Your file looks correct now. I don't see how the gpu information would have gotten lodged into the sn2s information, but it was clearly there. When you see that you have received a 32 bit application, can you check to see if the 32 bit app version information is correct? Also - at this point, you should not have any 'psuedo' gpu applications running. The reset should have corrected that.

There's no segregation of work assignment between 32 and 64 bit other then when the _0 copy gets send to a specific platform, any required wingmen go there to. With that in mind, and through noon, you may draw your own conclusions of batch or platform being the cause. If the wingman [if send out too], is also showing longer run times, it's the result that has heavier docking to do. If the chart stays within it's normal flux range and the wingman runs quite a bit faster, ... See http://bit.ly/WCGSNS light blue line how the work run time's been doing since February.

@knreed
Thank you. My machine is now also processing 32 bit work units successfully - they are quite rare on this 64 bit OS.

Lot of invalid results on my Mac OS X 10.8. (with GFAM only 2 sub-projects with invalid results) Putting them on hold and will try them later on Win XP. I hope it will work there...

Ni!