Hi,

after many years, I'm back at SETI@home.
I have several older computers and installed BOINC, but nowhere the GPU is used.

Computer 1:
https://setiathome.berkeley.edu/show_host_detail.php?hostid=8048670
Celeron J1900
16GB DDR3-RAM
Intel HD Graphics (onboard)
Win 10 Pro x64

Computer 2:
https://setiathome.berkeley.edu/show_host_detail.php?hostid=8048676
Core i7 840QM
24 GB DDR3-RAM
Nvidia Quadro FX 2800M, 1 GB (according to Wikipedia capable of CUDA 1.1)
Win 10 Pro x64

Computer 3:
https://setiathome.berkeley.edu/show_host_detail.php?hostid=8049006
Core 2 Duo E6600
3 GB RAM
Geforce 7900 GTO, 512 MB (without CUDA capability)
Win 10 Pro x86


The GPUs are not in use by BOINC/SETI@home.
Why? Too old?

The GPUs are not in use by BOINC/SETI@home.
Why? Too old?

Pretty much.

Neither the FX 2800M or the card it's based on (the Geforce 9800M GT) are listed as CUDA capable on NVidia's web site.


The Celeron, with the right drivers, should support OpenCL 1.2
However even on more powerful hardware the Intel drivers implementation of OpenCL leaves a lot to be desired.

---

Grant
Darwin NT

The Celeron, with the right drivers, should support OpenCL 1.2
However even on more powerful hardware the Intel drivers implementation of OpenCL leaves a lot to be desired.

I installed the Intel OpenCL runtime 16.1 on Computer 1.
Now GPU-Z shows me OpenCL capability.
Then I restarted the computer, but the GPU is still not in use.

Log:
cc_config.xml not found - using defaults
Starting BOINC client version 7.6.22 for windows_x86_64
log flags: file_xfer, sched_ops, task
Libraries: libcurl/7.45.0 OpenSSL/1.0.2d zlib/1.2.8
Running as a daemon (GPU computing disabled)
Running under account boinc_master
No usable GPUs found

Do I have to reinstall the BOINC client?

Running under account boinc_master
No usable GPUs found

Do I have to reinstall the BOINC client?

Yep, it's been installed as a service which makes GPU computing not possible.
Running as a daemon (GPU computing disabled)
That's why no suitable GPU devices are listed.

From http://boinc.berkeley.edu/wiki/GPU_computing
Warning: On Windows do not install BOINC in Protected Access Execution (PAE) mode aka service mode (6.4.5 - 7.0.28) or Service Install mode (7.0.64 and above). If you do, BOINC will not be able to detect or use your GPU.

Re-install, not as a service, and then things should be Ok.

EDIT- will be interesting to see if the Quadro makes the cut. On the NVidia page that lists cards that support CUDA, it's not listed. But on the card's own web page, it's listed as having 96 CUDA cores.

---

Grant
Darwin NT

OK, thank you for your support.

Computer 3:
Too old. No CUDA, no OpenCL, no chance.

Computer 2:
new Nvidia driver installed
BONIC re-installed as non-service
-> Now it works! The GPU was recognized and is computing. (OpenCL version 1.0)

Computer 1:
Intel OpenCL runtime 16.1 installed
BONIC re-installed as non-service
-> The GPU was recognized and tasks were downloaded, but GPU still doesn't work.

Now I have many error tasks:
https://setiathome.berkeley.edu/results.php?hostid=8048670&offset=0&show_names=0&state=6&appid=

Then I did:
- driver repair installaton
- restart
- BONIC re-installed again
- restart
-> The GPU still doesn't work.

