I recently downloaded and started running S@H. I noted there's a lot of validation inconclusive tasks for the nvidia results in particular. Any ideas?

http://setiathome.berkeley.edu/results.php?hostid=8037488&offset=0&show_names=0&state=3&appid=

TBar would be the one to ask.

Hopefully he will be by sometime.

Well, as Zalster can confirm, the Mac MB OpenCL Apps are basically broken in Mac OS 10.11.x. The Laptops were broken early on, the Desktops took until 10.11.4 before they started producing mostly Inconclusive results. If you want to receive good results with the Apps SETI sends you will have to go back to 10.10.5 on the Laptop and 10.11.3 on the Desktop. Otherwise there are CUDA Apps available that work on nVidia Macs and Most Hackintoshes. Zalster is running one of the older Apps on his Laptop, http://setiathome.berkeley.edu/results.php?hostid=7604284. I was hoping he would try the newer App so we could see if those detection warnings his and some other Laptops have would disappear. The newer Apps should be a little faster on the GUPPI VLARs although they are much slower than the Alpha Apps that are being tested. The currently available CUDA Apps work fine in El Capitan, hopefully the test Apps will be finished soon.

Here is a similar machine running a 980 in Yosemite, http://setiathome.berkeley.edu/results.php?hostid=6811981
The Mac CUDA Apps were submitted to Beta almost a month ago, still nothing on if they will ever be deployed. For now you will have to install the CUDA Apps manually if you want to avoid the Inconclusives.

post a link for the new app and I'll give it a try, Be sure to include any info I might need.

It's disappointing to learn that I'm wasting about half my computer time.

How do I go about manually installing the cuda app?

Here, http://www.arkayn.us/forum/index.php?topic=191.msg4411#msg4411
Since you're running the CUDA65 App all you'll have to do is paste in the New 75 App and change the two lines in the app_info file where the App name is listed. The new App uses the same libraries as the old App. Then just set the file permissions with the Terminal or run the BOINC installer again.

To install the Apps for the first time you will need to set the BOINC Manager to 'No New Tasks' and finish the existing tasks. There are links in the download package to download the Toolkit and extract the two needed libraries. Then paste the two 6.5 libraries and previously downloaded package files into /Library/Application Support/BOINC Data/projects/setiathome.berkeley.edu. Then run the BOINC installer again to set the correct file permissions. To revert back just finish the existing tasks and remove the files you added to setiathome.berkeley.edu.

This morning, the nvidia control panel informed me there was an update for the nvidia driver.

Prior to this I had notice the work units would be (postponed while the boinc manager was open) but once screensaver came on they would crunch

After the driver update, they are crunching while the boinc manager is open.

Greg, any update on your problem?

Were you able to make the changes TBar suggested?

Were you able to make the changes TBar suggested?

I'm waiting for the current tasks to run out then I will try to install the cuda version. Hopefully today.

It's not clear to me exactly what I need to do.

I have installed the cuda 6.5 dev tools, but what libraries do I need to extract?

What do I do with the CUDA75(App_Only).zip or CUDA42.7z files? Do I need both of them?

download the cuda75(app_only)

I believe what you are looking for in the toolbox is

libcudart.6.5.dylib
libcufft.6.5.dylib

Do you know the location of the setiathome folder?

Macintosh HD --> Library --> Application Support --> BOINC Data --> projects --> setiathome.berkeley.edu

The app_info should be in the setiathome folder as well, you will need to edit it with Textedit (right click and open with) to match what is in the zip folder and the libraries

After that, when you restart boinc, you will probably get an error message. Use the boinc installer to reinstall boinc, it corrects the permissions and then it should launch after that.

Hopefully then it will download and start to crunch new work.

It took me several attempts to figure this out but you will get the hang of it.

Here's what mine looks like, no guarantee it will work on yours but yours should look similar

