Right after I posted the above question, this thread got sidetracked with a question about HD4650 video cards and drifted away from the original "Updated OpenCL AstroPulse" apps subject and it never received a response. Now that a new AP tape is being split (09oc11ab) it seems like this would be a good time to bring it up again for all those that use Linux.

Updated Linux apps sent to Eric already. Up to him when they appear on beta. After their beta appearance few days to few weeks until they appear on main.

I myself wrote:

(...)
Here the app_info entries:

    <app>
        <name>astropulse_v7</name>
    </app>
    <file_info>
        <name>astropulse_7.09_windows_intelx86__opencl_intel_gpu_102.exe</name>
        <executable/>
    </file_info>
    <file_info>
        <name>libfftw3f-3-3-4_x86.dll</name>
        <executable/>
    </file_info>
    <file_info>
        <name>AstroPulse_Kernels_r2742.cl</name>
    </file_info>
    <file_info>
        <name>ap_cmdline_win_x86_SSE2_OpenCL_Intel.txt</name>
    </file_info>
    <app_version>
        <app_name>astropulse_v7</app_name>
        <version_num>704</version_num>
        <platform>windows_x86_64</platform>
        <avg_ncpus>0.04</avg_ncpus>
        <max_ncpus>0.2</max_ncpus>
        <plan_class>opencl_intel_gpu_102</plan_class>
       	<cmdline></cmdline>
        <coproc>
            <type>intel_gpu</type>
            <count>1</count>
        </coproc>
        <file_ref>
            <file_name>astropulse_7.09_windows_intelx86__opencl_intel_gpu_102.exe</file_name>
            <main_program/>
        </file_ref>
        <file_ref>
            <file_name>libfftw3f-3-3-4_x86.dll</file_name>
        </file_ref>
        <file_ref>
            <file_name>AstroPulse_Kernels_r2742.cl</file_name>
        </file_ref>
        <file_ref>
            <file_name>ap_cmdline_win_x86_SSE2_OpenCL_Intel.txt</file_name>
            <open_name>ap_cmdline.txt</open_name>
        </file_ref>
    </app_version>
    <app_version>
        <app_name>astropulse_v7</app_name>
        <version_num>709</version_num>
        <platform>windows_x86_64</platform>
        <avg_ncpus>0.04</avg_ncpus>
        <max_ncpus>0.2</max_ncpus>
        <plan_class>opencl_intel_gpu_102</plan_class>
       	<cmdline></cmdline>
        <coproc>
            <type>intel_gpu</type>
            <count>1</count>
        </coproc>
        <file_ref>
            <file_name>astropulse_7.09_windows_intelx86__opencl_intel_gpu_102.exe</file_name>
            <main_program/>
        </file_ref>
        <file_ref>
            <file_name>libfftw3f-3-3-4_x86.dll</file_name>
        </file_ref>
        <file_ref>
            <file_name>AstroPulse_Kernels_r2742.cl</file_name>
        </file_ref>
        <file_ref>
            <file_name>ap_cmdline_win_x86_SSE2_OpenCL_Intel.txt</file_name>
            <open_name>ap_cmdline.txt</open_name>
        </file_ref>
    </app_version>

<version_num>704</version_num> for the already downloaded tasks in BOINC & <version_num>709</version_num> for the next new tasks which BOINC will get.
In future if you have just only 709's in BOINC you can delete the whole:
<app_version>
...
<version_num>704</version_num>
...
</app_version> - entry.
(...)

Until now I thought it work like this, like I described in my above message.

I use now the above mentioned app_info.xml file entries.
BOINC downloaded new AP tasks for the Intel iGPU, but in BoincTasks Manager (v1.67) I see just v7.04 AP tasks.
I thought, BOINC (currently Client v7.4.42) use always the highest <version_num> entry in app_info.xml file (if more than just one is written in it).
So I should see now v7.04 and v7.09 AP tasks for Intel iGPU, or not?

I made something wrong?
Or what's up?

Thanks.

---

If you specify two app_versions which are identical other than the number, BOINC uses the first one.

                                                                   Joe

I'm a little bit confused at the moment. Seems like there are 4 different AP-apps for Nvidia cards:

Windows/x86 	7.10 (cuda_opencl_100) 	23 Apr 2015, 18:50:41 UTC 	81 GigaFLOPS
Windows/x86 	7.10 (cuda_opencl_cc1) 	23 Apr 2015, 18:50:41 UTC 	72 GigaFLOPS
Windows/x86 	7.10 (opencl_nvidia_100) 	23 Apr 2015, 18:50:41 UTC 	542 GigaFLOPS
Windows/x86 	7.10 (opencl_nvidia_cc1) 	23 Apr 2015, 18:50:41 UTC 	58 GigaFLOPS

What's the difference between those 4 and which one shall I use for my GTX750? (I'm about to set up my desktop as dedicated cruncher atm).

---

I'm a little bit confused at the moment. Seems like there are 4 different AP-apps for Nvidia cards:

Windows/x86 	7.10 (cuda_opencl_100) 	23 Apr 2015, 18:50:41 UTC 	81 GigaFLOPS
Windows/x86 	7.10 (cuda_opencl_cc1) 	23 Apr 2015, 18:50:41 UTC 	72 GigaFLOPS
Windows/x86 	7.10 (opencl_nvidia_100) 	23 Apr 2015, 18:50:41 UTC 	542 GigaFLOPS
Windows/x86 	7.10 (opencl_nvidia_cc1) 	23 Apr 2015, 18:50:41 UTC 	58 GigaFLOPS

What's the difference between those 4 and which one shall I use for my GTX750? (I'm about to set up my desktop as dedicated cruncher atm).

