Working app_info files

[bearcatrp]

Looking at the app info file above, the version number shows 656, not 6.56 like in the rest (around the middle by versionnum). Would this cause an issue?

[pirogue]

Looking at the app info file above, the version number shows 656, not 6.56 like in the rest (around the middle by versionnum). Would this cause an issue?

It shouldn't. Mine's working and it has <version_num>656</version_num>.

[nanoprobe]

Excellent. Thanks for the link.

One last question (hopefully for a little while), what program did you use to down clock your gpu memory frequency?

Reason I ask is my memory setting doesn't seem to be sticking, with CCC or Afterburner (card is a 7970 on beta 12.11 driver). Under load, the memory shoots up to default frequency. GPU frequency appears to be obeying what I set.

I'm using Afterburner. Check to make sure you have the newest version. (2.2.4 I believe) My memory settings stay where I set them.

[jspace]

This should be the relevant info for CEP2

<app>
	<name>cep2</name>
	<user_friendly_name>The Clean Energy Project - Phase 2</user_friendly_name>
</app>
<file_info>
	<name>wcgrid_cep2_6.40_windows_intelx86</name>
	<executable/>
</file_info>
<file_info>
	<name>wcgrid_cep2_qchem_6.40_windows_intelx86</name>
	<executable/>
</file_info>
<file_info>
	<name>wcgrid_cep2_graphics_6.40_windows_intelx86</name>
	<executable/>
</file_info>
<file_info>
	<name>cep2_qcaux_6.40.zip</name>
</file_info>
<file_info>
    <name>cep2_images_6.40.zip</name>
</file_info>
<app_version>
    <app_name>cep2</app_name>
    <version_num>640</version_num>
    <platform>windows_intelx86</platform>
    <avg_ncpus>1.000000</avg_ncpus>
    <max_ncpus>1.000000</max_ncpus>
    <flops>1982909446.217418</flops>
    <api_version>6.11.7</api_version>
    <file_ref>
        <file_name>wcgrid_cep2_6.40_windows_intelx86</file_name>
        <main_program/>
    </file_ref>
    <file_ref>
        <file_name>wcgrid_cep2_qchem_6.40_windows_intelx86</file_name>
        <open_name>wcgrid_cep2_qchem_prod_linux.x86</open_name>
    </file_ref>
    <file_ref>
        <file_name>wcgrid_cep2_graphics_6.40_windows_intelx86</file_name>
        <open_name>graphics_app</open_name>
    </file_ref>
    <file_ref>
        <file_name>cep2_images_6.40.zip</file_name>
        <open_name>Cep2_images.zip</open_name>
    </file_ref>
    <file_ref>
        <file_name>cep2_qcaux_6.40.zip</file_name>
        <open_name>qcaux2.zip</open_name>
    </file_ref>
</app_version>

[sRHunt3r]

Thanks a lot guys, I put my two 5970's in my hex rigs with 8 w/u's each and they seem to be trucking along nicely. I will let them run a day or so and if they stay solid possibly add the 5870 and 5770.

[=[PULSAR]=]

Ok maybe you guys can help me with this. Trying to run a GTX295 but with or without the app file still only uses one of the gpu's. Any idea what to change in the app_info?

[Snow Crash]

I have been running this app_info for about 2 weeks without any problems on a Win7 x64 rig, BOINC version 7.0.28 with a GTX-660Ti and a GTX-295 ... 2 WU per GPU core (so yes, the 660Ti is running 2 and the 295 is running 4 for a total of 6 WUs in progress at a time). Because this rig is a 980X I am also running HCC CPU. I have my min buffer set to .69 and my max additional buffer set to .07

<app_info>
   <app>
      <name>hcc1</name>
      <user_friendly_name>XtremeSystems Crunches Cancer</user_friendly_name>
   </app>

   <file_info>
      <name>wcg_hcc1_img_6.56_windows_intelx86__nvidia_hcc1</name>
      <executable/> 
   </file_info>
   <file_info>
      <name>hcckernel.cl.6.56</name>
      <executable/>
   </file_info>

