My personal feeling is that if I found a project taking about a week per work unit on a computer, I would decide that it was intended for faster computers and I would select a project with shorter work units, like Genome Comparison.

Hi joneill003,
As far as I know, the plan is still 'Early estimates are that Phase 2 may start around August of 2007' as stated in http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=10647

There has already been one change with the decision to run many more test proteins in Phase 1. The remaining proteins tend to be much larger than the earlier ones (at least, they take much longer to run). I would not be surprised if Phase 2 was delayed a bit to change the code, but I only know what I see here.

Lawrence

hi everybody,

just to let know everybody that there are truly long works units coming. It had been warned that coming to the end of the project some of the units would be very long in calculation, not all of them.

Indeed some units would crunch normally (about 14h on my slowest computer P4 2.8 mhz), but others have taken over 75 hours on my fastest computer.

I don't see why it should be a problem for a work unit to be long to complete, even more when there is no way to make them smaller. Results are still needed ... Time out is long enough to still complete the units. It would be a subject for concern if most units would fail but then I'm sure worldcommunity would know before any of us crunchers.

The reactions about long work units make me think that a central project with difficult crunching might get difficulty to be achieved just because people want units to be fast probably making them think reasearch go as fast as their units.

I think recurent timeout should really be the only reason for desactivating a project apart for personnal choices of course.

steve davol wrote:
The latter part of HCMD is very tough. It's going to take about 40% of the projects total CPU time estimate to do about 10% of the remainder of the project. On very slow CPU's reaching the first checkpoint which will be reflected in a jump from 0.2% to about 4-5%, can take many hours... after that you should see regular progress.

Each time you shut down the PC and not reaching the first checkpoint will in effect restart the work unit the next day. Hibernation (not stand-by), would preserve the memory state. Given you got nowhere, propose to first change your 'My Projects' profile to do a few short Genome Comparison, to test the speed (What is the CPU speed?). Once you changed the profile, go to Task Manager (Ctrl-Alt-Del keys simultaneous), select the Process tab and find the wcgridHCMD.exe process. Terminate that and it should send the interrupted result back and get a GC job.

Added: Getting worse: now it's 30% for the remaining 3%:

Hello steve davol,
The initial test of HCMD showed that they were going to have to rewrite Phase 2 a lot more than they had hoped, so they extended Phase 1 by adding on some much larger proteins to get all the data they needed for the Phase 2 programming. So HCMD now takes a lot of time. Safer not to run it on a 1.2 Ghz Celeron.

Lawrence