Installing a second GPU

[MarkOne]

Originally Posted by RONEDOGG

when i add a second gpu i always clean up the old. first i uninstall any gpu related drivers and soft ware. second i shut down pc and put in the second card. then i start back up and install drivers fresh for both cards. to avoid any conflicts at all.

it's what I will do, if I see no change tomorow,

[MarkOne]

I thing I found the problem they are P10101 works

[MarkOne]

now GPU2 get an other work unit it's running at 8k PPD, but I have an error that he can't send the result and put it in queue , something I should worry about ?

edit : gpu1 set the results successfully, and get a new work too .

when I should expect the gpu2 to try to send it again ?

[lowfat]

Originally Posted by MarkOne

now GPU2 get an other work unit it's running at 8k PPD, but I have an error that he can't send the result and put it in queue , something I should worry about ?

edit : gpu1 set the results successfully, and get a new work too .

when I should expect the gpu2 to try to send it again ?

Give it time. You really shouldn't need to worry about the GPU clients at all as long as you have a stable overclock on them.

[Zero82z]

Originally Posted by MarkOne

now GPU2 get an other work unit it's running at 8k PPD, but I have an error that he can't send the result and put it in queue , something I should worry about ?

edit : gpu1 set the results successfully, and get a new work too .

when I should expect the gpu2 to try to send it again ?

Mark, the only time you should ever worry about the clients is if you see UNSTABLE_MACHINE, EARLY_UNIT_END, or if the client just stops folding for no reason. Any other type of error, and it is most likely on Stanford's end so you just have to wait until it gets taken care of.

[MarkOne]

so how it work when it finish the second work it will send both results ?

[Zero82z]

The client will periodically retry sending the unit until it succeeds. Eventually it will get uploaded.

[MarkOne]

is that synchronize at your taste.....I means the PPD , I will call them the twin , so bad new don't have them anymore, I love those GPU

[lowfat]

If you use HFM.net instead of Fahmon you can see if the core how many times the core has succeeded/failed. That is the only thing I ever look at for my GPU clients.

[MarkOne]

the gpu2 send the last work successfully, the first work in queue was resent with always the same error , I will take it as 500 point lost : Server does not have record of this unit