Hi, I am using boinc 7.0.64 on fedora linux and I have set it to use all the cores all the time but it will only use 2 cores when the pc is under some other load, 3 cores when its under light load and only 4 cores when there is no load. There are four separate processes running but 2-3 of them will share a single core mostly meaning 2-c cores are at 100% with one or two cores at 0%.

I want boinc to use all cores all the time no matter how much load from other programs, because the processes are at nice 19 it does not matter.

here are my prefs file from /var/lib/boinc

<global_preferences>
<run_on_batteries>0</run_on_batteries>
<run_if_user_active>1</run_if_user_active>
<run_gpu_if_user_active>0</run_gpu_if_user_active>
<idle_time_to_run>0.000000</idle_time_to_run>
<suspend_cpu_usage>0.000000</suspend_cpu_usage>
<start_hour>0.000000</start_hour>
<end_hour>0.000000</end_hour>
<net_start_hour>0.000000</net_start_hour>
<net_end_hour>0.000000</net_end_hour>
<leave_apps_in_memory>0</leave_apps_in_memory>
<confirm_before_connecting>0</confirm_before_connecting>
<hangup_if_dialed>0</hangup_if_dialed>
<dont_verify_images>0</dont_verify_images>
<work_buf_min_days>0.300000</work_buf_min_days>
<work_buf_additional_days>0.900000</work_buf_additional_days>
<max_ncpus_pct>100.000000</max_ncpus_pct>
<cpu_scheduling_period_minutes>120.000000</cpu_scheduling_period_minutes>
<disk_interval>60.000000</disk_interval>
<disk_max_used_gb>3.500000</disk_max_used_gb>
<disk_max_used_pct>80.000000</disk_max_used_pct>
<disk_min_free_gb>2.500000</disk_min_free_gb>
<vm_max_used_pct>75.000000</vm_max_used_pct>
<ram_max_used_busy_pct>20.000000</ram_max_used_busy_pct>
<ram_max_used_idle_pct>20.000000</ram_max_used_idle_pct>
<max_bytes_sec_up>0.000000</max_bytes_sec_up>
<max_bytes_sec_down>0.000000</max_bytes_sec_down>
<cpu_usage_limit>100.000000</cpu_usage_limit>
<daily_xfer_limit_mb>0.000000</daily_xfer_limit_mb>
<daily_xfer_period_days>0</daily_xfer_period_days>
</global_preferences>

any ideas?

The science applications run at nice 19 as well, so they'll go out of the way of higher priority programs. There's not much you can do about that. Only GPU applications run at a slightly higher priority (Below Normal on Windows, nice 15 on Linux) because otherwise they'll not get allotted enough resources.

---

I experiencing absolutely same problem with Boinc 7.0.65 on Fedora 19 and 20 beta.
Boinc can't load all 4 cores of my Core i5 2500K. Only 2-3 cores fully loaded by it and 1-2 cores are idle.
Process priority change to highest doesn't help.


I noticed this issue only after upgrading from Fedora 17 (Boinc 7.0.29 I guess) to version 19 (Boinc 7.0.65). Then I double-checked with freshly installed OS (both versions 17 and 20 beta) - that gave me same result.
17 - loads processor fully, all 4 cores.
19/20 - only 2-3 cores are loaded simultaneously.

Hello everybody!

I just wanted to open another thread with the exact same problem when I saw this thread.

I'm also running Fedora 19 since this summer (fresh install). At first, everything worked well: Boinc started 4 tasks and always used all 4 (virtual, and 2 physical) cores up to 100%.

But 2-3 months ago I suddenly noticed a different behavior. Boinc still does start 4 tasks, but they only use 3 cores and my CPU usage stays at 75-80%.


I tried to resolve or reproduce the problem by doing a complete BOINC client and manager re-install. I also removed the /var/lib/boinc folder and reset my World Community Grid project configuration to "Maximum Output" (no restrictions) before finally reconnecting my client to the project. I didn't touch the local configuration - yet, the result is still the same: 4 tasks running, but they only use 75% of the CPU time.
Watching the process list in Gnome System Monitor, I noticed that there are usually two tasks - not always the same ones! - using a full core (used CPU time increases by 1s every second) and two tasks using half a core (CPU time increases by 0.5s/second).

Here is the beginning of the client log of this new installation / initialization: http://paste.robbi5.com/24d70cd5


Any help would be really appreciated!

The WCG projects that have tasks starting with FAHV and Ennnnn are multi-part, a worker that take the CPU time and a stager which just does nothing but launching the task segments from a second application. Those stagers use next to no CPU time, really should not be taking a thread for themselves. They are single threaded, so only use 1 processor per task. In BOINC Manager you'd only see 1 task running per job. So, this is novel, not read of this before. No GPGPU process that is secretly reserving a core?

---

Coelum Non Animum Mutant, Qui Trans Mare Currunt

I registered this issue in Red Hat Bugzilla 1 week ago - https://bugzilla.redhat.com/show_bug.cgi?id=1038283

No GPGPU process that is secretly reserving a core?

Nope, WCG is the only project connected, so there are no GPU tasks, just FightAIDS@Home, Mapping Cancer Markers and the Clean Energy Project. Still, the process list usually looks like this.

Issue was fixed and currently fix reached Fedora 20 stable repository.
https://bugzilla.redhat.com/show_bug.cgi?id=1038283