Welcome to MilkyWay@home

Posts by Floyd

1) Message boards : News : New Modfit Runs (Message 63090)
Posted 2 Feb 2015 by Floyd
Post:
After some more tinkering, I have discovered another parameter in the app_info that seems to have done the trick.

Specifically, I found I had to adjust <count>x</count> where x corresponds to the number or fraction of GPUs running a task.

I set this to 0.333 and the client happily crunched 3 concurrent tasks in about 30 seconds each instead of the 3 x 13 seconds for the equivalent individual tasks.

I then set it to 0.25 and the client crunches 4 concurrent tasks in about 39 seconds.

While I was there, I tried adding an entry for <flops> but while this changed the "Remaining: estimated" time, it was not a dynamic change, ie gave the same estimated time regardless of whether the client was running 1 or 4 concurrent WUs.

It would be useful if there were a fully documented list of app_info parameters and their likely effect - a definitive list rather than the piecemeal guesswork I have stumbled across in various forums. Oh well...
2) Message boards : News : New Modfit Runs (Message 63087)
Posted 2 Feb 2015 by Floyd
Post:
Thanks for the swift reply.

The reason for trying to fiddle is because I am seeing way less than 100% utilization in GPU-Z and was looking to improve efficiency.

The AMD card is currently crunching through modfit-fast WUs in about 13 seconds, but appears to get to 100% after about 10 seconds. I was guessing that this overrun was a small amount of CPU tidy-up, hence why the GPU usage drops to 0%. I was hoping that if I could double up the WU crunching, I could paper over this gap.

When the CPU is fully occupied with crunching another project, the MW WUs run to about 25 seconds - I haven't compared the WUs on the iGPU yet. I was wondering if a single CPU core would be enough to mother 2 WUs on the AMD card (MW@H) and one on the iGPU.

Of course things would be much easier to test if the sysadmins would sort out native support of the AMD card instead of me having to populate an app_info.
3) Message boards : News : New Modfit Runs (Message 63084)
Posted 1 Feb 2015 by Floyd
Post:
@swiftmallard

At the risk of hijacking the thread, I tried experimenting with your app_config entries without success.

I have an AMD R9 290X, so I have to use an app_info file just to get the applications running in the first place.

I was unable to get any more than the one WU running at a time regardless of my efforts.

I also have another project running on the Intel iGPU as well as a third project running on the CPU.

I am trying to optimise the overall output from the machine as a whole. What would you recommend to set for CPU usage for the GPU projects? Do they each really need a whole CPU core, or can the GPUs share a single core between them?
4) Message boards : Application Code Discussion : R9 290 is not recognized on linux x64 with proprietary drivers and OpenCL sdk installed (Message 63018)
Posted 16 Jan 2015 by Floyd
Post:
Posted in error
5) Message boards : Number crunching : Updated GPU Requirements (Currently not supporting GPU tasks) (Message 62885)
Posted 25 Dec 2014 by Floyd
Post:
Sorry, Jozef, but you are mistaken.

I have a Radeon R9 290X running tasks quite happily and fairly speedily as well.

I was guided by another forum member to create a custom app_info file for the purpose and put the relevant EXE files into the project folder manually.

See the second half of this thread for details:-

http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=3457

The BOINC manager reports the tasks as "Local:milkyway..." and cannot properly estimate the job length, but it crunches away regardless.

Of course it would be nice if the project technicians were to update the server code to recognise cards that have been around for years now, but that is another discussion.
6) Message boards : Number crunching : GPU tasks only uses 50% load on GTX780TI (Message 62867)
Posted 21 Dec 2014 by Floyd
Post:
Do you have a CPU core dedicated to each GPU as a supervisor?

If your CPU is maxed out doing other work, then it will starve the GPU processing.

As a quick test, you could temporarily suspend all CPU tasks and compare performance.

