate.
So buy your own damn computers and run it legitimately.
-- faye kane, girl brain
sexiest homeless astrophysicist you'll ever see naked
http://tinyurl.com/kanecave
PS:
You also don't seem to use the Nvidia GPU efficiently (if at all), since mine is at 2% utilization. That's enough reason to dump you right there. The whole reason I do BOINC is to contribute the massive vector processor in my Geforce 470 to science. Rewrite your parallel app in CUDA and you probably could do the whole thing on your own machines if you buy Tesla HPC cards and plug them into a few 6-core PCs.
PPS:
If the rest of y'all want to run eon, but at idle priority, just set persistent priority for the BOINC project to "idle" using a utility like taskinfo. Any daughter subtasks (like the science code) will inherit the idle priority. You can also allocate specific cores to BOINC instead of letting overzealous projects like this one wander around your CPU, grabbing resources.
Yes, I appreciate that you need fast parallel results for an app that models the current state of a simulation (i.e., that past results are irrelevant). I also realize that this requires changing the task's CPU-dispatch priority from "idle". This is completely legitimate.
So buy your own damn computers and run it legitimately.
-- faye kane, girl brain
sexiest homeless astrophysicist you'll ever see naked
http://tinyurl.com/kanecave
PS:
You also don't seem to use the Nvidia GPU efficiently (if at all), since mine is at 2% utilization. That's enough reason to dump you right there. The whole reason I do BOINC is to contribute the massive vector processor in my Geforce 470 to science. Rewrite your parallel app in CUDA and you probably could do the whole thing on your own machines if you buy Tesla HPC cards and plug them into a few 6-core PCs.
PPS:
If the rest of y'all want to run eon, but at idle priority, just set persistent priority for the BOINC project to "idle" using a utility like taskinfo. Any daughter subtasks (like the science code) will inherit the idle priority. You can also allocate specific cores to BOINC instead of letting overzealous projects like this one wander around your CPU, grabbing resources.