MW 0.82 (ati14) intelx86 - 3 WU - WORKING!

[Christopher Herr]

Hi, as the x86_64 application on CAL is currently not working and there might be others who want to run it anyway with the CAL application, e.g. because their card(s) do not support OpenCL, here is an app_info I adapted myself and tested it successfully on a HD7970 today:
Download the application and an included app_info here (click) (http://www.arkayn.us/forum/index.php?action=tpmod;dl=item42)

This former will reserve one thread/core for processing three tasks at once:


<app_info>
<app>
<name>milkyway</name>
<user_friendly_name>Milkyway@Home</user_friendly_name>
</app>
<file_info>
<name>milkyway_separation_0.82_windows_intelx86__ati14.e xe</name>
<executable/>
</file_info>
<app_version>
<app_name>milkyway</app_name>
<version_num>82</version_num>
<platform>windows_intelx86</platform>
<flops>5.0e11</flops>
<avg_ncpus>0.333333</avg_ncpus>
<max_ncpus>0.333333</max_ncpus>
<plan_class>ati14ati</plan_class>
<coproc>
<type>ATI</type>
<count>0.333333</count>
</coproc>
<cmdline>--process-priority 4</cmdline>
<file_ref>
<file_name>milkyway_separation_0.82_windows_intelx86__ati14.e xe</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>

This latter will reserve only as many cores/threads as the included app_info will also do, so you may have more cores free for processing other things, like POEM or SIMAP, for instance :D :


<app_info>
<app>
<name>milkyway</name>
<user_friendly_name>Milkyway@Home</user_friendly_name>
</app>
<file_info>
<name>milkyway_separation_0.82_windows_intelx86__ati14.e xe</name>
<executable/>
</file_info>
<app_version>
<app_name>milkyway</app_name>
<version_num>82</version_num>
<platform>windows_intelx86</platform>
<flops>5.0e11</flops>
<avg_ncpus>0.050000</avg_ncpus>
<max_ncpus>1.000000</max_ncpus>
<plan_class>ati14ati</plan_class>
<coproc>
<type>ATI</type>
<count>0.333333</count>
</coproc>
<cmdline>--process-priority 4</cmdline>
<file_ref>
<file_name>milkyway_separation_0.82_windows_intelx86__ati14.e xe</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>

As the OpenCL application is running slightly faster (about four seconds running two at once) for me on my HD7970, I use the OpenCL application myself, because those four seconds sum up to quite a large amount of wasted time and credits per day. :D

Speaking of credits: Did this sucker DA force the guys at MW to reduce credits? I remember having gained a lot more cpd with older and slower cards; okay they had dual GPUs, but anyway!?

Anyway, i thought some might need and like this, and as SETI.USA Hydra wants to get kicking again at Milkyway some time (soon?)... ;)

Hope this helps, Mike and all others trying to run the CAL application at the moment.

[Nuadormrac]

Yes, MW was substantially nerfed due to this credit new BS. The worst I've seen is on my GT650 paired with an i7 Ivy core, getting only 154 credits for 35 min crunching on a task, the CPU isn't so bad, around 2 hours for 159 credits (aprox 79 or so credits per hour), though I remember getting 200 credits per task on a much slower CPU (core 2 duo 1.7 GHz) which with an opti-app completed it in 1.5 hours for 200 credits...

I can only imagine these frackin credit parity people hit MW and hounded them till they nerfed their credits. On the GPU side, on nVidia cards, they're like the absolute worst credit granters out there, hands down, now....