We have a paper on the very topic that Jeff just mentioned :
Subodh Sharma, Sarvani Vakkalanka, Ganesh Gopalakrishnan, Robert M.
Kirby, Rajeev Thakur, and William Gropp, `` A Formal Approach to Detect
Functionally Irrelevant Barriers in MPI Programs,'' Recent Advances in
Parallel Virtual Machine and Message Passing Interface (EuroPVM/MPI),
Dublin, Ireland, 2008, LNCS 5205, pp. 265-273.
This is available from http://www.cs.utah.edu/formal_verification/ under
Publications.
The "FIB" facility (ability to detect functionally irrelevant barriers)
exists in verification tool ISP which can be downloaded from here
http://www.cs.utah.edu/formal_verification/ISP-release
Ganesh
Jeff Squyres wrote:
On Mar 5, 2009, at 9:29 AM, Shanyuan Gao wrote:
I am doing some research on MPI barrier operations. And I am ready
to do some performance test.
I wonder if there are any applications that using barriers a lot.
Please let me know if there
is any. Any comments are welcomed. Thanks!
I don't remember who originally said it, but I've repeated the
statement: any MPI program that relies on a barrier for correctness is
an incorrect MPI application.
There's anecdotal evidence that throwing in a barrier every once in a
while can help reduce unexpected messages (and other things), and
therefore improve performance a bit. But that's very application
dependent, and usually not frequent.
Why do you want to do a barrier? Particularly one that sounds like it
might be in your critical performance path?
