You are here

rosetta 3.1-how to run parallel tasks?

5 posts / 0 new
Last post
rosetta 3.1-how to run parallel tasks?
#1

Hi,
I found that the 3.1 version is more difficult to use the older one 2.3. Especially, the parallel doesn't work.
Isn't the newer version poorer than the previous release? How do the authors think about it? I never see any other kinds of newer suites like this case.

Fri, 2009-11-06 03:23
wtscrystal

What do you mean by "the parallel doesn't work?" The question is a little too vague to troubleshoot.

Fri, 2009-11-06 06:55
smlewis

> Hi,
> I found that the 3.1 version is more difficult to use the older one 2.3. Especially, the parallel doesn't work.
> Isn't the newer version poorer than the previous release? How do the authors think about it? I never see any other kinds of newer suites like this case.

Just like lennyly have said:
~~#FF0000:Thanks, smlewis.

Yes, "scons bin mode=release extras=mpi" can make execution code with supporting MPI. But how to run it, I mean the command options specification? It seems Rosetta3 changes a lot in command options from Rosetta 2.3.

I only find a piece of such information, such like:

-jdist_rerun -process: :condor -process: :process......

but it does not seem to work on my cluster machine, which well serves Rosetta 2.3 in MPI mode.

thanks, len ~~

how to run ?

The user guide looks like nonsense and useless. And the command line also have any indications about how to use the options for the parallel binary code.
Why?
If you use any of the command under linux, just like "ls, mv, etc", there are many mannuals for their use and various options.

If you haven't complete this suite, don't release it. Don't let the user do the try-and-error thing, and waste time to guess how to use it.

In addition, in the user guide, there is a simple protocol walkthourgh, I read many times and didn't know what the code is and where it is located. I just found two word about it, that is "at FILE". What does it means?

Sun, 2009-11-08 02:54
wtscrystal

The parallel code runs more or less automatically. The decision to run in parallel or single-processor is a compile-time decision, not a runtime decision. So, if you decided parallel and compiled it that way, you don't need to give Rosetta more command-line flags to get it to run in parallel. You run the executeable under mpirun, same as 2.x.

Mon, 2009-11-09 07:38
smlewis

I switched from rosetta2.3 to rosetta3.1 recently. I found that rosetta3.1 could run faster than rosetta2.3 on the same job using one node. Maybe that is the improved part. But I used to do perturbative docking and ensemble docking, could i do the same kind of thing in rosetta3.1? I could see that some of the flags are gone.

In rosetta2.3, I used to do prepacking before docking. Could I do the same kind of thing in rosetta3.1?

Anyone knows my answer? please share them with me. Thanks a lot.

> Hi,
> I found that the 3.1 version is more difficult to use the older one 2.3. Especially, the parallel doesn't work.
> Isn't the newer version poorer than the previous release? How do the authors think about it? I never see any other kinds of newer suites like this case.

Thu, 2010-01-07 11:47
lqzhang