<app_info>
 <app>
   <name>setiathome_v8</name>
 </app>
  <file_info>
    <name>MBv8_8.05r3344_sse41_x86_64-apple-darwin</name>
    <executable/>
  </file_info>
  <app_version>
    <app_name>setiathome_v8</app_name>
    <platform>x86_64-apple-darwin</platform>
    <version_num>800</version_num>
    <file_ref>
      <file_name>MBv8_8.05r3344_sse41_x86_64-apple-darwin</file_name>
      <main_program/>
    </file_ref>
  </app_version>
  <app>
     <name>setiathome_v8</name>
  </app>
    <file_info>
        <name>setiathome_x41zi_x86_64-apple-darwin_cuda75</name>
        <executable/>
    </file_info>
    <file_info>
        <name>libcudart.6.5.dylib</name>
        <executable/>
    </file_info>
    <file_info>
        <name>libcufft.6.5.dylib</name>
        <executable/>
    </file_info>
   <app_version>
        <app_name>setiathome_v8</app_name>
        <platform>x86_64-apple-darwin</platform>
        <version_num>800</version_num>
        <plan_class>cuda65</plan_class>
        <avg_ncpus>0.1</avg_ncpus>
        <max_ncpus>0.1</max_ncpus>
        <coproc>
            <type>CUDA</type>
            <count>1</count>
        </coproc>
        <file_ref>
            <file_name>setiathome_x41zi_x86_64-apple-darwin_cuda75</file_name>
            <main_program/>
         </file_ref>
        <file_ref>
            <file_name>libcudart.6.5.dylib</file_name>
        </file_ref>
        <file_ref>
            <file_name>libcufft.6.5.dylib</file_name>
        </file_ref>
     </app_version>
</app_info>

I found the two libraries.

Copied them to /Library/Application Support/BOINC Data/projects/setiathome.berkeley.edu

There was no app_info file in the sesetiathome.berkeley.edu folder. I coped the one from the CUDA75(App_Only) download.

I found the two libraries.

Copied them to /Library/Application Support/BOINC Data/projects/setiathome.berkeley.edu

There was no app_info file in the sesetiathome.berkeley.edu folder. I coped the one from the CUDA75(App_Only) download.

The package contains everything you need to run a GPU & CPU task, except the two libraries libcudart.6.5.dylib & libcufft.6.5.dylib. Just copy everything that is in the CUDA75(App_Only) folder plus libcudart.6.5.dylib & libcufft.6.5.dylib to the setiathome.berkeley.edu folder. The app_info in is the CUDA75(App_Only) folder, you do not need to modify it unless you want to run a different App.

Looking at Zalster's results it seems the cuda75 App isn't much different than the cuda65 App. For some reason the Laptop is still giving the detection warnings at times. That's something Jason will have to look into.

Yes, it makes sense to me now.

I got everything in the right place and the bonix permission error and then reinstall.

Now Seti is doing the communication deferred thing and not downloading tasks. I will give it some time and see what happens.

All the nvidia stuff got aborted. I am not getting any intel GPU task.

Wed Jul 27 10:02:00 2016 | SETI@home | [error] Missing coprocessor for task blc3_2bit_guppi_57449_40804_HIP78709_OFF_0004.24469.831.17.26.88.vlar_1; aborting
Wed Jul 27 10:02:00 2016 | SETI@home | [error] Missing coprocessor for task blc3_2bit_guppi_57449_40804_HIP78709_OFF_0004.24469.831.17.26.184.vlar_1; aborting
Wed Jul 27 10:02:00 2016 | SETI@home | [error] Missing coprocessor for task blc3_2bit_guppi_57449_40804_HIP78709_OFF_0004.24469.831.17.26.123.vlar_0; aborting
Wed Jul 27 10:02:00 2016 | SETI@home | [error] Missing coprocessor for task blc3_2bit_guppi_57449_40804_HIP78709_OFF_0004.24469.831.17.26.190.vlar_0; aborting
Wed Jul 27 10:02:00 2016 | SETI@home | [error] Missing coprocessor for task blc3_2bit_guppi_57449_40804_HIP78709_OFF_0004.24469.831.17.26.201.vlar_1; aborting
Wed Jul 27 10:02:00 2016 | SETI@home | [error] Missing coprocessor for task blc3_2bit_guppi_57449_40804_HIP78709_OFF_0004.24469.831.17.26.129.vlar_0; aborting
Wed Jul 27 10:02:00 2016 | SETI@home | [error] Missing coprocessor for task blc3_2bit_guppi_57449_41140_HIP78709_0005.24410.831.17.26.190.vlar_1; aborting
Wed Jul 27 10:02:00 2016 | SETI@home | [error] Missing coprocessor for task blc3_2bit_guppi_57449_40804_HIP78709_OFF_0004.24469.831.17.26.196.vlar_0; aborting
Wed Jul 27 10:02:00 2016 | SETI@home | [error] Missing coprocessor for task blc3_2bit_guppi_57449_40804_HIP78709_OFF_0004.24469.831.17.26.135.vlar_0; aborting
Wed Jul 27 10:02:00 2016 | SETI@home | [error] Missing coprocessor for task blc3_2bit_guppi_57449_40804_HIP78709_OFF_0004.24469.831.17.26.94.vlar_1; aborting
Wed Jul 27 10:02:00 2016 | SETI@home | [error] Missing coprocessor for task blc3_2bit_guppi_57449_40469_HIP78709_0003.2186.416.17.26.223.vlar_2; aborting
Wed Jul 27 10:02:00 2016 | SETI@home | [error] Missing coprocessor for task 25my10aa.19272.6243.4.31.10_2; aborting
Wed Jul 27 10:02:00 2016 | SETI@home | [error] Missing coprocessor for task blc3_2bit_guppi_57449_40804_HIP78709_OFF_0004.24469.831.17.26.108.vlar_0; aborting
Wed Jul 27 10:02:00 2016 | SETI@home | [error] Missing coprocessor for task blc3_2bit_guppi_57449_40804_HIP78709_OFF_0004.24469.831.17.26.102.vlar_1; aborting
Wed Jul 27 10:02:00 2016 | SETI@home | [error] Missing coprocessor for task blc3_2bit_guppi_57449_40804_HIP78709_OFF_0004.24469.831.17.26.207.vlar_1; aborting