   <file_info>
      <name>wcg_hcc1_img_6.56_windows_intelx86</name>
      <executable/>
   </file_info>


   <app_version>
      <app_name>hcc1</app_name>
      <version_num>656</version_num>
      <platform>windows_intelx86</platform>
      <plan_class>nvidia_hcc1</plan_class>
      <avg_ncpus>1.0</avg_ncpus>
      <max_ncpus>1.0</max_ncpus>
      <coproc>
         <type>CUDA</type>
         <count>.5</count>
      </coproc>
      <file_ref>
          <file_name>wcg_hcc1_img_6.56_windows_intelx86__nvidia_hcc1</file_name>
          <main_program/>
      </file_ref>
      <file_ref>
         <file_name>hcckernel.cl.6.56</file_name>
         <open_name>hcckernel.cl</open_name>
      </file_ref>
   </app_version>

   <app_version>
      <app_name>hcc1</app_name>
      <version_num>656</version_num>
      <platform>windows_intelx86</platform>
      <avg_ncpus>1.000000</avg_ncpus>
      <max_ncpus>1.000000</max_ncpus>
      <flops>3231932486.733310</flops>
      <api_version>6.13.0</api_version>
      <file_ref>
         <file_name>wcg_hcc1_img_6.56_windows_intelx86</file_name>
         <main_program/>
      </file_ref>
   </app_version>
</app_info>

[=[PULSAR]=]

What version of boinc are you running x86 or x64

[pirogue]

Ok maybe you guys can help me with this. Trying to run a GTX295 but with or without the app file still only uses one of the gpu's. Any idea what to change in the app_info?

Try adding
<use_all_gpus>1</use_all_gpus>
to the <options></options> section of your cc_config.xml file. This one goes in the root of the BOINC Data folder.
If you don't have one, here is an example:

<?xml version="1.0"?>
<cc_config>
<options>
<use_all_gpus>1</use_all_gpus>
</options>
</cc_config>


From http://boinc.berkeley.edu/wiki/Client_configuration
<use_all_gpus>0|1</use_all_gpus>
If 1, use all GPUs (otherwise only the most capable ones are used)

[Snow Crash]

What version of boinc are you running x86 or x64

x64

[=[PULSAR]=]

Still just GPU 0 working for this, I tried disabling SLI too.

EDIT: Duh nevermind I got it, now how to you increase your work buffer?

[Snow Crash]

BOINC Manager ... Tools --> Computing Preferences ... Network Usage tab

[jspace]

Got FAAH working

<app>
    <name>faah</name>
    <user_friendly_name>FightAIDS@Home</user_friendly_name>
</app>
<file_info>
    <name>wcg_faah_autodock_6.40_windows_intelx86</name>
    <executable/>
</file_info>
<file_info>
    <name>wcg_faah_autodock_graphics_6.40_windows_intelx86</name>
    <executable/>
</file_info>
<file_info>
    <name>faah_protease_6.40.dat</name>
</file_info>
<file_info>
    <name>faah_image_6.40.tga</name>
</file_info>
<app_version>
    <app_name>faah</app_name>
    <version_num>640</version_num>
    <platform>windows_intelx86</platform>
    <avg_ncpus>1.000000</avg_ncpus>
    <max_ncpus>1.000000</max_ncpus>
    <flops>3894434573.022523</flops>
    <api_version>6.3.3</api_version>
    <file_ref>
        <file_name>wcg_faah_autodock_6.40_windows_intelx86</file_name>
        <main_program/>
    </file_ref>
    <file_ref>
        <file_name>wcg_faah_autodock_graphics_6.40_windows_intelx86</file_name>
        <open_name>graphics_app</open_name>
    </file_ref>
    <file_ref>
        <file_name>faah_protease_6.40.dat</file_name>
        <open_name>protease.dat</open_name>
    </file_ref>
    <file_ref>
        <file_name>faah_image_6.40.tga</file_name>
        <open_name>wc_grid_s1b2.tga</open_name>
    </file_ref>
</app_version>

Working on HFCC & DSFL next. We'll see if the machine that grabbed some starts complaining it can't get those "lost tasks".

