home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.parallel
- Path: sparky!uunet!destroyer!gatech!hubcap!fpst
- From: standers@tera.cs.umn.edu (Steve Anderson)
- Subject: Re: Question on the CM5 CMfortran optimisation
- Message-ID: <1993Jan26.160325.8738@hubcap.clemson.edu>
- Apparently-To: comp-parallel@uunet.uu.net
- Summary: CMF compiler optimization may be wrong
- Keywords: CMfortran, CM5,MPP
- Sender: news@news2.cis.umn.edu (Usenet News Administration)
- Nntp-Posting-Host: tera.cs.umn.edu
- Organization: University of Minnesota, Minneapolis, CSci dept.
- References: <1993Jan20.124731.7725@hubcap.clemson.edu> <1993Jan25.163635.11981@hubcap.clemson.edu>
- Date: Mon, 25 Jan 1993 21:40:48 GMT
- Approved: parallel@hubcap.clemson.edu
- Lines: 18
-
- I should amplify on my previous note. While it is true that some CMF
- codes fail without "-g", lots of codes run fine. It is unfortunate
- that one such code should be mine :) Such problems are to be expected
- in a beta release of anything, even e-mail. Actually, the compiler
- is in a useful state right now, and the eagerly awaited 2.1 release
- promises to increase performance in several more important ways than
- by the instruction block size (The primary effect of -g at this
- release.)
-
- Also, to respond to the original question better: CMF "on a node" will be
- in the next release of CMMD. Until then, there is no vectorizing non-data
- parallel compiler.
-
- Steve Anderson
-
-
-
-
-