Last Log:
19.07.2016 12:57:53 | | cc_config.xml not found - using defaults
19.07.2016 12:57:53 | | Starting BOINC client version 7.6.22 for windows_x86_64
19.07.2016 12:57:53 | | log flags: file_xfer, sched_ops, task
19.07.2016 12:57:53 | | Libraries: libcurl/7.45.0 OpenSSL/1.0.2d zlib/1.2.8
19.07.2016 12:57:53 | | Data directory: C:\Programme\BOINC\Daten
19.07.2016 12:57:53 | | Running under account xxxxx
19.07.2016 12:57:53 | | OpenCL: Intel GPU 0: Intel(R) HD Graphics (driver version 10.18.10.4252, device version OpenCL 1.2, 1221MB, 1221MB available, 358 GFLOPS peak)
19.07.2016 12:57:53 | | OpenCL CPU: Intel(R) Celeron(R) CPU J1900 @ 1.99GHz (OpenCL driver vendor: Intel(R) Corporation, driver version 5.2.0.10002, device version OpenCL 1.2 (Build 10002))
19.07.2016 12:57:53 | | Host name: xxxxxx
19.07.2016 12:57:53 | | Processor: 4 GenuineIntel Intel(R) Celeron(R) CPU J1900 @ 1.99GHz [Family 6 Model 55 Stepping 3]
19.07.2016 12:57:53 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 sse4_2 movebe popcnt rdrandsyscall nx lm vmx tm2 pbe smep
19.07.2016 12:57:53 | | OS: Microsoft Windows 10: Professional x64 Edition, (10.00.10586.00)
19.07.2016 12:57:53 | | Memory: 15.71 GB physical, 18.08 GB virtual
19.07.2016 12:57:53 | | Disk: 223.23 GB total, 68.01 GB free
19.07.2016 12:57:53 | | Local time is UTC +2 hours
19.07.2016 12:57:53 | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 8048670; resource share 100
19.07.2016 12:57:53 | SETI@home | General prefs: from SETI@home (last modified 19-Jul-2016 11:58:19)
19.07.2016 12:57:53 | SETI@home | Host location: none
19.07.2016 12:57:53 | SETI@home | General prefs: using your defaults
19.07.2016 12:57:53 | | Reading preferences override file
19.07.2016 12:57:53 | | Preferences:
19.07.2016 12:57:53 | | max memory usage when active: 8042.95MB
19.07.2016 12:57:53 | | max memory usage when idle: 14477.32MB
19.07.2016 12:57:53 | | max disk usage: 68.02GB
19.07.2016 12:57:53 | | suspend work if non-BOINC CPU load exceeds 25%
19.07.2016 12:57:53 | | (to change preferences, visit a project web site or select Preferences in the Manager)
19.07.2016 12:57:58 | SETI@home | Sending scheduler request: To fetch work.
19.07.2016 12:57:58 | SETI@home | Requesting new tasks for Intel GPU
19.07.2016 12:57:59 | SETI@home | Scheduler request completed: got 0 new tasks
19.07.2016 12:58:03 | | Suspending computation - CPU is busy
19.07.2016 12:58:13 | | Resuming computation
19.07.2016 12:58:23 | | Suspending computation - CPU is busy
19.07.2016 12:59:04 | | Resuming computation
19.07.2016 13:02:25 | | Remote desktop in use; disabling GPU tasks
19.07.2016 13:02:34 | | Suspending computation - CPU is busy
19.07.2016 13:02:44 | | Resuming computation
19.07.2016 13:03:59 | SETI@home | update requested by user
19.07.2016 13:04:00 | SETI@home | Sending scheduler request: Requested by user.
19.07.2016 13:04:00 | SETI@home | Not requesting tasks: don't need (CPU: job cache full; Intel GPU: GPUs not usable)
19.07.2016 13:04:02 | SETI@home | Scheduler request completed



GPU not usable? Why?

19.07.2016 13:02:25 | | Remote desktop in use; disabling GPU tasks

GPU not usable? Why?

Because you are using Windows' remote desktop connection which replaces the GPU driver with a basic one that doesn't support GPU computing. You have to use something else for remote management. VNC, for example, should for fine.

You can also set up BOINC so that it can be controlled remotely.

Chrome Remote Desktop works fine also.

BTW: I wouldn't use the GPU, just the 4 cores. Using all cores plus the GPU will produce a lot of heat (possibly throttling the CPU) and you may even end up with less work done because the GPU will use some CPU also. Just my 2 cents.

---

Because you are using Windows' remote desktop connection which replaces the GPU driver with a basic one that doesn't support GPU computing.

OK, that could be an explanation.
But, why the GPU doesn't work immediately after the restart? (there is a auto-login)
There is no remote connection yet.

You have to use something else for remote management.

:-(
I need the Windows remote desktop for other reasons.

Chrome Remote Desktop works fine also.

Schleich di mit dem Sauglump. ;-)

BTW: I wouldn't use the GPU, just the 4 cores. Using all cores plus the GPU will produce a lot of heat (possibly throttling the CPU) and you may even end up with less work done because the GPU will use some CPU also. Just my 2 cents.

OK, that's a good point.
At the moment my J1900 is running in turbo mode.

But, why the GPU doesn't work immediately after the restart? (there is a auto-login)
There is no remote connection yet.

There is usually a message in the Event Log that states the reason for such things.

The log snippet you posted earlier is a bit weird though. Supposedly you have cache full of CPU tasks yet there is no message saying the client is starting any CPU tasks. I can't see any reason why it would not have started CPU tasks during the time the CPU was not busy.

Because you are using Windows' remote desktop connection which replaces the GPU driver with a basic one that doesn't support GPU computing.

OK, that could be an explanation.
But, why the GPU doesn't work immediately after the restart? (there is a auto-login)
There is no remote connection yet.

You have to use something else for remote management.

:-(
I need the Windows remote desktop for other reasons.

Chrome Remote Desktop works fine also.

Schleich di mit dem Sauglump. ;-)

BTW: I wouldn't use the GPU, just the 4 cores. Using all cores plus the GPU will produce a lot of heat (possibly throttling the CPU) and you may even end up with less work done because the GPU will use some CPU also. Just my 2 cents.

OK, that's a good point.
At the moment my J1900 is running in turbo mode.

My J1900 system has never had a problem running 4 CPU + iGPU tasks while staying in Boost at 2.41GHz. It is an ASRock Q1900-ITX. Which has a fairly large heatsink.
When using an iGPU sometimes there can be problems when Intel updates the driver. Causing SETI@home apps to no longer work or generate garbage results.

Current temps with 4CPU + iGPU at 100% load.
Room 25ºC
CPU 33ºC
iGPU 39ºC
CPU cores 41-42ºC.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

Nvidia CUDA site has a list of acceptable GPU.

I don't track their business/cad gpu.

But their consumer gpu start at 8400 with 256 or more vram. So don't bother with anything older.

---