Are the cards overheating and self-throttling perhaps?
7) Message boards : Number crunching : ATI GPU R600 (R38xx) does not support OpenCL (Message 62781)
Posted 4 Dec 2014 by Floyd
Post:
Thanks for the download link, but I have already located a copy on a sister machine. My problems have moved on a little since then.

I have discovered that your app_info is coded for two of the four available applications. I have disabled the unconfigured applications in my profile until I can resolve the issue, either by the sysadmins fixing the generic problem or by using an app_info which covers all four applications.

Are you able/willing to supply a comprehensive app_info - I would have a go myself, but there are parameters in there that I have no idea what they are for.

I have also now discovered that the client does not seem willing to update to the server automatically on a frequent-enough schedule to keep the cache flowing. The sister machine seems to run a scheduler every 60 seconds, clearing completed tasks and requesting new ones, but this new installation doesn't seem to run any automatic scheduler.

Is that configured (or missing) in the app_info?


Contact rate is set by the cache settings that you use.

This should be the CPU app version of Milkyway.


<file_info>
<name>milkyway_separation_1.20_windows_x86_64.exe</name>
<executable />
</file_info>
<app_version>
<app_name>milkyway</app_name>
<version_num>120</version_num>
<file_ref>
<file_name>milkyway_separation_1.20_windows_x86_64.exe</file_name>
<main_program/>
</file_ref>
</app_version>

<file_info>
<name>milkyway_separation__modified_fit_1.36_windows_x86_64.exe</name>
<executable />
</file_info>
<app_version>
<app_name>milkyway_separation__modified_fit</app_name>
<version_num>136</version_num>
<file_ref>
<file_name>milkyway_separation__modified_fit_1.36_windows_x86_64.exe</file_name>
<main_program/>
</file_ref>
</app_version>

Paste them below the ATI versions that are already there.

<app_info>
<app>
<name>milkyway_separation__modified_fit</name>
</app>
<file_info>
<name>milkyway_separation__modified_fit_1.36_windows_x86_64__opencl_ati_101.exe</name>
<executable />
</file_info>
<app_version>
<app_name>milkyway_separation__modified_fit</app_name>
<version_num>136</version_num>
<avg_ncpus>0.040000</avg_ncpus>
<max_ncpus>0.567833</max_ncpus>
<plan_class>opencl_ati_101</plan_class>
<file_ref>
<file_name>milkyway_separation__modified_fit_1.36_windows_x86_64__opencl_ati_101.exe</file_name>
<main_program/>
</file_ref>
<coproc>
<type>ATI</type>
<count>1.0000</count>
</coproc>
</app_version>
<file_info>
<name>milkyway_separation__modified_fit_1.36_windows_x86_64.exe</name>
<executable />
</file_info>
<app_version>
<app_name>milkyway_separation__modified_fit</app_name>
<version_num>136</version_num>
<file_ref>
<file_name>milkyway_separation__modified_fit_1.36_windows_x86_64.exe</file_name>
<main_program/>
</file_ref>
<app>
<name>milkyway</name>
</app>
<file_info>
<name>milkyway_separation_1.20_windows_x86_64__opencl_amd_ati.exe</name>
<executable />
</file_info>
<app_version>
<app_name>milkyway</app_name>
<version_num>120</version_num>
<avg_ncpus>0.040000</avg_ncpus>
<max_ncpus>0.567833</max_ncpus>
<plan_class>opencl_ati</plan_class>
<file_ref>
<file_name>milkyway_separation_1.20_windows_x86_64__opencl_amd_ati.exe</file_name>
<main_program/>
</file_ref>
<coproc>
<type>ATI</type>
<count>1.0000</count>
</coproc>
</app_version>
<file_info>
<name>milkyway_separation_1.20_windows_x86_64.exe</name>
<executable />
</file_info>
<app_version>
<app_name>milkyway</app_name>
<version_num>120</version_num>
<file_ref>
<file_name>milkyway_separation_1.20_windows_x86_64.exe</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>

If you want NBODY, I cannot help you on that one.



