I have a rig with Dual Radeon 7950's. I installed the BOINC manager, and added my project. Under the Task menu, it's showing all the jobs as "GPU suspended -comp..." Im assuming this means the GPU's are not being used. Is there a way to "activate" the GPU's so BOINC will use them?

Sorry, these are 7950's running under Windows 7.

http://boinc.berkeley.edu/wiki/GPU_computing.

In essence, what you need is an AMD videocard driver that supports OpenCL under your form of Linux. I am assuming you mean Linux, if not, please explain, we're not psychic.

---

This is a Windows box.

[2] AMD AMD Radeon HD 7850/7870 series (Pitcairn) (2048MB) driver: 1.4.1848 OpenCL: 1.2

Can't tell if it's working or not,.Sais suspended, and it's not using any time currently.

Ok, that seems to work. At least I now have 2 jobs running (presumably both on GPU)...

It looks like both GPU's are running by default.
I can't yet tell if the CPU or GPU is faster, I don't see any way to directly time the rate at which "work" is accomplished.

On this specific box, I have a little crappy AMC Sempron 145 processor (2.8Ghz). Probably don't care if the CPU runs or not.
On my other boxes, which only have CPU's, I would rather they use 100% CPU 100% of the time, regardless of other workload (which shouldn't be any).
Im not familiar with XML, so I don't know if im editing the files correctly or not. Can I tell the Linux boxes to use the 100% cpu/time while the Windows 7 box to use GPU's 100%?

The window you showed is for local preferences. These set the preferences on that computer only. Local preferences override web preferences.

If you want to set preferences for all computers, you either use the preferences on the project websites, or an account manager such as BAM! or Grid Republic.

---

Sounds like a good idea to set the preferences locally, since the Windows boxes have GPU's with crappy CPU's, while the Linux boxes are all server grade with higher-end CPU's and no GPUs...

In either case, I want to get it to use 100% of the local resources, CPU for Linux and GPU for Windows.

Ill see if I can edit the 13% and make that 100% on the Windows box. The Linux box doesn't have any GUI available, so im just going to edit the XML file and replace 13 with 100.

Ill see if I can edit the 13% and make that 100% on the Windows box.

Do know that the value here is an integer. Which means that in the case of 13% on a single core CPU, you've told BOINC to use zero CPU.
And then it falls back to a value set through some project's global computing preferences, named "On multiprocessors, use N of the processors", where N is just a number, not a percentage.

Then there's a fail-safe, for if you set both these values to zero, but have told BOINC to use the CPU. It'll then switch to using 1 CPU (core) at minimum.

---

I don't know how it got set to 13% in the first place, but if setting to 0 is the best option, ill do that instead.

No, set it to 100

---

Watch out. I just noticed that the option to set a set value of multiprocessors (not a percentage) is no longer available in the project's global computing preferences.

This means that newer clients don't use this value anymore, and haven't been for a while now. Old clients (BOINC 4, 5, 6) that still support this option will need to have their global_prefs.xml file edited by hand, the tag <max_ncpus>N</max_ncpus> added, where N is the maximum amount of CPUs in that system. Then that system shouldn't try to get newer preferences in from the website, as then that value is lost again in the XML file.

By the way, BOINC XML files are easily done by hand. It is based on Extensible Markup Language, but developed specifically for BOINC. It requires no specific XML editor to write or edit, an ASCII text editor such as Notepad, or Notepad++ is enough. Do save all files with ANSI encoding, instead of UTF-8 or other Unicode. Then BOINC won't have a problem with it.

---

Unless im not understanding correctly, my 2x Raedon 7850's are not running anywhere near as good as my CPU rigs, based on the "average credit" listings.
The GPU's are only getting aboput 1950 while the CPU rig is getting about 12000.
Heck, even my plain-jane desktop is getting around 4000 (about 2x better than the dual GPU rig).

Im really pissed again...
My 128 core box is only running 16 threads.
I just can't seem to figure out why.

Where are you finding this info?
The boinc_client has been running for 30 mins, and has been showing only 16 threads of workload the entire time.
How do I get it to "refresh" and send/receive new work?

top - 15:28:41 up 27 days, 4:41, 1 user, load average: 16.09, 16.12, 18.89
Tasks: 2193 total, 17 running, 2176 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.0%us, 0.1%sy, 12.5%ni, 87.4%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st
Mem: 529223744k total, 9398920k used, 519824824k free, 1761116k buffers
Swap: 4194296k total, 0k used, 4194296k free, 4019228k cached