[Sparky]

OK Mine seems to be quite happy right now, except I have a few HCC units that started on the CPU and are just sitting there (apparently once the CPU starts them the GPU won't take over). Do I have to abort them or...?

[nanoprobe]

OK Mine seems to be quite happy right now, except I have a few HCC units that started on the CPU and are just sitting there (apparently once the CPU starts them the GPU won't take over). Do I have to abort them or...?

GPU should take 1 CPU core to use by default.

[Sparky]

Right, but I mean I have a few that are 20%, 40%, etc done via CPU (before I put in app_info with 8 GPU threads) and now the GPU won't touch them and the CPU cores are either busy doing their part of the GPU work or just sitting idle waiting for the GPU units.

*edit* what I did was tell BOINC no new tasks, and suspended all other units in the queue. BOINC then processed the remaining HCC units that were CPU only. After that I resumed GPU work on the rest and resumed new tasks. Bingo, no more waiting units for the CPU anymore. I did notice that if a GPU unit is partway through and you suspend it that it will restart it from the beginning when told to resume.

[my87csx481]

I'm using Afterburner. Check to make sure you have the newest version. (2.2.4 I believe) My memory settings stay where I set them.

I *think* it stays....just a reporting issue maybe

Afterburner and Overdrive and gpu-z show memory at 1500 but, furmark and kombustor show memory at the downclocked value.

Whatever, it's crankin' away....for now.

(sorry for the late post and messin' up the flow)

[Sparky]

Got back from work today and found 7 of the 8 threads were hung since last night and had never even actually started the units. Dang it. I should have checked the PC this morning. 14 hours of 7x units wasted.

*edit* and I just had my first ever driver recovery. I'm running an old driver though, I should probably update that.

[Snow Crash]

A few of us have been getting good results and no hangs with the latest beta ... if you're going to update, I'd say go all the way!

[bluestang]

^^ What he said...12.11beta4

[Sparky]

K I'm downloading those now and will install them shortly. I had another unit hang, thankfully caught it before it wasted too much time.

[bluestang]

Make sure you uninstall old, run Driver Sweeper, then install new

[Sparky]

Well I can say I already see one improvement. There is next to no display lag with these drivers. I was using 12.4 previously and when crunching on the GPU even scrolling webpages was really choppy lol. Now it is a ton smoother, not perfect but still so much better.

[0ne.shot]

Can anyone confirm the ATI GPU only app_info from page one works? I put it in the directory and the BOINC client didn't like it. I'm running the GPU + CPU HCC with 8 CPUs on the GPU so the CPU tasks just keep piling up.

[Sparky]

Works just fine for me once I got it squared away. Here's what I did.

Set a profile for HCC only and have that profile set for your PC(s) that have the ATI GPU(s).
DO NOT set the app_info yet. Let the PC work through the units that you have until all that is left is HCC in the queue. If you don't do this you'll have miserable headaches (Been there, done that, didn't even get the shirt for it).
Once ALL units in queue on that rig are HCC, shut off BOINC and put the app_info in place. Fire BOINC up again.

Now if you're doing 8 GPU only, no CPU work, you'll probably notice you'll have a few partially done HCC units just sitting there while all the other units blast through the GPU. Once option is to just abort those few that were started via CPU, but I didn't want to do that. Instead, I told BOINC to not fetch more work, then selected all units that haven't been started yet and suspended them. Then I told BOINC to snooze the GPU only. BOINC then switched the CPU over to the partially finished CPU units. The suspend on everything else kept the CPU from starting anything new, so once the CPU units were done it went idle. Then I resumed the GPU, resumed all tasks, and resumed BOINC fetching work.