LHC@home

I have an older Pentium machine with 82 MB RAM that downloaded a WU and began processing it.

Originally, the time to completion was before the WU deadline. However, after running it for nearly 25 hours, achieving ~9% completion, the time to completion has increased (by an additional 8 hours!), and I fear it will not get done in time.

I would like to move it to my Pent. 4 laptop and was wondering if there is anything special I need to do?

I see the init_data.xml file contains a path to the files which are on a different drive in the older machine and I will edit this. Anything else?

Expiration is in four days.

Thanks,
Charles

---

I have an older Pentium machine with 82 MB RAM that downloaded a WU and began processing it.

Originally, the time to completion was before the WU deadline. However, after running it for nearly 25 hours, achieving ~9% completion, the time to completion has increased (by an additional 8 hours!), and I fear it will not get done in time.

I would like to move it to my Pent. 4 laptop and was wondering if there is anything special I need to do?

I see the init_data.xml file contains a path to the files which are on a different drive in the older machine and I will edit this. Anything else?

Expiration is in four days.

Moving a single workunit to another machine? You would need to do lots of manual edits to client_state.xml; and I'd almost guarantee you'll screw up in the process. Moving the slot directory for that workunit is just the beginning.

---

You would need to do lots of manual edits to client_state.xml; and I'd almost guarantee you'll screw up in the process.

Oh ye of little faith!

Thanks for the quick response.
If you know of somewhere which gives an overview I would be interested in seeing it, none-the-less.


I take it, what I am suggesting is not recommended, but I'd hate to run it out, invest four days and not have it finish and was hoping to avoid that especially in light of how seldom work is available for LHC.

I have managed to successfully finish WU for SETI and Spinhenge on this old machine and assumed I would not be sent any work unless it was able to finish in a timely manner. It has never happened in the 2+ years I have run BOINC.

Do you think I should abort it now and cut my losses or try to get done as much as possible?

---

What I've done in the past is to:

Close up BOINC as much as possible on the machine that's going to be too slow - No New Tasks, network activity disabled, that sort of thing.

Shut down any copy of BOINC that's running on the fast machine that's going to take over the processing. If it's a service install, disable the auto-start of the BOINC service: if if runs as single user or all users, remove/rename any startup icon.

Take the whole BOINC folder from the slow machine, and move it to the fast one. Use a network connection or USB stick: if you have to use a CDR, compress (zip, rar, 7-zip etc.) the folder first, so you don't get any problems with read-only files.

On the fast machine, put the BOINC folder in a convenient place of its own - don't let it interact with the existing installation in any way.

Locate boincmgr.exe in the guest folder, and double-click it to run it. BOINC processing should pick up exactly where it was suspended on the old machine - once you've checked that it's running cleanly, re-enable networking and allow the WUs to upload/report as normal. Once they've finished, shut down the visiting copy of BOINC, and re-enable the usual copy. That's all.

re init_data.xml: are you say that the data folders are outside the normal BOINC folder tree, or just that the whole installation is in a non-standard location? If the data folders are outside the BOINC tree, then you'll probably have to make matching changes to the xml file(s): but if they're in a single tree, then I think you can just copy and run. I haven't changed drives, but I moved it from C:\\Program files to C:\\ when I did an in-situ upgrade from XP to Vista, and it ran immediately with no errors and no hand-editing (including 3 CPDN models).

The only other thing to watch out for is if you run processor-specific optimised SETI applications: make sure that the temporary host has at least a compatible set of SIMD extensions (MMX, SSEn) of the opt. app. the donor is running - not usually a problem if you're moving to a faster machine, but I had to be careful once when I finished off a Xeon-Core 2 test run on an old P4.

Thank You Richard.

Actually, that is very similar to how I had to update BOINC the last time on the laptop and probably again after I get the other issue sorted.