Na just keep it linux (I need to catch up to pututu and some of the others running only win products) ;-)

I know that this is built with CUDA. If you start working on OpenCL, please know that OpenCL ICD relies on dynamic libraries. It is double edged sword for gpu apps. While you eliminate dependency on specific system libraries, the OpenCL icd on system may depend on version other than what you statically lined in. I know that versioned symbols of libc is supposed to prevent such conflicts, it did not for me on another project. I had to recompile to get rid of sigsegv.
Good luck. And thanks for working on accelerated apps!

Thx for the GPU App.

For me it works fine so far. 75 valid tasks 0 invalid :)

Just low to mid 70% GPU util so I'm running 2x. Now mid 90%.

<app_config>
  <app>
    <name>GetDecics</name>
     <gpu_versions>
      <gpu_usage>0.5</gpu_usage>
      <cpu_usage>0.1</cpu_usage>
     </gpu_versions>
  </app>
</app_config>

Just low to mid 70% GPU util so I'm running 2x. Now mid 90%.

<app_config>
  <app>
    <name>GetDecics</name>
     <gpu_versions>
      <gpu_usage>0.5</gpu_usage>
      <cpu_usage>0.1</cpu_usage>
     </gpu_versions>
  </app>
</app_config>

Interesting... I forgot you could control those things through the client. I just did the same thing. GPU is running a little hotter and my browser is a little sluggish... I might have to go back to running 1x.

So is 70% considered low? Should I try to optimize the utilization? During development, having too many blocks or threads per block caused crashes due to resource problems. I eventually settled on some values that seemed to work well, but maybe they are not optimal.

But I am pressing on with the OpenCL version first; I will worry about optimization later.

Just low to mid 70% GPU util so I'm running 2x. Now mid 90%.

<app_config>
  <app>
    <name>GetDecics</name>
     <gpu_versions>
      <gpu_usage>0.5</gpu_usage>
      <cpu_usage>0.1</cpu_usage>
     </gpu_versions>
  </app>
</app_config>

Interesting... I forgot you could control those things through the client. I just did the same thing. GPU is running a little hotter and my browser is a little sluggish... I might have to go back to running 1x.

So is 70% considered low? Should I try to optimize the utilization? During development, having too many blocks or threads per block caused crashes due to resource problems. I eventually settled on some values that seemed to work well, but maybe they are not optimal.

But I am pressing on with the OpenCL version first; I will worry about optimization later.

Hotter is to be expected. My Ti is in a dedicated system so slow browsing won't bother me.

I'd consider it low, so I added another task even at the expense of another lost CPU thread feeding the GPU exe.

Some other projects have options the user can configure, I'm guessing something like the blocks and threads per block that you mentioned. Collatz, Amicable and even SETI's user apps can have custom settings that can push GPUs harder. Amicable has a kernel size setting on BOINC's user preferences. Collatz and SETI can look at text files in the project folder.

Memory usage in Nvidia X server is showing about 1.5GB with 2 tasks so plenty of resources left. 567mb per task. 0-1% PCI-E usage. Just 115-118 watts via nvidia-smi on a 1080Ti. Same OC uses 185-195 watts with E@H.

Anyone know what's going on at Formula Boinc please? They don't seem to be including points from the new GPU app :-(

There is no way to separate it without reading every users tasks, which is impossible with hidden computers.

Okay ta - maybe it is just taking a while. Nothing has updated yet in fact Formula Boinc, Boincstats and even my new badge on project site all seem to be 24hrs behind

Are points from the GPU app exported please Eric?

There is only once per day external update to FreeDC/BOINCSats but FB can read the team points every hour which worked for the Sprint. Marathon may just look once per day.

Okay ta - maybe it is just taking a while. Nothing has updated yet in fact Formula Boinc, Boincstats and even my new badge on project site all seem to be 24hrs behind

Are points from the GPU app exported please Eric?

I didn't think the GPU credits were treated any different during export. Looking at it a little further, the update_stats is a periodic task run once a day and according to time stamps that occurs at about 6:00 UTC (just a little while ago). So I think that explains the delay. I also grepped for your name and you are definitely in the export file.

All sorted now. Thanks Eric.

If it sounds a little mercenary, the points are the only thing we get for all the money spent on hardware and electricity. Oh wait, make that points plus a warm furry feeling that we are helping to advance science.

I just wanted to thank everyone for your comments, suggestions, etc. It's all very helpful. If I haven't responded enough, it's only because I've been too busy with life's other responsibilities (and learning about OpenCL), but know that I am definitely reading and considering all your suggestions.