Your machine isn't showing the nVidia driver anymore. Apparently you installed the Toolkit? You need to reinstall the driver again, http://www.nvidia.com/object/macosx-cuda-7.5.30-driver.html
You do not need to install the toolkit...

The intel GPU uses OpenCL which is Broken in El Capitan, it will give mostly incorrect results. The intel GPU will also cause the CPU cores to run much slower than normally. There are numerous threads about it here. People have found they get better results by not using the intel GPU. As you can see there isn't an intel GPU App in the CUDA75(App_Only) folder, so, you won't receive any iGPU tasks. The machine will work better without it, just read a few of the threads and you'll see the conclusion is to not use it if you have another GPU.

...For some reason the Laptop is still giving the detection warnings at times. That's something Jason will have to look into.

Interesting side issue, and probably out of our application scope:

v8 task detected
Cuda error 'Couldn't get cuda device count
' in file 'cuda/cudaAcceleration.cu' in line 138 : no CUDA-capable device is detected.
setiathome_CUDA: cudaGetDeviceCount() call failed.
setiathome_CUDA: No CUDA devices found
setiathome_CUDA: Found 0 CUDA device(s):
In cudaAcc_initializeDevice(): Boinc passed DevPref 1
setiathome_CUDA: CUDA Device 1 specified, checking...
Device cannot be used
Cuda device initialisation retry 1 of 6, waiting 5 secs...
...
...
setiathome_CUDA: Found 0 CUDA device(s):
In cudaAcc_initializeDevice(): Boinc passed DevPref 1
setiathome_CUDA: CUDA Device 1 specified, checking...
Device cannot be used
Cuda initialisation FAILED, Initiating Boinc temporary exit (180 secs)
...
...
[more retries and failures later...]

cudaGetDeviceCount() is the first Cuda runtime call, and occurs logically before any Cuda device properties are even known. Failure of that implies either there are simply no devices, the Cuda runtime libraries, or driver are broken in some way.

One possible exception might be if something similar to nVidia Optimus technology (in proprietary Apple form) is sending the nv GPU to sleep, the driver/runtime/OS just responding by removal of the device (which it shouldn't do, even an asleep device is still a present device).

After that check of device count, there is a trivial bug in baseline code initialisation that can cause an issue if you happened to have more than 8 Cuda devices detected ... obviously not that (doesn't happen, and that would be afterwards anyway).

I would suggest looking at power options... beyond that disappearing devices is squarely in the realm of OS + drivers, so could be something worth chasing up with the web driver devs, or with Apple.

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

I see what I did. I installed the current cuda driver. Waiting for the scheduler to send some nvidia work units.

It's too bad about the iGPU. I didn't see any specific errors from it but I guess I will be better off without it for now.

It looks like I am not getting and GPU tasks now. Did Seti blacklist my GPU tasks because of the last aborts?

It looks that way. The details page is showing;
SETI@home v8 (anonymous platform, NVIDIA GPU)
Number of tasks completed 0
Max tasks per day 3
Number of tasks today 30
Consecutive valid tasks 0

Not sure if it will send more before tomorrow or not.
You will probably have to report a completed GPU task before it will send anymore today.
That can be arranged, by running a cpu task on the GPU...but you would have to edit the client_state.xml file to accomplish that feat.