NBODY is one of the two unconfigured applications. The other is MilkyWay@Home (you have given me the two "Separation" applications).

I am dedicating the CPU processing to a different project (WCG) so do not need the CPU aspect of MW@H.

I have been fiddling with settings trying to sort out conflicting global settings with project-specific ones and think I have finally got the cache setting right to keep up the MW@H feed.

I guess I can live with my new status quo until the general problem is sorted centrally.

Thanks very much for your help.
8) Message boards : Number crunching : ATI GPU R600 (R38xx) does not support OpenCL (Message 62778)
Posted 4 Dec 2014 by Floyd
Post:
Thanks for the download link, but I have already located a copy on a sister machine. My problems have moved on a little since then.

I have discovered that your app_info is coded for two of the four available applications. I have disabled the unconfigured applications in my profile until I can resolve the issue, either by the sysadmins fixing the generic problem or by using an app_info which covers all four applications.

Are you able/willing to supply a comprehensive app_info - I would have a go myself, but there are parameters in there that I have no idea what they are for.

I have also now discovered that the client does not seem willing to update to the server automatically on a frequent-enough schedule to keep the cache flowing. The sister machine seems to run a scheduler every 60 seconds, clearing completed tasks and requesting new ones, but this new installation doesn't seem to run any automatic scheduler.

Is that configured (or missing) in the app_info?
9) Message boards : Number crunching : ATI GPU R600 (R38xx) does not support OpenCL (Message 62775)
Posted 4 Dec 2014 by Floyd
Post:
Update:-

I found the following EXE files on another machine running MW@H:-

milkyway_separation__modified_fit_1.32_windows_x86_64__opencl_amd_ati
milkyway_separation__modified_fit_1.36_windows_x86_64__opencl_ati_101
milkyway_separation_1.02_windows_x86_64__ati14
milkyway_separation_1.02_windows_x86_64__opencl_amd_ati

and the client is now running tasks.

However, the task names are preceded by "Local".

Am I storing up problems for when the applications are updated/replaced?
10) Message boards : Number crunching : ATI GPU R600 (R38xx) does not support OpenCL (Message 62774)
Posted 4 Dec 2014 by Floyd
Post:
I am yet another poor soul trying to contribute effort using an R9 290X in a Windoze 7 rig.

Tried your app_info (thanks), but I'm getting an error about:-

"File referenced in app_info.xml does not exist: milkyway_separation_1.20_windows_x86_64__opencl_amd_ati.exe"


Do I need to download this EXE manually? If so, where from?

Before I inserted the app_info file, I had the following installed:-

Boinc 7.4.27 (x64)
Catalyst 14.301
AMD APP SDK 2.9-1

I am also bizarrely gettting a message:-

"app version refers to missing GPU type NVIDIA"

I am hoping there is a simple answer (maybe the sysadmins can fix the project itself?).

Any ideas?
11) Message boards : Number crunching : New BOINC Manager = no MWAH WUs (Message 54011)
Posted 13 Apr 2012 by Floyd
Post:
The answer is there on the BOINC site for all to see if you actually read the release notes:-

"GPU detection will no longer work when BOINC is installed as a service, or protected application execution"

So either BOINC 7 as a service OR gpu crunching, but not both.
12) Message boards : Number crunching : New BOINC Manager = no MWAH WUs (Message 53994)
Posted 12 Apr 2012 by Floyd
Post:
My machine was happily crunching away steady numbers of WUs until I over-installed BOINC 7.0.25.

The BOINC event log reported no WUs available, but otherwise no communication issues.

Once I uninstalled 7.0.25 and reinstalled 6.12.34, the client downloaded a full set of WUs and resumed action as normal.

Has anybody else seen this behaviour or is my problem my very own?




©2024 Astroinformatics Group