Hello,

I have two GPU's running tasks under BOINC.
The main one is an NVIDIA (the one I play game on too),
the second one is an AMD one.

I use the following config in my cc_config.xml:

<cc_config>
<options>
<exclusive_app>cpuocct64.exe</exclusive_app>
<exclusive_app>OCCT.exe</exclusive_app>
<exclusive_app>Crysis.exe</exclusive_app>
<exclusive_app>bf4.exe</exclusive_app>
<exclusive_app>SC2_x64.exe</exclusive_app>
</options>
</cc_config>


To autopause crunching while I play the game.

What I would like is to exclude the AMD gpu, to make it crunch when I play
a game, because it is not used for 3d rendering.

Is it possible to control that in the cc_config ?

I was thinking about <exclude_gpu> but that will not make it I think.

Thanks in advance for your help.

Note: I use windows 10 4 bits.

http://boincstats.com/en/stats/-5/user/detail/839477/projectList

Is it possible to control that in the cc_config ?

I think the short answer is no, see here

You could set up a pre-gaming script to restart boinc in AMD only mode such as described here

Hi,
I created a cmd script to switch between two different configurations as suggested.

But my config where I want to disable the NVIDIA is not working,
it looks like it is ignored (no suggestion on reading this in the log file):

<cc_config>
<options>
<exclusive_app>cpuocct64.exe</exclusive_app>
<exclusive_app>OCCT.exe</exclusive_app>
<exclusive_app>Crysis.exe</exclusive_app>
</options>
</cc_config>

<exclude_gpu>
<type>nvidia</type>]
</exclude_gpu>



Can't I exclude the entire NVIDIA ?
Am I obliged to do it for all projects ?


Below the script I use:

echo "for info: boinc exe are installed in the E: drive"
echo "for info: boinc data are installed in the E:\BOINC drive location"

cd E:\
E:

e:boinccmd --get_project_status


cd E:\BOINC\
E:



IF EXIST checkpoint.txt (
echo checkpoint.txt exists
echo now restoring to all gpu usage
copy /Y cc_config_all_gpu.xml cc_config.xml
del checkpoint.txt
) ELSE (
echo checkpoint.txt does not exist
echo now using non nvidia gpu usage
copy /Y cc_config_exclude_nvidia.xml cc_config.xml
echo this_is_for_remembering_gpu_usage_selection>checkpoint.txt
)


cd E:\
E:


echo "now reading the new configuration file for BOINC"
e:boinccmd --read_cc_config

Looks like this is not working either using sub project details ...


<exclude_gpu>
<url>http://setiathome.berkeley.edu/</url>
<device_num>0</device_num>
<type>nvidia</type>
<app>einsteinbinary_BRP4</app>
</exclude_gpu>

<exclude_gpu>
<url>http://boinc.fzk.de/poem/</url>
<device_num>0</device_num>
<type>nvidia</type>
<app>einsteinbinary_BRP4</app>
</exclude_gpu>


<exclude_gpu>
<url>http://www.gpugrid.net/</url>
<device_num>0</device_num>
<type>nvidia</type>
<app>einsteinbinary_BRP4</app>
</exclude_gpu>


<exclude_gpu>
<url>http://einstein.phys.uwm.edu/</url>
<device_num>0</device_num>
<type>nvidia</type>
<app>einsteinbinary_BRP4</app>
</exclude_gpu>



What to do ?

Here is the BOINC log:

28-12-15 20:39:20 | | CUDA: NVIDIA GPU 0: GeForce GTX TITAN X (driver version 361.43, CUDA version 8.0, compute capability 5.2, 4096MB, 4025MB available, 8479 GFLOPS peak)
28-12-15 20:39:20 | | OpenCL: NVIDIA GPU 0: GeForce GTX TITAN X (driver version 361.43, device version OpenCL 1.2 CUDA, 12288MB, 4025MB available, 8479 GFLOPS peak)
28-12-15 20:39:20 | | OpenCL: AMD/ATI GPU 0: Hawaii (driver version 1912.5 (VM), device version OpenCL 2.0 AMD-APP (1912.5), 4096MB, 4096MB available, 3200 GFLOPS peak)
28-12-15 20:39:20 | | OpenCL CPU: Intel(R) Core(TM) i7-5930K CPU @ 3.50GHz (OpenCL driver vendor: Advanced Micro Devices, Inc., driver version 1912.5 (sse2,avx), device version OpenCL 1.2 AMD-APP (1912.5))


28-12-15 21:01:21 | | Re-reading cc_config.xml
28-12-15 21:01:21 | | Not using a proxy
28-12-15 21:01:21 | | Config: don't compute while cpuocct64.exe is running
28-12-15 21:01:21 | | Config: don't compute while OCCT.exe is running
28-12-15 21:01:21 | | Config: don't compute while Crysis.exe is running
28-12-15 21:01:21 | | log flags: file_xfer, sched_ops, task
28-12-15 21:03:25 | SETI@home | Sending scheduler request: To fetch work.

