Tansferring WU to new machine? |
![]() |
| log in |
Message boards : Number crunching : Tansferring WU to new machine?
| Author | Message |
|---|---|
|
What's the best way, very specifically, to transfer un-crunched WU from one machine to another? | |
| ID: 1133003 · | |
|
I think BOINC is specifically designed against that. I know you can't easily transfer work between machines. | |
| ID: 1133005 · | |
|
The easiest way is probably: | |
| ID: 1133007 · | |
The easiest way is probably: Even easier than that. Since BOINC's uninstaller is non-destructive, you can do it like this: 1. Uninstall BOINC on old machine. This leaves behind the data directory where all workunits are stored. 2. Move data directory to the new machine. 3. Install BOINC on the new machine and make sure the installer points to the correct data directory (if it isn't the default) BOINC will automatically pick up where it left off. Note that if you were using optimized apps, you may want to look into re-installing if your new machine supports a newer/better instruction set. | |
| ID: 1133008 · | |
What's the best way, very specifically, to transfer un-crunched WU from one machine to another? You can transfer the identity of a machine to a different machine, including un-crunched work. If you have a slow machine with tons of work, and a fast machine that is empty, you could actually swap identities. What you can't do is take some work on machine "A" and give it to machine "B." One form of widespread cheating in SETI Classic was to have one machine download work, crunch it on multiple machines offline, and then give it back to the original machine to return -- making that original machine seem mind-bogglingly fast. | |
| ID: 1133041 · | |
|
Thanks for the feedback. | |
| ID: 1133067 · | |
What's the best way, very specifically, to transfer un-crunched WU from one machine to another? I used to do this, not to cheat but because machine "A" (at work) had internet access and machine "B" (at home) didn't. I never realised it was considered cheating or that anyone was looking at the performance of machine "A". ____________ Brian. | |
| ID: 1133068 · | |
|
sharing WU's is no longer viable since your machine will eventually want to return the work and you can't have 2 machines with the same ID | |
| ID: 1133087 · | |
I used to do this, not to cheat but because machine "A" (at work) had internet access and machine "B" (at home) didn't. I never realised it was considered cheating or that anyone was looking at the performance of machine "A". I don't think this was ever considered as cheating. What I can remember is that some people ran a task on one pc up to 99%, than distribute that task to multiple machines and let those machines finish the last % and return the result. And that was cheating... ____________ | |
| ID: 1133400 · | |
|
Wouldn't the machine just realize it can't use the old WU's an abandon them? You know clientID and all. Maybe if you did the merger clients thing BEFORE you load them up. | |
| ID: 1133415 · | |
Thanks for the feedback. Depending on what state BOINC is in now you can either do a project detach/attach or just remove BOINC and reinstall it. Then attach to the project. If you still have the data from the old machine intact you might be able to try again. If BOINC aborted the tasks on the new machine then you are out of luck. ____________ SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the BP6/VP6 User Group today! | |
| ID: 1133445 · | |
Message boards : Number crunching : Tansferring WU to new machine?
| Copyright © 2013 University of California |