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
29411 rwatkins 39 19 18172 3472 1716 R 100.4 0.0 14:14.41 milkyway_separa
29442 rwatkins 39 19 5032 3840 924 R 100.4 0.0 13:08.91 milkyway_separa
29334 rwatkins 39 19 5032 3328 408 R 100.1 0.0 32:25.31 milkyway_separa
29337 rwatkins 39 19 5032 3328 408 R 100.1 0.0 31:25.90 milkyway_separa
29339 rwatkins 39 19 5032 3324 408 R 100.1 0.0 31:25.77 milkyway_separa
29341 rwatkins 39 19 5032 3328 408 R 100.1 0.0 31:25.92 milkyway_separa
29362 rwatkins 39 19 5032 3324 408 R 100.1 0.0 30:20.92 milkyway_separa
29413 rwatkins 39 19 5032 3840 924 R 100.1 0.0 14:06.01 milkyway_separa
29440 rwatkins 39 19 18172 3468 1716 R 100.1 0.0 13:19.56 milkyway_separa
29351 rwatkins 39 19 5032 3328 408 R 99.8 0.0 30:21.01 milkyway_separa
29357 rwatkins 39 19 5032 3324 408 R 99.8 0.0 30:20.91 milkyway_separa
29360 rwatkins 39 19 5032 3324 408 R 99.8 0.0 30:20.94 milkyway_separa
29365 rwatkins 39 19 5032 3324 408 R 99.8 0.0 30:20.88 milkyway_separa
29405 rwatkins 39 19 18172 3184 1432 R 99.8 0.0 14:26.95 milkyway_separa
29445 rwatkins 39 19 18172 3468 1716 R 99.8 0.0 12:42.13 milkyway_separa
29451 rwatkins 39 19 18172 3468 1716 R 99.8 0.0 12:35.11 milkyway_separa
29199 rwatkins 24 4 172m 17m 5668 S 4.5 0.0 0:26.90 boinc_client

Im really pissed again...
My 128 core box is only running 16 threads.
I just can't seem to figure out why.

This is why I requested you post the contents of global_prefs.xml.

There seems to be a setting of 12.5% somewhere in your preferences.
128 x 0.125 = 16
24 x 0.125 = 3

I don't have the override file. Only "global_prefs.xml". PS, these files are really hard to read...

> cat global_prefs.xml
<global_preferences>

<source_project>http://setiathome.berkeley.edu/</source_project>
<source_scheduler>http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi</source_scheduler>

<mod_time>1429302363</mod_time>
<max_ncpus_pct>13</max_ncpus_pct>
<cpu_usage_limit>100</cpu_usage_limit>
<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>3</idle_time_to_run>
<suspend_if_no_recent_input>0</suspend_if_no_recent_input>
<suspend_cpu_usage>25</suspend_cpu_usage>
<work_buf_min_days>0.1</work_buf_min_days>
<work_buf_additional_days>0.5</work_buf_additional_days>
<cpu_scheduling_period_minutes>60</cpu_scheduling_period_minutes>
<disk_interval>60</disk_interval>
<disk_max_used_gb>0</disk_max_used_gb>
<disk_min_free_gb>1</disk_min_free_gb>
<disk_max_used_pct>90</disk_max_used_pct>
<ram_max_used_busy_pct>50</ram_max_used_busy_pct>
<ram_max_used_idle_pct>90</ram_max_used_idle_pct>
<leave_apps_in_memory>0</leave_apps_in_memory>
<vm_max_used_pct>75</vm_max_used_pct>
<max_bytes_sec_down>0</max_bytes_sec_down>
<max_bytes_sec_up>0</max_bytes_sec_up>
<daily_xfer_limit_mb>0</daily_xfer_limit_mb>
<daily_xfer_period_days>0</daily_xfer_period_days>
<dont_verify_images>0</dont_verify_images>
<confirm_before_connecting>0</confirm_before_connecting>
</global_preferences>

I don't have the override file. Only "global_prefs.xml". PS, these files are really hard to read...

> cat global_prefs.xml
<global_preferences>

<source_project>http://setiathome.berkeley.edu/</source_project>
<source_scheduler>http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi</source_scheduler>

<mod_time>1429302363</mod_time>
<max_ncpus_pct>13</max_ncpus_pct>
<cpu_usage_limit>100</cpu_usage_limit>
<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>3</idle_time_to_run>
<suspend_if_no_recent_input>0</suspend_if_no_recent_input>
<suspend_cpu_usage>25</suspend_cpu_usage>
<work_buf_min_days>0.1</work_buf_min_days>
<work_buf_additional_days>0.5</work_buf_additional_days>
<cpu_scheduling_period_minutes>60</cpu_scheduling_period_minutes>
<disk_interval>60</disk_interval>
<disk_max_used_gb>0</disk_max_used_gb>
<disk_min_free_gb>1</disk_min_free_gb>
<disk_max_used_pct>90</disk_max_used_pct>
<ram_max_used_busy_pct>50</ram_max_used_busy_pct>
<ram_max_used_idle_pct>90</ram_max_used_idle_pct>
<leave_apps_in_memory>0</leave_apps_in_memory>
<vm_max_used_pct>75</vm_max_used_pct>
<max_bytes_sec_down>0</max_bytes_sec_down>
<max_bytes_sec_up>0</max_bytes_sec_up>
<daily_xfer_limit_mb>0</daily_xfer_limit_mb>
<daily_xfer_period_days>0</daily_xfer_period_days>
<dont_verify_images>0</dont_verify_images>
<confirm_before_connecting>0</confirm_before_connecting>
</global_preferences>

Your preferences are set at Seti (see blue hilight). Go to the Seti site and change the following under your account: Computing preferences, Default preferences, Use as most ___ % of the CPUs to 100. Right now it is set to 13% (see red hilight).

may not work if seti is no longer attached

may have to make change at mw@h

Thanks for pointing that out. I thought about that a while after posting, but I was out so couldn't post it.