But my config where I want to disable the NVIDIA is not working,
it looks like it is ignored (no suggestion on reading this in the log file):

<cc_config>
<options>
<exclusive_app>cpuocct64.exe</exclusive_app>
<exclusive_app>OCCT.exe</exclusive_app>
<exclusive_app>Crysis.exe</exclusive_app>
</options>
</cc_config>

<exclude_gpu>
<type>nvidia</type>]
</exclude_gpu>

I think you need to use a different option, or maybe specify a project to exclude_gpu nvidia use.

From the tome of Boinc wiki

 <ignore_nvidia_dev>N</ignore_nvidia_dev>
    Ignore (don't use) a specific NVIDIA GPU. You can ignore more than one. Replaces <ignore_cuda_dev/>. Example:

<ignore_nvidia_dev>0</ignore_nvidia_dev>

will ignore the first NVIDIA GPU in the system. 

Nice script, hope it works.

Oh I got it now,
the command

boinccmd --read_cc_config

seems to only read the commands in the cc_config.xml
that are between the

<cc_config>

and

</cc_config>

The problem is that the

<exclude_gpu> or <ignore_nvidia_dev> are outside the one above.

When I include it inside, then I get an error in the log:
(sorry for the french :-)

28-12-15 21:09:28 | | Re-reading cc_config.xml
28-12-15 21:09:28 | | Balise non reconnue dans cc_config.xml: <exclude_gpu>
28-12-15 21:09:28 | | Balise non reconnue dans cc_config.xml: <exclude_gpu>
28-12-15 21:09:28 | | Balise non reconnue dans cc_config.xml: <exclude_gpu>
28-12-15 21:09:28 | | Balise non reconnue dans cc_config.xml: <exclude_gpu>

Any idea how to sort this out ?

YAHOOOOOOOOOOO, this is working fine using proper syntax as below ^^



<cc_config>
<options>
<exclusive_app>cpuocct64.exe</exclusive_app>
<exclusive_app>OCCT.exe</exclusive_app>
<exclusive_app>Crysis.exe</exclusive_app>

<exclude_gpu>
<url>http://setiathome.berkeley.edu/</url>
<device_num>0</device_num>
<type>NVIDIA</type>
<app>setiathome_v7</app>
</exclude_gpu>

<exclude_gpu>
<url>http://setiathome.berkeley.edu/</url>
<device_num>0</device_num>
<type>NVIDIA</type>
<app>astropulse_v7</app>
</exclude_gpu>

<exclude_gpu>
<url>http://boinc.fzk.de/poem/</url>
<device_num>0</device_num>
<type>NVIDIA</type>
<app>poempp</app>
</exclude_gpu>

<exclude_gpu>
<url>http://boinc.fzk.de/poem/</url>
<device_num>0</device_num>
<type>NVIDIA</type>
<app>poemcl</app>
</exclude_gpu>

<exclude_gpu>
<url>http://www.gpugrid.net/</url>
<device_num>0</device_num>
<type>NVIDIA</type>
<app>acemdshort</app>
</exclude_gpu>

<exclude_gpu>
<url>http://einstein.phys.uwm.edu/</url>
<device_num>0</device_num>
<type>NVIDIA</type>
<app>einsteinbinary_BRP6</app>
</exclude_gpu>

<exclude_gpu>
<url>http://einstein.phys.uwm.edu/</url>
<device_num>0</device_num>
<type>NVIDIA</type>
<app>hsgamma_FGRP4</app>
</exclude_gpu>


</options>
</cc_config>




Result in the BOINC log:
28-12-15 21:25:48 | SETI@home | Config: excluded GPU. Type: NVIDIA. App: setiathome_v7. Device: 0
28-12-15 21:25:48 | SETI@home | Config: excluded GPU. Type: NVIDIA. App: astropulse_v7. Device: 0
28-12-15 21:25:48 | Poem@Home | Config: excluded GPU. Type: NVIDIA. App: poempp. Device: 0
28-12-15 21:25:48 | Poem@Home | Config: excluded GPU. Type: NVIDIA. App: poemcl. Device: 0
28-12-15 21:25:48 | GPUGRID | Config: excluded GPU. Type: NVIDIA. App: acemdshort. Device: 0
28-12-15 21:25:48 | Einstein@Home | Config: excluded GPU. Type: NVIDIA. App: einsteinbinary_BRP6. Device: 0
28-12-15 21:25:48 | Einstein@Home | Config: excluded GPU. Type: NVIDIA. App: hsgamma_FGRP4. Device: 0

If you care to read the wiki it is obvious that the exclude_gpu section has to be inside the options section.

