The XML files are "operating system neutral" - so give it a try.
I was able to run optimised applications on my first Linux machine just by changing the names of the application from those used on the Windows seed machine.
Currently you are working from something that is blatantly wrong, with no idea where to look.
And I stand by the statement - Chrome is NOT the best test for syntax tool for checking the BOINC XML files - they do not properly obey the standard XML file syntax, thus while no "standard" syntax rules are being bent or broken that does not mean that the BOINC specific rules are intact. Indeed the lack of a proper XML file header will probably PREVENT Chrome from identifying the syntax requirements as being "XML", thus it will parse the file as a simple, plain text file, which of course will have no errors.

---

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

The XML files are "operating system neutral" - so give it a try.
I was able to run optimised applications on my first Linux machine just by changing the names of the application from those used on the Windows seed machine.
Currently you are working from something that is blatantly wrong, with no idea where to look..

Bob,
Thank you for jogging my thinking. While I can certainly give it a try, probably due to another screwup on my part, I am not getting any GPU tasks from Boinc/Seti this morning even though Boinc/SetiBeta sent me a batch after I allowed new tasks. My current guess is nothing will show up until Thursday when a bunch "ghost cache" jobs will expire.

So will get setup and see but probably won't have anything to post till later on Thursday.

Thank you Bob, Mike and anyone else for your patience.

Tom

---

A proud member of the OFA (Old Farts Association).

I am sorry. I tried Mike's example after editing the file names to the ones I had. And Boinc threw a "syntax error" fit that I was never able to clear. As a result I wasn't getting any Seti tasks much less Gpu tasks.

I tried hitting the reset after I made a couple of copies of setiathome directory. Nothing doing. So I set Seti to "get no tasks" and deleted the entire directory. Allowed new tasks, hit manual update once it had, basically nothing happened.

However, an hour later when I took a look I had Stock Seti in place and was full of both cpu and gpu work.

I'm going to postpone the Lunatics experiment until I have a chance to take a breath and try to re-create the asymmetrical gpu experiment in stock seti now that I am currently doing that.

I also have a new-to-me box coming and want to get all the transplanting done before I upgrade to Lunatics again. I get too much cpu production boost to not go back.

Tom

---

A proud member of the OFA (Old Farts Association).

I got impatient so I re-installed my copy of the beta6 distribution and then applied a global replace of r3557 to r3584 using notepad. I had downloaded the r3584 app previously so I just copied that subdirectory into the Boinc/Seti directory. Then I started up Boinc.

All the previously present GPU tasks for Seti had disappeared. I still have lots of cpu tasks. But I am now worried that I have managed to create yet 100 more gpu ghost tasks.

:(

I guess that is why its called "Beta6" :)

The app_info.xml file is once again, gigantic.

I wonder if I needed to run Boinc/Seti for a while before I applied the r3557 to r3584 update.

Tom

---

A proud member of the OFA (Old Farts Association).

All the previously present GPU tasks for Seti had disappeared. I still have lots of cpu tasks. But I am now worried that I have managed to create yet 100 more gpu ghost tasks.

:(

Tom

I went away and Boinc and the central scheduler got busy and loaded me up with maybe a hundred GPU tasks. So I am back where I started before I started trying to tinker.

Before I started trying to "improve" things I did discover that if I put an acceptable set of parameters in the mb*sog.txt for the Gtx 750 Ti , the Gtx 1060 didn't appear to slow down at all. So that may very well become my default if I want to run two asymmetrical gpus under the Lunatics distro under Windows.

I still have some hardware changes to make before I put the Gtx 750 ti back in this box.

Tom

---

A proud member of the OFA (Old Farts Association).

It's not that hard to trim the app_info that the Lunatics installer generates down to bare minimum of the actual apps you use. If you are running the SoG app, there is no need for all the CUDA sections. If you are running a 64 bit machine, there is no need for the X86 sections. There is no need for app versions other than the latest SoG version. I have one section for x64 bit MB SoG and one section for x64 bit AP. You only need one section for the CPU app and one section for the GPU app for each project. I trimmed the Lunatics generated app_info from 40 kB down to 5 kB.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

I guess that is why its called "Beta6" :)

The app_info.xml file is once again, gigantic.

If it weren't, there wouldn't be any point in having the installer.

The idea behind the installer was it would allow anyone to install the optimized application, and allow them to retain all their existing cache, as well as in progress WUs. Hence the large app_info.xml file.

