Hi!

I only want to run the CUDA60 WUs on my Nvidia-Card.
So i need a working app_info.xml
Actually there are 8.01 WUs on my PC (cuda60) - and these i want in the XML-File

The files i found via google or in the lunatics package are very complex, several 100 lines of code and in most cases for windows.

I tried a little bit, getting only aborted WUs.
Maybe someone has a app_info.xml ready for Linux/CUDA?
Then i would have a basis to start.

Best Regards
MagicEye

---

First you need a working Linux/CUDA application, as far as I am aware there is no working CUDA 6.0 application for any platform, so you would have to compile your own from the source (and the source is thousands of lines, not hundreds), you would then have to debug it to get it to run properly, and it sounds as if you are a novice when it comes to CUDA programing.
However as far as I am aware the default (aka "stock") application CUDA application for Linux is CUDA 5, which should run very well with the need for an app_info.xml file.

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

As MagicEye says, there is a v8.01 cuda60 Linux/64 application shown on the stock applications page:

Linux/x86_64 8.01 (cuda60) 18 May 2016, 1:10:51 UTC 799 GigaFLOPS

If he has the application already, we can only speculate as to why he might want to convert from stock to anonymous platform.

He has aborted nearly all opencl_nvidia_sah and opencl_nvidia_SoG tasks, despite the survivor being almost an order of magnitude faster than cuda60 (example from computer 8053404).

Still, if MagicEye really wants to convert, he has all the information he needs.

The files are already present on his machine - make a backup copy in a safe place.
The instructions for writing the app_info.xml file are documented in The format of app_info.xml
The detailed data needed to complete the template will be present in the <app_version> segment for cuda60 and associated <file> blocks in the client_state.xml file active on his machine now.

It's just a matter of piecing the jigsaw together.

I´ve had run severval cuda60 workunits before on Ubuntu 15.10 without issues. But didn´t use an app_config/info.xml file...

A small installation help

http://setiathome.berkeley.edu/forum_thread.php?id=79358

---

#MagicEye
-- I have used the app_info.xml (below) to run a Linux x41zi cuda60 version. It probably is not exactly what you would need but may perhaps guide you to a working .xml file. In particular, the libcudart and libcufft might very well be different for your cuda60 application; or if your .xml file doesn't have the lib references then that would be the first thing to fix. And, of course, your exact application name may differ.

The "code" block shown is NOT the whole app_info.xml but just the portion relevant to the cuda60 application. If you have other sections for CPU applications or even for v7 applications they would remain unchanged.

../snip/..
        <app>
         <name>setiathome_v8</name>
        </app>
        <file_info>
           <name>setiathome_x41zi_x86_64-pc-linux-gnu_cuda60</name>
           <executable/>
        </file_info>

        <file_info>
           <name>libcudart.so.6.0</name>
           <executable/>
        </file_info>
        <file_info>
           <name>libcufft.so.6.0</name>
           <executable/>
        </file_info>
       <app_version>
           <app_name>setiathome_v8</app_name>
           <version_num>800</version_num>
           <platform>x86_64-pc-linux-gnu</platform>
           <plan_class>cuda60</plan_class>
           <avg_ncpus>0.5</avg_ncpus>
           <max_ncpus>1.0</max_ncpus>
           <coproc>
                <type>CUDA</type>
                <count>1.0</count>
           </coproc>
           <file_ref>
             <file_name>setiathome_x41zi_x86_64-pc-linux-gnu_cuda60</file_name>
             <main_program/>
           </file_ref>
           <file_ref>
              <file_name>libcudart.so.6.0</file_name>
           </file_ref>
           <file_ref>
              <file_name>libcufft.so.6.0</file_name>
           </file_ref>
        </app_version>

I no longer use the cuda60 application, having switched to the opencl_nvidia_SoG application for better production. My own benchmarks showed SoG to be much better on Green Bank data and only slightly slower on Arecibo data so I went with the better overall performance for my system.
(To use the x41zi application required going "anonymous platform" for me with the extra complications of that choice but that would not apply to your use of the stock cuda60 application.)

---

Thank you Gene, that looks like a good basis to adapt it to the apps on my PC.
(I think this is the better way than to build up the XML from Scratch with lots of errors...)

@others:
The reason for this wish is, that only the CUDA60 WUs run without influence of the graphics output. With OpenCL it might get more credit/h, but i cannot use the PC, because of nearly 1 second delay for everything on the desktop. Scrolling is a nightmare when OpenCL is active.

And since the original idea of seti@home is to use the idle computing power, i want to run Seti when i use the PC and to shut off the PC, when i dont use it. (except for things like the wow event, than it can run 24/7)

---