Edit, ah the reading did work, or maybe it did not ;0)

---

Coelum Non Animum Mutant, Qui Trans Mare Currunt

Oh lol,

When I re read the config where the exclude_gpu is not in,
they are still disabled ^^

How to force re-enabling them ? ^^

Note: Exiting BOINC and restarting it fixes the side effect,
but is there a way using the config read ?

Config read does not work for everything... client restart is often required.

---

Coelum Non Animum Mutant, Qui Trans Mare Currunt

Config read does not work for everything... client restart is often required.

Ok, I found the command to exit the client,
but I don't know how to start the client,
like when it starts with windows or when I click the BOINC icon (I want the GUI icon on the bottom right bar).

but I don't know how to start the client,
like when it starts with windows or when I click the BOINC icon (I want the GUI icon on the bottom right bar).

Add this:

"C:\program files\boinc\boincmgr.exe"

For if you want to start BOINC Manager, which starts the client.

Or

"C:\program files\boinc\boinc.exe" --detach_console

For if you want to just start the client.

We're putting quotes around the path, because of the space in the path.
--detach_console closes the command line window.

Of course, this depends on whether a French version names the directory like this as well. Else rename to the French version. It also depends on whether you installed BOINC in the default directory, else change the path to where you did install it to.

---

but I don't know how to start the client,
like when it starts with windows or when I click the BOINC icon (I want the GUI icon on the bottom right bar).

Add this:

"C:\program files\boinc\boincmgr.exe"

For if you want to start BOINC Manager, which starts the client.

Or

"C:\program files\boinc\boinc.exe" --detach_console

For if you want to just start the client.

We're putting quotes around the path, because of the space in the path.
--detach_console closes the command line window.

Of course, this depends on whether a French version names the directory like this as well. Else rename to the French version. It also depends on whether you installed BOINC in the default directory, else change the path to where you did install it to.

The intersection of Windows, boinc and i is not a large knowledge space, so i defer to Ageless here. I did open a VM though and looked at the dark side - to see what exactly the sequence of steps (who starts what) and with what parameters.

i see from "wmic process" (win10)

boinc.exe                "C:\Program Files\BOINC\boinc.exe" --redirectio --launched_by_manager

on Ubuntu 14.04 "ps" shows

/usr/bin/boinc --check_all_logins --redirectio --dir /var/lib/boinc-client

I don't know if this is helpful (or confusing) - but the -redirectio sends output to the log files, and the --dir is the boinc data directory.

Again the boinc wiki has details.

I don't know if this is helpful (or confusing) - but the -redirectio sends output to the log files, and the --dir is the boinc data directory.

Under Windows, starting the client (boinc.exe) only will always write the output to the log file (stdoutdae.txt), so no need for the --redirectio, that's one for Linux (or all Unix) specifically.

--launched_by_manager is shown by process grabbers if the client was launched by BOINC Manager. One can also manually just start the client, without the manager.
Is that still being used by BOINC 7.6 though? I don't see it in the Help output of BOINC 7.6.22

The command-line options for boinc are intended for debugging.
The recommended command-line interface is a separate program,'boinccmd'.
Run boinccmd in the same directory as boinc.

Usage: boinc [options]
    --abort_jobs_on_exit           when client exits, abort and report jobs
    --allow_remote_gui_rpc         allow remote GUI RPC connections
    --allow_multiple_clients       allow >1 instances per host
    --attach_project <URL> <key>   attach to a project
    --check_all_logins             for idle detection, check remote logins too
    --daemon                       run as daemon (Unix)
    --detach_console               detach from console (Windows)
    --detach_project <URL>         detach from a project
    --dir <path>                   use given dir as BOINC home
    --exit_after_app_start N       exit N seconds after an app starts
    --exit_after_finish            exit right after finishing a job
    --exit_before_start            exit right before starting a job
    --exit_before_upload           exit right before starting an upload
    --exit_when_idle               exit when there are no results
    --fetch_minimal_work           fetch only 1 job per device
    --file_xfer_giveup_period N    give up on file xfers after N sec
    --gui_rpc_port <port>          port for GUI RPCs
    --gui_rpc_unix_domain          use Unix domain for GUI RPCs
    --help                         show options
    --insecure                     disable app sandboxing (Unix)
    --master_fetch_interval N      limiting period of master retry
    --master_fetch_period N        reload master URL after N RPC failures
    --master_fetch_retry_cap N     exponential backoff limit
    --no_gpus                      don't check for GPUs
    --no_gui_rpc                   don't allow GUI RPC, don't make socket
    --no_info_fetch                don't fetch project list or client version info
    --no_priority_change           run apps at same priority as client
    --pers_giveup N                giveup time for persistent file xfer
    --pers_retry_delay_max N       max for file xfer exponential backoff
    --pers_retry_delay_min N       min for file xfer exponential backoff
    --redirectio                   redirect stdout and stderr to log files
    --reset_project <URL>          reset (clear) a project
    --retry_cap N                  exponential backoff limit
    --run_cpu_benchmarks           run the CPU benchmarks
    --run_by_updater               set by updater
    --sched_retry_delay_max N      max for RPC exponential backoff
    --sched_retry_delay_min N      min for RPC exponential backoff
    --show_projects                show attached projects
    --skip_cpu_benchmarks          don't run CPU benchmarks
    --start_delay X                delay starting apps for X secs
    --suppress_net_info            don't send network addrs to server
    --unsigned_apps_ok             allow unsigned apps (for testing)
    --update_prefs <URL>           contact a project to update preferences
    --version                      show version info