---

Grant
Darwin NT

It's not that hard to trim the app_info that the Lunatics installer generates down to bare minimum of the actual apps you use. If you are running the SoG app, there is no need for all the CUDA sections. If you are running a 64 bit machine, there is no need for the X86 sections. There is no need for app versions other than the latest SoG version. I have one section for x64 bit MB SoG and one section for x64 bit AP. You only need one section for the CPU app and one section for the GPU app for each project. I trimmed the Lunatics generated app_info from 40 kB down to 5 kB.

Keith,
I haven't tried it yet, but I do have a stripped down copy of the app_info.xml file. Its a LOT easier to read. Not claiming I understand it (yet). I have potential hardware changes in the air right now (I ordered the wrong box/motherboard version). So I am letting the Lunatic auto-pilot.

Thank you.
Tom

---

A proud member of the OFA (Old Farts Association).

I mainly trimmed my app_info down so that I could experiment with the different reschedulers. But it IS A LOT easier to read now.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

ok, I have "trimmed" my app_info.xml file. And tried to add the "multibeam_nv_config.xml file.

Based on GPU-Z's memory reports on both gpu's I appear to not be successfully reading and using the "multibeam_nv_config.xml" file.

Here is the app_info.xml file.

<app_info> 
    <app>
        <name>astropulse_v7</name>
    </app>
    
     <app>
        <name>astropulse_v7</name>
    </app>
    <file_info>
        <name>AP7_win_x86_SSE2_OpenCL_NV_r2887.exe</name>
        <executable/>
    </file_info>
    <file_info>
        <name>libfftw3f-3-3-4_x86.dll</name>
        <executable/>
    </file_info>
    <file_info>
        <name>AstroPulse_Kernels_r2887.cl</name>
    </file_info>
    <file_info>
        <name>ap_cmdline_win_x86_SSE2_OpenCL_NV.txt</name>
    </file_info>
    <app_version>
        <app_name>astropulse_v7</app_name>
        <version_num>710</version_num>
        <platform>windows_x86_64</platform>
        <avg_ncpus>0.04</avg_ncpus>
        <max_ncpus>0.2</max_ncpus>
        <plan_class>opencl_nvidia_100</plan_class>
       	<cmdline></cmdline>
        <coproc>
            <type>CUDA</type>
            <count>1</count>
        </coproc>
        <file_ref>
            <file_name>AP7_win_x86_SSE2_OpenCL_NV_r2887.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_r2887.cl</file_name>
        </file_ref>
        <file_ref>
            <file_name>ap_cmdline_win_x86_SSE2_OpenCL_NV.txt</file_name>
            <open_name>ap_cmdline.txt</open_name>
        </file_ref>
    </app_version>
    
    
      <app_version>
        <app_name>setiathome_v8</app_name>
        <version_num>800</version_num>
        <platform>windows_intelx86</platform>
        <api_version>7.5.0</api_version>
        <file_ref>
            <file_name>MB8_win_x64_SSE3_VS2008_r3330.exe</file_name>
            <main_program/>
        </file_ref>
        <file_ref>
            <file_name>libfftw3f-3-3-4_x64.dll</file_name>
        </file_ref>
        <file_ref>
            <file_name>mb_cmdline_win_x64_SSE3_VS2008.txt</file_name>
            <open_name>mb_cmdline.txt</open_name>
        </file_ref>
    </app_version>

  <app_version>
        <app_name>setiathome_v8</app_name>
        <version_num>805</version_num>
        <platform>windows_intelx86</platform>
        <api_version>7.5.0</api_version>
        <file_ref>
            <file_name>MB8_win_x64_SSE3_VS2008_r3330.exe</file_name>
            <main_program/>
        </file_ref>
        <file_ref>
            <file_name>libfftw3f-3-3-4_x64.dll</file_name>
        </file_ref>
        <file_ref>
            <file_name>mb_cmdline_win_x64_SSE3_VS2008.txt</file_name>
            <open_name>mb_cmdline.txt</open_name>
        </file_ref>
    </app_version>


    <app>
        <name>Setiathome_v8</name>
    </app>
    <file_info>
                   <file_name>MB8_win_x86_SSE3_OpenCL_NV_SoG_r3584.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>MultiBeam_Kernels_r3584.cl</file_name>
        </file_ref>
        <file_ref>
            <file_name>mb_cmdline_win_x86_SSE3_OpenCL_NV_SoG.txt</file_name>
<open_name>mb_cmdline.txt</open_name>
        