No, there is only one NV AP app there, the 7.10 app,

there are four different planclasses, two for old Boinc's that don't do OpenCL detection, that use the assumption that if a device has Cuda support,
it also has OpenCL support, two for modern Boinc that can detect OpenCL support,
The cc1 ones are for Legacy NV GPUs, there is a OpenCL driver Bug with 340.xx drivers for legacy NV GPUs, so they are restricted to 337.xx and earlier drivers.

Claggy

Thx Claggy!

---

What's the difference between those 4 and which one shall I use for my GTX750? (I'm about to set up my desktop as dedicated cruncher atm).

Rather than using any of those, use the dropbox link Raistmer posted earlier, then merge the app_info's:

Message 1670053

If you have existing work, you'll need to add app_version sections for the app_version of the work you have.

Claggy

Wingman's PC make errors with stock apps on AMD GPU: http://setiathome.berkeley.edu/show_host_detail.php?hostid=7560694.

If someone have an idea what's wrong, please send a PM to him. :-)

Thanks.

---

Wingman's PC make errors with stock apps on AMD GPU: http://setiathome.berkeley.edu/show_host_detail.php?hostid=7560694.

If someone have an idea what's wrong, please send a PM to him. :-)

If you actually looked at the stderr.txt of the errored tasks you'd see:

Error tasks for computer 7560694

<core_client_version>7.4.42</core_client_version>
<![CDATA[
<message>
WU download error: couldn't get input files:
<file_xfer_error>
<file_name>ap_15ja12ab_B1_P1_00185_20150430_03559.wu</file_name>
<error_code>-224 (permanent HTTP error)</error_code>
<error_message>permanent HTTP error</error_message>
</file_xfer_error>

</message>

<core_client_version>7.4.42</core_client_version>
<![CDATA[
<message>
WU download error: couldn't get input files:
<file_xfer_error>
<file_name>ap_15ja12ab_B1_P0_00186_20150430_03521.wu</file_name>
<error_code>-224 (permanent HTTP error)</error_code>
<error_message>permanent HTTP error</error_message>
</file_xfer_error>

</message>

<core_client_version>7.4.42</core_client_version>
<![CDATA[
<message>
app_version download error: couldn't get input files:
<file_xfer_error>
<file_name>libfftw3f-3.dll</file_name>
<error_code>-224 (permanent HTTP error)</error_code>
<error_message>permanent HTTP error</error_message>
</file_xfer_error>

</message>

Claggy

I saw.

And your proposed solution?

---

I saw.

And your proposed solution?

No solution needed. BOINC is fault-tolerant, and will make resends as needed. At least one of the workunits has already validated.

I saw.

And your proposed solution?

Suggest they get a more reliable internet connection?

I certainly Not going to PM that to them.

Claggy

I saw.

And your proposed solution?

No solution needed. BOINC is fault-tolerant, and will make resends as needed. At least one of the workunits has already validated.

Good catch.

1 task with a ratio of survival 3/2.

3 error results:
http://setiathome.berkeley.edu/show_host_detail.php?hostid=7560694 (PC with a weak web connection?)
http://setiathome.berkeley.edu/show_host_detail.php?hostid=2360076 (a Win98SE PC which make just errors)
http://setiathome.berkeley.edu/show_host_detail.php?hostid=6621400 (PC with a NV Quadro NVS 295 card which trash all AP tasks which he can catch)

2 successfully results.

To sum, 'a regular SETI day'.

I saw.

And your proposed solution?

Suggest they get a more reliable internet connection?

I certainly Not going to PM that to them.

Claggy

Maybe the DSL router need a cold reboot?

---

What's the difference between those 4 and which one shall I use for my GTX750? (I'm about to set up my desktop as dedicated cruncher atm).

Rather than using any of those, use the dropbox link Raistmer posted earlier, then merge the app_info's:

Message 1670053

If you have existing work, you'll need to add app_version sections for the app_version of the work you have.

Claggy

I dl'd the app from Raistmer, moved all files to the Seti folder and edited the app info (changed all numbers for the app itself and the kernels from the old numbers to the new ones). But there seems to be a problem because today I got an AP and it immediatly errored out:
http://setiathome.berkeley.edu/result.php?resultid=4129202176

---

Most probably some so called virus protection crapware. Exclude the BOINC directory and the BOINC data directory from scan and all will be fine. Although, depending on how bad programmed the AV stuff installed is, you may have to reinstall the new AP application.

---

Aloha, Uli

Thx Ulrich! I never had this problem before with Avira or Avast but I recently switched to Panda so maybe things are different there. I will check at evening when I'm back home.

---

Yep, Panda was the problem. For some reason it classified the .exe as a virus and locked it. Hope it works now.

---

Hehe, what will be when in the future we will have automatic biological antiviruses... In some sense life itself is a virus and definitely will be exterminated by such over-carefull antivirus...

BTW, now we have new Linux OpenCL AP deployed on Beta. So try to test it.
And whose who know what they do could take binaries from where.