I've updated most of the server side stuff (and seems like I broke everything in the process). I think I've cleaned up most of the problems, but if you notice anything else weird post it here.

Hello,

I notice a too high estimated CPU-usage for 0.23 GPU seperation since the server upgrade.

Please show also here: http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=2312

I hope, you can change this...

---

I attached to MilkyWay. The server sent the correct "milkyway_0.23_windows_x86_64__ati13ati.exe" and "brook64a_ati.dll" for my computer but 2 minutes later also sent "milkyway_0.23_windows_intelx86__ati13ati.exe" and "brook32a_ati.dll". The correct 64-bit application is running, so I deleted the surplus 32-bit files.

I currently have 320 tasks in my cache. This is a great improvement from the previous limit of 48. My phoneline has degraded lately and ADSL is dropping out many times a day. It needs to be manually reconnected so a 4.5 hour cache is much more convenient for me than a 40 minute cache. Still will get idle periods overnight but much less than before.

Bit hazy on this one - could be wrong as I've had the auto log in checkmarked, and its ages since I looked for Log in/log out for the forum.

Cant see one, I assume a log out box should be there for those who want it - but the implication could be that there is no log in box showing if needed. Someone check me here, but .... could be no one can log in if needed.

Regards
Zy

Got compute error Exit status -1073741792 (0xffffffffc0000020) on 43 tasks and the 2 32-bit application files have downloaded again. Server is sending 32-bit GPU tasks to my 64-bit computer? Something awry there, shouldn't be getting 32-bit GPU tasks. I'll whack in an app_info.xml file if it keeps happening, that should tame the errant server.

Cache is now down to 140 tasks, sadly it looks like the big cache was only a temporary aberration.

No GPU work?
upload/download server milkyway Disabled

Since the larger cache limits have been instituted MW is running much more smoothly here. I also notice that my machines are NO LONGER hammering the server every minute or two trying to fill caches that the server wouldn't allow. I would think that server load should be diminished because of this.

Are you saying that you have smooth flowing GPU WU's right now???
All 6 of mine have been sitting idle only offered cpu for hrs!!!

I also get no work for my GPUs :-(. They are doing alternative Project since hours.

I already have more than a thousand WU and I just tried to increase the number of day of cache in my boinc and I had new ones.

Are you saying that you have smooth flowing GPU WU's right now???
All 6 of mine have been sitting idle only offered cpu for hrs!!!

Yes, all my machines are downloading WUs consistently. Cache sizes vary between 155 and 525 WUs depending on GPU speed. They're holding at between an 11 and 12 hour cache. A difference may be that I always use an app_info.xml file to optimize GPU usage.

OMG I have been running the stock app and didn't think about that!!
FIXED
THANK YOU!!!

OMG I have been running the stock app and didn't think about that!!
FIXED
THANK YOU!!!

You're very welcome. Good luck hitting that 1,000,000 RAC!

OMG I have been running the stock app and didn't think about that!!
FIXED
THANK YOU!!!

You're very welcome. Good luck hitting that 1,000,000 RAC!

Ähm what does that mean??
How should it loook like the "app_info.xml"

sry but I am a newby...

This is the one I used that fixed me right up!

<app_info>
<app>
<name>milkyway</name>
</app>
<file_info>
<name>astronomy_0.23_ATI_x64.exe</name>
<executable/>
</file_info>
<file_info>
<name>brook64.dll</name>
<executable/>
</file_info>

<app_version>
<app_name>milkyway</app_name>
<version_num>23</version_num>
<plan_class>ati13ati</plan_class>
<flops>1.0e11</flops>
<avg_ncpus>0.05</avg_ncpus>
<max_ncpus>1</max_ncpus>
<coproc>
<type>ATI</type>
<count>1</count>
</coproc>
<cmdline></cmdline>
<file_ref>
<file_name>astronomy_0.23_ATI_x64.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>brook64.dll</file_name>
</file_ref>
</app_version>

<app_version>
<app_name>milkyway</app_name>
<version_num>22</version_num>
<flops>1.0e11</flops>
<avg_ncpus>0.05</avg_ncpus>
<max_ncpus>1</max_ncpus>
<coproc>
<type>ati13ati</type>
<count>1</count>
</coproc>
<cmdline></cmdline>
<file_ref>
<file_name>astronomy_0.23_ATI_x64.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>brook64.dll</file_name>
</file_ref>
</app_version>

</app_info>

OMG I have been running the stock app and didn't think about that!!
FIXED
THANK YOU!!!

You're very welcome. Good luck hitting that 1,000,000 RAC!

How should it loook like the "app_info.xml"

sry but I am a newby...

You can DL the optimized apps here:

http://www.arkayn.us/milkyway/index.html

They should contain the appropriate default app_info.xml

tried using the various optimised apps, just gives computation errors (no error shown in messages, just premature finish)

still no GPU work here :(

Right now I have over 2100 WU cached, or about 3 days worth.

---

tried using the various optimised apps, just gives computation errors (no error shown in messages, just premature finish)
still no GPU work here :(

From your app messages it looks like you're missing a file in the project directory (one that's referenced in the app_info.xml).

Also using the standard app all 3 of your GPUs are showing invalid results. I'd suspect too high OCing:

CAL Runtime: 1.4.1332
Found 3 CAL devices

Device 0: ATI Radeon HD5800 series (Cypress) 1024 MB local RAM (remote 64 MB cached + 128 MB uncached)
GPU core clock: 969 MHz, memory clock: 1300 MHz
1600 shader units organized in 20 SIMDs with 16 VLIW units (5-issue), wavefront size 64 threads
supporting double precision

Device 1: ATI Radeon HD5800 series (Cypress) 1024 MB local RAM (remote 64 MB cached + 128 MB uncached)
GPU core clock: 969 MHz, memory clock: 1300 MHz
1600 shader units organized in 20 SIMDs with 16 VLIW units (5-issue), wavefront size 64 threads
supporting double precision

Device 2: ATI Radeon HD5800 series (Cypress) 1024 MB local RAM (remote 64 MB cached + 128 MB uncached)
GPU core clock: 969 MHz, memory clock: 1300 MHz
1600 shader units organized in 20 SIMDs with 16 VLIW units (5-issue), wavefront size 64 threads
supporting double precision

Edit: There's 10 invalids there at the moment on the 1 machine. All 3 GPUs are involved:

http://milkyway.cs.rpi.edu/milkyway/results.php?hostid=275082&offset=0&show_names=0&state=4&appid=

Judging by all the replies it looks like I broke everything more than I thought.