<file_name>ultibeam_NV_config.xml</file_name>
</file_ref>

    </app_version>
    <app_version>
        <app_name>setiathome_v8</app_name>
        <version_num>820</version_num>
        <platform>windows_intelx86</platform>
        <api_version>7.5.0</api_version>
        <avg_ncpus>0.04</avg_ncpus>
        <max_ncpus>0.2</max_ncpus>
        <plan_class>opencl_nvidia_SoG</plan_class>
        	<cmdline></cmdline>	
	<coproc>
            <type>CUDA</type>
            <count>1</count>
        </coproc>
        <file_ref>
            <file_name>MB8_win_x86_SSE3_OpenCL_NV_SoG_r3584.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>MultiBeam_Kernels_r3584.cl</file_name>
        </file_ref>

        <file_ref>            <file_name>mb_cmdline_win_x86_SSE3_OpenCL_NV_SoG.txt</file_name>

  <file_name>mb_cmdline.txt</file_name>
 

 <file_name>Multibeam_NV_config.xml</file_name>
 </file_ref>

 </app_version>
</app_info> 

Here is the Multibeam_NV_config.xml file

<device0>
	<period_iterations_num>4</period_iterations_num>
	<spike_fft_thresh>4096</spike_fft_thresh>
	<sbs>1024</sbs>
        <oclfft_plan>
       		<global_radix>256</global_radix>
		<local_radix>16</local_radix>
		<workgroup_size>256</workgroup_size>
		<max_local_size>512</max_local_size>
		<localmem_banks>64</localmem_banks>
		<localmem_coalesce_width>64</localmem_coalesce_width>
	</oclfft_plan>
</device0>
<device1>
	<period_iterations_num>4</period_iterations_num>
	<spike_fft_thresh>2048</spike_fft_thresh>
	<sbs>512</sbs>
        <oclfft_plan>
      		<global_radix>256</global_radix>
		<local_radix>16</local_radix>
		<workgroup_size>256</workgroup_size>
		<max_local_size>512</max_local_size>
		<localmem_banks>64</localmem_banks>
		<localmem_coalesce_width>64</localmem_coalesce_width>
	</oclfft_plan>
</device1>

Here is the mb*sog.txt file

-sbs 384 -tune 1 64 1 4 -period_iterations_num 4 -spike_fft_thresh 2048 -oclfft_tune_gr 256 -oclfft_tune_lr 16 -oclfft_tune_wg 256 -oclfft_tune_ls 512 -oclfft_tune_bn 64 -oclfft_tune_cw 64

I used the -sbs 384 in the mb*sog.txt file to help me recognize where the -sbs was being pulled from. My standard "compromise" version of this setup is -sbs 512 which allows the 1060 to run "pretty good."

Hope someone can help me figure out my errors.

Thank you.
Tom

---

A proud member of the OFA (Old Farts Association).

One thing that catches my eye

        <file_ref>
            <file_name>mb_cmdline_win_x86_SSE3_OpenCL_NV_SoG.txt</file_name>
<open_name>mb_cmdline.txt</open_name>
        
<file_name>ultibeam_NV_config.xml</file_name>
</file_ref>

You might want to add an "M" in your XML name.

EDIT: and your <file_ref> tags are not opened/closed properly.

Your file still looks far too long for what you are trying to achieve!

The section at the very top may confuse BOINC:

  <app>
        <name>astropulse_v7</name>
    </app>
    
     <app>
        <name>astropulse_v7</name>
    </app>

I'm not sure how well BOINC manages multiple empty entries for the same tag.
You also need to take great care over EXACT file names you use - capitalisation has to be consistent, in a file name "M" is not the same as "m", and very easy to get confused (Windows often presents file names with upper case first letters when in reality the file name is all lower case.)

---

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

Hi,
I want to THANK YOU for finding those typos. Duh <slaps forhead>.

I applied all the proposed changes including the <file_ref></file_ref> around each and every file. No luck.

I tried consolidating all three parameter files (mb*sog.txt, multi*config.xml and mb_cmdline.txt) into one <file_ref> the way the AP entry above had it. No luck.

I have tried changing the <open_name> to <file_name> (for mb_cmdine.txt). No luck.

In no case, has the memory of the Gpu's changed from the "baseline" at -sbs 384 or anything else I set it at.

I am unclear about what one responder meant when he said the "app_info.xml" file is still too large?

I thank you for your patience. If you want, I can post every combination I have tried. Or can someone post a windows version that appears to be working with appropriate comments embedded?

Or we can take another break Or give it up as a bad job.

Its too bad, my RAC has been rising even with out having both Gpu's tuned to their most productive :)

I suppose I can always revert the compromise mb*sog.txt or simply pull the GTX 750 Ti. Either one works, just not as well as I wanted it to.

Tom

---

A proud member of the OFA (Old Farts Association).

