So I was thinking about putting in an application for that 48 core Magny Cours rig that AMD is offering for free and use it for bigadv crunching. My internet connection is slow as all hell. It takes me 20+ minutes to upload a regular A2 or A3 unit, let alone a bigadv unit. As close as I can figure this 48 core machine would finish a bigadv unit in about 17-18 hours, then on my connection could take over an hour to upload while the CPUs sit idle.
What I'm thinking is to set up 2 clients which would not run simultaneously. Each client would run with the -oneunit flag. When one unit is complete or nearly complete, a script would run to start the other client. The new client would download a new WU and begin crunching while the original client would finish up and upload the completed work, then terminate.
@ the beta testers:
Any idea if there is a change in the works to download a new WU earlier rather than waiting for the upload to complete?
What I'm thinking is to set up 2 clients which would not run simultaneously. Each client would run with the -oneunit flag. When one unit is complete or nearly complete, a script would run to start the other client. The new client would download a new WU and begin crunching while the original client would finish up and upload the completed work, then terminate.
@ the beta testers:
Any idea if there is a change in the works to download a new WU earlier rather than waiting for the upload to complete?