Hey Crew,
I've got a few questions that may seem odd, but you guys are likely to know the answer, so we go........
I'm about to swap out another Nvidia card for a new AMD card.
Last time I did this I didn't think I had a choice but to let all the NVIDIA GPU tasks simply expire.
But this time I am thinking that I might be able to cut all the "zipped" NVIDIA tasks from the project folder (cache ??) and use a USB stick to transfer them onto another PC running the NVIDIA tasks.
Would this work ?? or are the tasks now allocated to that pc when downloaded ??
Going one step further, could I "cut and transfer" the tasks currently running that are in the "slots" folders, after suspending the project via the manager ??
And if so,
do tasks from slot 0 for example have to go into slot 0 folder on the "pasting" (receiving) machine ?
The "cutting" (donor) pc has 25 slots, and "pasting" receiving machine only has 8 slots.
Don't really want to abort the tasks or let them time out (wasting WCG bandwidth), but I do want to swap cards out today at some stage.
Your advice welcomed.
Thanks,
underwater

You NEED version 7.044 (or later)
Once installed - make sure that BOINC client it is NOT running before doing the following steps with the XML files.
Look for a "cc_config" XML file which is usually found in C:\ProgramData\BOINC - (Windows will have the ProgramData folder hidden by default) (open it with notepad to edit)
If "cc_config" not in Boinc folder create an XML file with the following, if it is there ensure the following option is there "<use_all_gpus>1</use_all_gpus>"

<cc_config>
<options>
<use_all_gpus>1</use_all_gpus>
</options>
</cc_config>

That little XML file tells BOINC to use all gpus in your computer - not just the best one (usually in pci-e slot 1)
So now you can utilize both GPU's, how to best use them ??

Another XML file, "app_config" needs to be in place for manipulating gpu task setting to run multiple tasks per gpu
It should be located at C:\ProgramData\BOINC\projects\www.worldcommunitygrid.org

<app_config>
<app>
<name>hcc1</name>
<gpu_versions>
<gpu_usage>XXX</gpu_usage> (XXX = how many work units per gpu card)
<cpu_usage>YYY</cpu_usage> (YYY = how much CPU to allocate per GPU task
</gpu_versions>
</app>
</app_config>

Once saved I fire up Boinc and let it download a few WU's then deactive network activity while testing the new app_config settings.
This will show any compuatation errors and not submit straight away so you wouldn't know about them 'til too late.

Any dramas, just shout out, some one will be able to help.

The BOINC Manager now shows Running (0.5C + 0.25ATI GPUs) on each for the four GPU Tasks. Most interesting.
This frees up two C threads for individual tasking.
The ATIs now equates to 4 x .25 or equates to one GPU.
How do I make use of the other four 0.25ATIs or does it work that way?

FYI - My second computer is running LinuxMint Cinnamon DE.

Could you, please, post the first 35 - 40 rows of Event Log? Or, re-start your PC and post entire Event Log

Thanks

post edited and settings changed