The simplest thing for you to do is STOP FIDDLING.
Delete app_info.xml and the other files you have broken.
Run the current version of the Lunatics installer to get a clean set of working files, and let the system RUN for a few days.
This will give you a baseline result to see what impact any changes you do make.

Once you've got a good baseline on the system after DAYS of running, not minutes, copy the working files, and keep them safe, before making ANY changes.

---

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

Run the current version of the Lunatics installer to get a clean set of working files, and let the system RUN for a few days.

Bob,
1) I have several WEEKS of baseline thank you.
2) I HAVE the original install file for this.
3) I WAS TOLD to edit that file down to make it easier to understand and work with. It certainly looks a lot less complicated to me.
4) I have enough experience playing with the -sbs parameter on my GPU's to immediately SEE if the parameter have taken effect when I inspect the results of a old task/new task with Gpu-Z. If the -sbs parameter has taken effect then the changes I have made are working. Since the effects of the -sbs parameter are not visible, I can reasonably assume nothing has been successful.

Your right. I am going to "stop fiddling" (with Lunatics distro). I have demonstrated to my satisfaction that the Stock SOG is surprisingly fast. And gpu processing does drown out cpu processing on this machine (I don't have enough cores to compensate).

Seti Screen saver strikes back (I miss my graphics) <queue Star Wars "the empire strikes back theme"> ROFLing....

Tom

---

A proud member of the OFA (Old Farts Association).

I am pleased to report that in stock Seti at home the documented "Multibeam_NV_config.xml" file does appear to be working.

The Gtx 1060 is showing "normal" memory usage under inspection by Gpu-z for the -sbs setting of 1024.

The Gtx 750 Ti is showing "normal" memory usage under inspection by Gpu-z for the -sbs setting of 512.

Both of these numbers are "optimum" based on other advice in other threads.

As one Nobel Prize Economist said this isn't optimum but it is "satisficed".

I might take another stab at doing this under the Lunatics distribution once I have completed my course in boinc xml files and the anonymous platform. Brought to you by DIY University ;)

After all, the Lunatics distribution produces a significantly faster result in both my cpu and my gpu(s). And I do lust after that speed up.

Tom

---

A proud member of the OFA (Old Farts Association).

We know it is working.
I tested it years ago.

---

With each crime and every kindness we birth our future.

We know it is working.
I tested it years ago.

That's great that you tested it years ago. There is a process in software testing called "regression testing". Hopefully it is fully automated. And it discovers if the latest "fixes" break something else.

I believe I have evidence that "it" is no longer working.

I was wrong in my report. I failed to remove the -sbs xxx entry from the mb*sog.txt file.

When I did so, the memory reported used for both GPU's fell to "default" levels for each gpu. (after clicking the "read config files" on the boinc menu and after going to new task).

When I put the -sbs 1024 back in the mb*sog.txt file both gpu memory usages returned to the previous levels. (after clicking the "read config files" on the boinc menu and after going to new task).

So the "Multibeam_NV_config.xml" file that I constructed based on the documentation isn't working at least for the -sbs parameter.

Sigh.

The only good news I can report is the GTX 750 Ti is not too unhappy working under the parameters that are best for the GTX 1060 card. The 750 is buzzing through the SOG files at about 15-18 minutes just like it used to under the "best" parameters I thought I had.

Tom

---

A proud member of the OFA (Old Farts Association).

I finally snagged a GTX 1060 6GB shorty card to add to my existing dual 1070s. I am going to have to try and get this Multibeam_NV_config.xml working in my Ryzen system. Or have my app_config detuned for all cards. I'll post what my results are when I receive the 1060 and get it installed.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

I am going to have to try and get this Multibeam_NV_config.xml working ...

It should not be too difficult to edit app_info.xml
You should have the file Multibeam_NV_config.xml (even empty) in SETI@home directory before trying to start BOINC with changed app_info.xml

Note: I don't think Caps matter in filenames on Windows but the exact name from inside the .exe (Copy/Paste) is:
MultiBeam_NV_config.xml

I think it is enough to change only the first (topmost) section, from MB8_win_x86_SSE3_OpenCL_NV.aistub this is:

