Hello,

I've got a HD 6990 overclocked, with a max output of 6 Teraflops but the projects I'm running are not using more than 4% - 25% of my GPU Core.

Milkyway@home is the only one which reaches 97% on both cores and can output over 120,000 credits a day (would be more but don't get enough work).

The following I'm having issues with:

Eistein@home - Only using 25% at max
Poem@home - Only using 20% at max

That's leaving nearly 80% free on both GPU cores, and going to waste.

My CPU has 8 cores and there at 99% usage, the computer stills runs smoothly =)

Any idea how to increase the output for these projects?

I should note, the cc_config & app_info.xml doesn't seem to exist nor when created make any changes to my settings. (Windows 8 Retail Preview, with 12.6 drivers.) - Latest version of BOINC, older versions won't install.

I should note, the cc_config & app_info.xml doesn't seem to exist nor when created make any changes to my settings.

Where did you create them? cc_config.xml belongs in the BOINC data directory and the different app_info.xml files in the respective project subdirectories.

BTW, Einstein@home doesn't need an app_info.xml; there you can use the preferences to run more than one task on each GPU.

Gruß,
Gundolf

---

Computer sind nicht alles im Leben. (Kleiner Scherz)

I created the cc_config.xml in: D:\ProgramData\BOINC\
and the app_info.xml in: D:\ProgramData\BOINC\projects\boinc.fzk.de_poem\

Where do I find the preferences to adjust for Einstein@home? I can't find them in the boinc manager.

Where do I find the preferences to adjust for Einstein@home? I can't find them in the boinc manager.

On Your Account page on the Einstein website, where it says Einstein@Home preferences.

Claggy

Which option do I change to increase the work units per GPU? I can't seem to find it on the options available.

Which option do I change to increase the work units per GPU? I can't seem to find it on the options available.

The one at the Bottom that says: GPU utilization factor of BRP apps

GPU utilization factor of BRP apps
DANGEROUS! Only touch this if you are absolutely sure of what you are doing!
Wrong setting might even damage your computer! Use solely on your own risk!
Min: 0.0 / Max: 1.0 / Default: 1.0

Set it at 0.5 for two Wu's at one, 0.33 for 3 Wu's at once.

Claggy

Thank you Claggy,

I've now got Eistein@home using 95% of my GPU Core.

How do I go about poem@home as I need to run about 4 WU's per GPU core to achieve 80%+ usage.

I tried following guides online regarding the app_info.xml but it's not working along with the cc_config.xml (neither existed originally and are not being read by BOINC or Processes related)

I created the cc_config.xml in: D:\ProgramData\BOINC\
and the app_info.xml in: D:\ProgramData\BOINC\projects\boinc.fzk.de_poem\

Any idea's, and thank you for very much for the help your providing me.

What are your contents of the cc_config.xml and app_info.xml?

Claggy

My app_info.xml is:

<app_info>
<app>
<name>poemcl</name>
<user_friendly_name>POEM++ OpenCL</user_friendly_name>
</app>
<file_info>
<name>poemcl_0.1_windows_intelx86__opencl_ati_100</name>
<executable/>
</file_info>

<app_version>
<app_name>poemcl</app_name>
<version_num>1</version_num>
<plan_class>opencl_ati_100</plan_class>
<avg_ncpus>0.5</avg_ncpus>
<max_ncpus>1</max_ncpus>
<flops>2.1e10</flops>
<coproc>
<type>ATI</type>
<count>0.25</count>
</coproc>
<cmdline></cmdline>
<file_ref>
<file_name>poemcl_0.1_windows_intelx86__opencl_ati_100</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>

and the cc_config.xml:

<cc_config>
<options>
<report_results_immediately>1</report_results_immediately>
</options>
<log_flags>
</log_flags>
</cc_config>

app_info.xml and cc_config.xml both look fine, Does Boinc on startup show a line with the following?:

Poem@Home | Found app_info.xml; using anonymous platform".

Claggy

Yes it does show the line, here's a copy of the log around that area:

17/06/2012 19:13:47 | | OpenCL: ATI GPU 0: Cayman (driver version CAL 1.4.1734 (VM), device version OpenCL 1.2 AMD-APP (937.2), 2048MB, 2016MB available)
17/06/2012 19:13:47 | | OpenCL: ATI GPU 1: Cayman (driver version CAL 1.4.1734 (VM), device version OpenCL 1.2 AMD-APP (937.2), 2048MB, 2016MB available)
17/06/2012 19:13:47 | Poem@Home | Found app_info.xml; using anonymous platform
17/06/2012 19:13:47 | Poem@Home | [error] No application found for task: windows_intelx86 1 opencl_ati_100; discarding
17/06/2012 19:13:47 | Poem@Home | [error] No application found for task: windows_intelx86 1 opencl_ati_100; discarding
17/06/2012 19:13:47 | Poem@Home | [error] No application found for task: windows_intelx86 1 opencl_ati_100; discarding
17/06/2012 19:13:47 | Poem@Home | [error] State file error: result poempp_gpucrystal_1339452306_2060820658_0 not found for task

You probably lost that work because there was no platform tag in your app_info, add the following line before the <plan_class> line:

<platform>windows_intelx86</platform>

and restart Boinc, if resend lost tasks is enabled at poem the server should resend those lost tasks,

Claggy

Thank you, I've added the line to the app_info.xml and my projects have appeared in the Transfer Tab (They took over 8 hours to process and would of been a shame to have lost them.)

Seems to be all sorted =)

but the projects I'm running are not using more than 4% - 25% of my GPU Core.

That's not bad I think.
There was a grid project in the past (few weeks ago I don't remember which it was) - reporting, that only about 1% of the peak performance of the GPU were effectively used in practise.

I have the same-ish problem. I have a GT530 card, and Process Explorer only reports that I am using 17%. I have know that the card can take 100% usage for up to 15 minutes. How do I get the card to be used at 100%? I understand the idea of the app_info files, but I don't know what to write in it for a NVidia card. Any help would be greatly appreciated. Thank you.

Project: Poem@Home
Dell XPS 8300
Intel I7-2600
NVidia GT530

---

http://boincstats.com/signature/-1/bam/17970/sig.png

I have a GT530 card, and Process Explorer only reports that I am using 17%

Process Explorer only shows CPU activity, not any load of a GPU. For that you'll have to use something like GPU-Z: http://www.techpowerup.com/gpuz/.

And then, the work being done by the GPU is done on all cores of the GPU. The 'load' you may see shows mostly memory use. Running more than one task on a GPU can be done, but will slow the work down as the GPU has to switch between running the jobs alternately. So it won't run them at the same time as you'd do with 2 cores of a CPU.

---

Installed GPU-Z and Open Hardware Monitor. Ran both with Poem@home and I understand what you mean. I have no idea what Process Explorer was saying for the GPU load, but I know to ignore it now. GPU is running an average of 64% with nothing else running, so I don't think there would be any advantage to trying to double process my GPU. Thank you for your assistance.