Newsgroups: comp.parallel.mpi
From: kohr@donner.mcs.anl.gov (Dave Kohr)
Subject: Re: Concurrency doubts
Organization: MCS, Argonne National Laboratory
Date: Fri, 6 Sep 1996 18:0:26 GMT
Message-ID: <84203282619666@donner.mcs.anl.gov>

In article <Pine.SGI.3.91.960905115218.5716C-100000@hurricane>,
Luke A. Lonergan <luke@hurricane> wrote:
> Is there a strong reason not to implement MPI-2 thread safe?

One reason might be if you have to pay the cost of thread safety even when
you're not using threads.  A common way for this to happen is when a
library locks and unlocks its data structures regardless of how many
threads there are.  I know there are ways around this problem, so you could
argue that this is more of a quality of implementation issue, but I think
it still needs to be addressed if you want to support threads but get the
best performance in all cases.

Dave Kohr    Hacker/Researcher    Argonne National Laboratory, MCS Division
Building 203, Room C-246   Phone: (630) 252-4243   E-mail: kohr@mcs.anl.gov
See also my WWW Home Page:      http://www.mcs.anl.gov/home/kohr/index.html
                      "Surfing the Silicon Prairie."