    <app>
        <name>setiathome_v8</name>
    </app>
    <file_info>
        <name>MB8_win_x86_SSE3_OpenCL_NV_SoG_r3557.exe</name>
        <executable/>
    </file_info>
    <file_info>
        <name>libfftw3f-3-3-4_x86.dll</name>
	<executable/>
    </file_info>
    <file_info>
        <name>MultiBeam_Kernels_r3557.cl</name>
    </file_info>
    <file_info>
        <name>mb_cmdline_win_x86_SSE3_OpenCL_NV_SoG.txt</name>
    </file_info>
    <app_version>
        <app_name>setiathome_v8</app_name>
        <version_num>819</version_num>
        <platform>windows_intelx86</platform>
        <api_version>7.5.0</api_version>
        <avg_ncpus>0.04</avg_ncpus>
        <max_ncpus>0.2</max_ncpus>
        <plan_class>opencl_nvidia_SoG</plan_class>
        	<cmdline></cmdline>	
	<coproc>
            <type>CUDA</type>
            <count>1</count>
        </coproc>
        <file_ref>
            <file_name>MB8_win_x86_SSE3_OpenCL_NV_SoG_r3557.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>MultiBeam_Kernels_r3557.cl</file_name>
        </file_ref>
        <file_ref>
            <file_name>mb_cmdline_win_x86_SSE3_OpenCL_NV_SoG.txt</file_name>
            <open_name>mb_cmdline.txt</open_name>
        </file_ref>
    </app_version>

First Copy/Paste two existing parts (<file_info> and <file_ref>, I used where it say MultiBeam_Kernels_r3557.cl):

    <app>
        <name>setiathome_v8</name>
    </app>
    <file_info>
        <name>MB8_win_x86_SSE3_OpenCL_NV_SoG_r3557.exe</name>
        <executable/>
    </file_info>
    <file_info>
        <name>libfftw3f-3-3-4_x86.dll</name>
	<executable/>
    </file_info>
    <file_info>
        <name>MultiBeam_Kernels_r3557.cl</name>
    </file_info>

    <file_info>
        <name>MultiBeam_Kernels_r3557.cl</name>
    </file_info>

    <file_info>
        <name>mb_cmdline_win_x86_SSE3_OpenCL_NV_SoG.txt</name>
    </file_info>
    <app_version>
        <app_name>setiathome_v8</app_name>
        <version_num>819</version_num>
        <platform>windows_intelx86</platform>
        <api_version>7.5.0</api_version>
        <avg_ncpus>0.04</avg_ncpus>
        <max_ncpus>0.2</max_ncpus>
        <plan_class>opencl_nvidia_SoG</plan_class>
        	<cmdline></cmdline>	
	<coproc>
            <type>CUDA</type>
            <count>1</count>
        </coproc>
        <file_ref>
            <file_name>MB8_win_x86_SSE3_OpenCL_NV_SoG_r3557.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>MultiBeam_Kernels_r3557.cl</file_name>
        </file_ref>

        <file_ref>
            <file_name>MultiBeam_Kernels_r3557.cl</file_name>
        </file_ref>

        <file_ref>
            <file_name>mb_cmdline_win_x86_SSE3_OpenCL_NV_SoG.txt</file_name>
            <open_name>mb_cmdline.txt</open_name>
        </file_ref>
    </app_version>

Then change in the Copies (both <file_info> and <file_ref>) the MultiBeam_Kernels_r3557.cl with MultiBeam_NV_config.xml

    <app>
        <name>setiathome_v8</name>
    </app>
    <file_info>
        <name>MB8_win_x86_SSE3_OpenCL_NV_SoG_r3557.exe</name>
        <executable/>
    </file_info>
    <file_info>
        <name>libfftw3f-3-3-4_x86.dll</name>
	<executable/>
    </file_info>
    <file_info>
        <name>MultiBeam_Kernels_r3557.cl</name>
    </file_info>

    <file_info>
        <name>MultiBeam_NV_config.xml</name>
    </file_info>

    <file_info>
        <name>mb_cmdline_win_x86_SSE3_OpenCL_NV_SoG.txt</name>
    </file_info>
    <app_version>
        <app_name>setiathome_v8</app_name>
        <version_num>819</version_num>
        <platform>windows_intelx86</platform>
        <api_version>7.5.0</api_version>
        <avg_ncpus>0.04</avg_ncpus>
        <max_ncpus>0.2</max_ncpus>
        <plan_class>opencl_nvidia_SoG</plan_class>
        	<cmdline></cmdline>	
	<coproc>
            <type>CUDA</type>
            <count>1</count>
        </coproc>
        <file_ref>
            <file_name>MB8_win_x86_SSE3_OpenCL_NV_SoG_r3557.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>MultiBeam_Kernels_r3557.cl</file_name>
        </file_ref>

        <file_ref>
            <file_name>MultiBeam_NV_config.xml</file_name>
        </file_ref>

        <file_ref>
            <file_name>mb_cmdline_win_x86_SSE3_OpenCL_NV_SoG.txt</file_name>
            <open_name>mb_cmdline.txt</open_name>
        </file_ref>
    </app_version>

! DON'T type anything, only Copy/Paste !
 

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)