From: "Lee, Bo-sung" Newsgroups: comp.parallel.mpi Subject: Re: [Q] Out-of-order calls in MPI vs. PVM Date: Sat, 03 Jul 1999 23:43:06 +0900 Organization: ETRI Supercomputer Center / Seoul National Univ. Sender: bslee6@210.216.149.243 Message-Id: <377E217A.10B59676@pvmcube4.snu.ac.kr> References: <7l0u34$q9g$1@mozo.cc.purdue.edu> <7l8547$qh0$1@milo.mcs.anl.gov> Reply-To: bslee@pvmcube4.snu.ac.kr Mime-Version: 1.0 Content-Type: text/plain; charset=EUC-KR Content-Transfer-Encoding: 7bit Xref: ukc comp.parallel.mpi:5276 Dear Rusty Lusk. I developed my CFD code using PVM and MPI as the first method you decribed. From now, I think my programs have no problem and I need not non-blocking send-recv functions in my code. But, today I translated my code using MPI_SENDRECV functions as you suggested. And I benchmarked two versions. As you said that the second method is the compact version of the third, so I think the second is more faster than first. But, the first is slightly faster than the second - I think more test needed. Let me know the performance of the alternatives. Bye. Sorry for my poor English. -- Lee, Bo-sung Ph.D Candidate in Aerospace Eng. Seoul National Univ. Invited Reseacher in ETRI/Supercomputing Center mailto:bslee@pvmcube4.snu.ac.kr mailto:bslee@bslee.hpcnet.ne.kr http://www.shinbiro.com/~bslee6