---

I don't know if this is helpful (or confusing) - but the -redirectio sends output to the log files, and the --dir is the boinc data directory.

Under Windows, starting the client (boinc.exe) only will always write the output to the log file (stdoutdae.txt), so no need for the --redirectio, that's one for Linux (or all Unix) specifically.

--launched_by_manager is shown by process grabbers if the client was launched by BOINC Manager. One can also manually just start the client, without the manager.
Is that still being used by BOINC 7.6 though? I don't see it in the Help output of BOINC 7.6.22

If we mean "Is boincmgr.exe starting boinc with those parameters?" yes absolutely at 7.6.9

If you mean "Does boinc.exe ignore or behave differently using that parameter?" hmmm - a quick look at source code suggests that it is not ignored (anywhere), and has some function in gui_rpc_server_ops.cpp and client_state.cpp especially in the (even darker) Apple world.

Hi,
Using the exit for the client makes the gui offuscated after 3 kills,
it gives a warning and the tasks gets aborted (GPU missing, aborted).

29/12/2015 09:22:18 | Poem@Home | [error] Missing coprocessor for task poempp_1vii_1451334726_1095421_0; aborting



Is there another option to close the GUI before to avoid this bad situation ?

It looks like restarting is not really an option now (seeing the side effect on gpu tasks behing aborted automatically when boinc is not allowed to use it).

I think we need an extra option in the cc_config.xml on top of <exclude_gpu>,
something like <include_gpu>.

Or the config parser should be smart enough to activate the GPU work when parsing again if the <exclude_gpu> option is absent.


cd E:\
E:

echo "now exiting BOINC"
e:boinccmd --quit
echo "now starting BOINC"
e:boincmgr.exe

I have switched the gui to english ^^

To make certain GPU detection decisions a restart of the client (boinc.exe) is required. The GUI (boincmgr.exe) does not need to restart.
If we're only starting one instance of a BOINC + BOINC Manager, they both use the default RPC port to communicate with each other.

This means that you can exit the client and restart it, without having to exit and restart the GUI. As soon as the client restarts it will populate the already running BOINC Manager again.

In other words, there is no need to close the GUI through a script. It's reusable.

I have just tried 7 times in a row to quit BOINC from command line, leave the manager running and restart BOINC from a shortcut on my desktop. It repopulated the GUI every time, although some times it would take longer than others.

You can of course force BOINC to run with the same port number as the GUI, by doing:

boinc.exe --gui_rpc_port 31416

---

I also noticed that it takes like 1 minute to reconnect to the locahost.

I still get the error from boinc manager using this:

echo "now exiting BOINC"
e:boinccmd --quit
echo "now starting BOINC"
e:boinc.exe

echo "now exiting BOINC"
e:boinccmd --quit
echo "now starting BOINC"
e:boinc.exe

Why are you not using the full path to the BOINC executable on E: ?
And why aren't you using the dividers '\'?

Normally, you either added the full path to the BOINC directory to the %PATH% variable, or you point to the executable using the full path, like e:\the directory BOINC is installed in\boinc.exe

And just for the sake of making sure it's continuing to use the correct port number, do use e:\the directory BOINC is installed in\boinc.exe --gui_rpc_port 31416

Even if you managed to put BOINC in the root of the E: drive, you'll use e:\boinc.exe, with the back-slash.

---

BOINC is not in the PATH list.

E: Is my ramdrive dedicated to BOINC, boinc.exe is in the root.

Anyway, this is syntax stuff, not related to the problem I mention.