Hello :)
I waited a long time for GPU app. Great job!
I don't getting any GPU tasks. I have drivers version 418.39. Linux 4.13.16-hiveos. My machine: https://numberfields.asu.edu/NumberFields/show_host_detail.php?hostid=1465179

5: 23-Mar-2019 10:04:14 (low) [] CUDA: NVIDIA GPU 0: GeForce GTX 1060 3GB (driver version 418.39, CUDA version 10.1, compute capability 6.1, 3019MB, 2943MB available, 4053 GFLOPS peak)
6: 23-Mar-2019 10:04:14 (low) [] CUDA: NVIDIA GPU 1: GeForce GTX 1060 3GB (driver version 418.39, CUDA version 10.1, compute capability 6.1, 3019MB, 2943MB available, 4053 GFLOPS peak)
7: 23-Mar-2019 10:04:14 (low) [] CUDA: NVIDIA GPU 2: GeForce GTX 1060 3GB (driver version 418.39, CUDA version 10.1, compute capability 6.1, 3019MB, 2943MB available, 4053 GFLOPS peak)
8: 23-Mar-2019 10:04:14 (low) [] CUDA: NVIDIA GPU 3: GeForce GTX 1060 3GB (driver version 418.39, CUDA version 10.1, compute capability 6.1, 3019MB, 2943MB available, 4053 GFLOPS peak)
9: 23-Mar-2019 10:04:14 (low) [] CUDA: NVIDIA GPU 4: GeForce GTX 1060 3GB (driver version 418.39, CUDA version 10.1, compute capability 6.1, 3019MB, 2943MB available, 4053 GFLOPS peak)
10: 23-Mar-2019 10:04:14 (low) [] CUDA: NVIDIA GPU 5: GeForce GTX 1060 3GB (driver version 418.39, CUDA version 10.1, compute capability 6.1, 3019MB, 2943MB available, 4053 GFLOPS peak)
11: 23-Mar-2019 10:04:14 (low) [] CUDA: NVIDIA GPU 6: GeForce GTX 1060 3GB (driver version 418.39, CUDA version 10.1, compute capability 6.1, 3019MB, 2943MB available, 4053 GFLOPS peak)
12: 23-Mar-2019 10:04:14 (low) [] CUDA: NVIDIA GPU 7: GeForce GTX 1060 3GB (driver version 418.39, CUDA version 10.1, compute capability 6.1, 3019MB, 2943MB available, 4053 GFLOPS peak)

And I getting this:

74: 23-Mar-2019 10:17:03 (low) [NumberFields@home] Sending scheduler request: Requested by user.
75: 23-Mar-2019 10:17:03 (low) [NumberFields@home] Requesting new tasks for NVIDIA GPU
76: 23-Mar-2019 10:17:05 (low) [NumberFields@home] Scheduler request completed: got 0 new tasks
77: 23-Mar-2019 10:17:05 (user notification) [NumberFields@home] This project is using an old URL.  When convenient, remove the project, then add http://numberfields.asu.edu/NumberFields/

Sorry, I missed this earlier... Did you ever get it to work? Could it have been there was no work available at the time? There should be plenty of work now. I also don't think that old url message should affect anything, I have seen that before and it didn't affect downloading of work. But you could always try resetting the project anyways to see if it helps.

Yes, now it working, probably yesterday there was out of work, but I'm limited by CPU and can use only 4 GPU so i need to buy CPU with more cores :)

Yes, it does seem to take a whole CPU core to keep each GPU WU busy

Yes, now it working, probably yesterday there was out of work, but I'm limited by CPU and can use only 4 GPU so i need to buy CPU with more cores :)

Does it help if you set cpu_usage to .1 in app_config.xml?

Yes it helps :), but also I see one more strange behavior. On default configuration, in BOINC manager I see 5 WUs running:
(0.981 CPUs + 1 NVIDIA GPU (device 0))
(0.981 CPUs + 1 NVIDIA GPU (device 1))
(0.981 CPUs + 1 NVIDIA GPU (device 2))
(0.981 CPUs + 1 NVIDIA GPU (device 3))
(0.981 CPUs + 1 NVIDIA GPU (device 4))
but in practice it working only on one GPU:

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0       840      G   /usr/lib/xorg/Xorg                             6MiB |
|    0     29346      C   ...Decics_3.00_x86_64-pc-linux-gnu__cuda30   195MiB |
|    0     29347      C   ...Decics_3.00_x86_64-pc-linux-gnu__cuda30   195MiB |
|    0     29348      C   ...Decics_3.00_x86_64-pc-linux-gnu__cuda30   195MiB |
|    0     29349      C   ...Decics_3.00_x86_64-pc-linux-gnu__cuda30   195MiB |
|    0     31804      C   ...Decics_3.00_x86_64-pc-linux-gnu__cuda30   195MiB |
|    1       840      G   /usr/lib/xorg/Xorg                             6MiB |
|    2       840      G   /usr/lib/xorg/Xorg                             6MiB |
|    3       840      G   /usr/lib/xorg/Xorg                             6MiB |
|    4       840      G   /usr/lib/xorg/Xorg                             6MiB |
|    5       840      G   /usr/lib/xorg/Xorg                             6MiB |
|    6       840      G   /usr/lib/xorg/Xorg                             6MiB |
|    7       840      G   /usr/lib/xorg/Xorg                             6MiB |
+-----------------------------------------------------------------------------+

^I have seen the same behaviour on a 2-GPU system. It shows one task for each device, but it's actually crunching both on a single card while the other one is left idle.

EDIT: For now I have simply added an exclusion for the second GPU (GTX 980) and assigned it to another project. The first (GTX 1660 Ti) is now crunching two tasks at the same time with a noticeable boost in GPU utilization (and throughput? projected credit/day is 4+ million). Too bad each seems to require a full CPU thread.

Would it be possible for the Server Status page to separate the CPU/GPU WU counts out? So we can tell if there are GPU WU's available?