Hello everyone
I've encountered a problem yesterday. My BOINC just stopped using my GPU for no reason.
This is my setup:
Intel Core i7 @ 3.4 GHz Boost up to 4 GHz
32 GB RAM
NVIDIA GeForce GTX 970 (Gainward Phantom Edition)
Windows 10
BOINC 7.6.22 x64 --> not running as a service
asteroids@home

So I've been running BOINC since a few days and everything was going fine, both my GPU and my CPU were always in use. I did not change any settings, but the GPU just suddenly stopped being used by BOINC.
The Event-Log says that it detected the GPU, so that can't be the problem.
The GPU is still working (I tested it multiple times) so that is also not the problem.
I tried changing the local and the Web-Settings so it would really use the GPU but that also did not change anything at all, the GPU ist still sitting there and doing nothing while the CPU gets all the work. I'm honestly at the edge of giving up, because no matter what I do, I just can't find a way to make it work again.

Could you please exit BOINC & restart it, then post the first 40 lines or so from the event log? That gives us a lot more information.

Asteroids@Home has work for your GPU?
Your preferences didn't change?
Windows 10 didn't install newer drivers?

---

30.04.2016 16:42:55 | | cc_config.xml not found - using defaults
30.04.2016 16:42:55 | | Starting BOINC client version 7.6.22 for windows_x86_64
30.04.2016 16:42:55 | | log flags: file_xfer, sched_ops, task
30.04.2016 16:42:55 | | Libraries: libcurl/7.45.0 OpenSSL/1.0.2d zlib/1.2.8
30.04.2016 16:42:55 | | Data directory: C:\ProgramData\BOINC
30.04.2016 16:42:55 | | Running under account Jan
30.04.2016 16:42:56 | | CUDA: NVIDIA GPU 0: GeForce GTX 970 (driver version 364.72, CUDA version 8.0, compute capability 5.2, 4096MB, 3347MB available, 4340 GFLOPS peak)
30.04.2016 16:42:56 | | OpenCL: NVIDIA GPU 0: GeForce GTX 970 (driver version 364.72, device version OpenCL 1.2 CUDA, 4096MB, 3347MB available, 4340 GFLOPS peak)
30.04.2016 16:42:56 | | Host name: predator
30.04.2016 16:42:56 | | Processor: 8 GenuineIntel Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz [Family 6 Model 60 Stepping 3]
30.04.2016 16:42:56 | | 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 fma cx16 sse4_1 sse4_2 movebe popcnt aes f16c rdrandsyscall nx lm avx avx2 vmx smx tm2 pbe fsgsbase bmi1 smep bmi2
30.04.2016 16:42:56 | | OS: Microsoft Windows 10: Core x64 Edition, (10.00.10586.00)
30.04.2016 16:42:56 | | Memory: 31.92 GB physical, 36.67 GB virtual
30.04.2016 16:42:56 | | Disk: 117.99 GB total, 31.62 GB free
30.04.2016 16:42:56 | | Local time is UTC +2 hours
30.04.2016 16:42:56 | | VirtualBox version: 5.0.16
30.04.2016 16:42:56 | Asteroids@home | URL http://asteroidsathome.net/boinc/; Computer ID 314131; resource share 100
30.04.2016 16:42:56 | | General prefs: from http://wuprop.boinc-af.org/ (last modified 24-Apr-2016 04:27:09)
30.04.2016 16:42:56 | | Host location: none
30.04.2016 16:42:56 | | General prefs: using your defaults
30.04.2016 16:42:56 | | Reading preferences override file
30.04.2016 16:42:56 | | Preferences:
30.04.2016 16:42:56 | | max memory usage when active: 24515.69MB
30.04.2016 16:42:56 | | max memory usage when idle: 29418.83MB
30.04.2016 16:42:56 | | max disk usage: 25.00GB
30.04.2016 16:42:56 | | suspend work if non-BOINC CPU load exceeds 80%
30.04.2016 16:42:56 | | (to change preferences, visit a project web site or select Preferences in the Manager)
30.04.2016 16:42:56 | | Contacting account manager at http://pool.gridcoin.co/
30.04.2016 16:42:58 | | Account manager: Welcome to the GridCoin.co Pool beta test! Please report any problems to Katiee on #GridCoin IRC.
30.04.2016 16:42:58 | | Account manager: Login Successful
30.04.2016 16:42:58 | | Account manager: Updated Host (791dde3bf037fe1a2304c1ce7de331d6) | ID: 2814
30.04.2016 16:42:58 | | Account manager: Only 1 project in client
30.04.2016 16:42:58 | | Account manager: Checking project(s)..
30.04.2016 16:42:58 | | Account manager: 1 Project(s) updated to Pool
30.04.2016 16:42:58 | | Account manager: 1 Project(s) settings refreshed from Pool
30.04.2016 16:42:58 | | Account manager contact succeeded

These is my event log after restarting BOINC.
I chose asteroids@home because it is the only Gridcoin-whitelisted NVIDIA-GPU project. Besides that, it used to work before (I alway only used asteroids@home as project) so I'd say yes, it has work.
My preferences are still the same.
I manage my drivers manually, they haven't changed since I installed BOINC. I'm running the most recent available driver for my GPU.

What does the Event Log say when you click 'update' for Asteroids? Asteroids must not be suspended and new work must be allowed.

All right, can you post the next 40 lines after that, please? :-)

---

01.05.2016 19:26:38 | Asteroids@home | update requested by user
01.05.2016 19:26:41 | Asteroids@home | Sending scheduler request: Requested by user.
01.05.2016 19:26:41 | Asteroids@home | Not requesting tasks: don't need (CPU: job cache full; NVIDIA GPU: )
01.05.2016 19:26:43 | Asteroids@home | Scheduler request completed

This is the log after requesting an update. I noticed that apparently, there is no job cache for the NVIDIA GPU. What does that mean? I know for sure that asteroids@home has work for my type of GPU because I have a friend who does BOINC too with the exact same GPU as I have.

I updated the project and posted the log after that in this thread. Maybe it helps already. I'll run BOINC for a longer time tomorrow, so I'll post a longer log tomorrow if the current one isn't helpful.

I noticed that apparently, there is no job cache for the NVIDIA GPU. What does that mean?

BOINC only has one cache, where it stores all tasks for CPU and GPU. Your cache is full with CPU tasks. Only until it has lowered the amount of tasks in the cache to below the low water mark (the value asked for in the first preference "Store at least N days of work") will BOINC ask for more work again.

Normally GPU tasks are loaded first to fill the cache, CPU tasks second.
Perhaps that you've run through the GPU work already and are now running through your CPU cache.

And if you only want to run with the GPU, you can set so in the Asteroid project preferences. (Normally, uncheck CPU. Any work in cache for the CPU still has to be run or aborted, but no more work will be asked).

---