--- /dev/null
+
+ $Name: fa_34_26_5 $ - $Id: readme.pvm_3.4,v 1.3 2001/09/17 21:18:19 wrp Exp $
+
+
+20-August-2001
+
+The pvm/mpi complib programs have been substantially updated with
+release 3.4. See readme.v34t0 for more information. With version
+3.4, the MPI programs are mp34comp*, mu34comp*, etc.
+
+A major effect of this change is to disable automatic sequence type
+(protein/DNA) recognition with pv34compfa/mp34compfa. By default,
+protein libraries are assumed. Thus, pv34compfa/mp34compfa require
+the "-n" command line option when running pv34compfa/mp34compfa on DNA
+sequence libraries. This issue does not occur with the other
+programs, which will recognize the appropriate sequence type, because
+it is determined by the program (e.g. pv34compfx requires
+DNA:protein).
+
+================
+pv4comp* - July, August, 2000
+
+As noted in readme.pvm_3.3 - the major problem that users have had
+with the PVM/MPI version of the programs is in reading database files
+on the nodes. All previous versions of the program (pvcompfa,
+pv3compfa, etc) had the nodes read the databases in parallel. Thus,
+the database file had to be visible to the nodes, typically through
+NFS on modern clusters of workstations.
+
+This strategy caused some problems. It did not work on beowulf-type
+systems, where most of the nodes are in an isolated local network and
+do not have NFS access to the outside world. And it made it
+complicated to read more than one database file. Because specialized
+functions were used, the nodes could not read the full set of library
+file formats available to the other fasta programs.
+
+These problems have been addressed by significantly changing the the
+way the pv4comp*/mp4comp* programs read the second "reference"
+library. With these versions, both databases, but specifically the
+reference library, are read by a manager process. The manager process
+then sends the sequences to the workers. This solves problems with
+NFS reads from the workers (they don't do any), and uses exactly the
+same functions as the other fasta programs, so the full set of
+database formats can be read. In addition, the FASTLIBS database
+abbreviations are available. This also should also solve problems with
+searches of very long sequences (bacterial genomes); they can now be
+broken up into smaller pieces with the -N ##### option, as with
+fasta33/tfastx33.
+
+Thus, you are encouraged to use the pv4comp*/mp4comp* versions of the
+programs, which should run more like fasta33.
+
+================
+Program summary:
+
+Programs to produce conventional scores and alignments:
+
+pv4compfa protein vs protein, DNA vs DNA
+pv4compsw protein vs protein, DNA vs DNA
+pv4compfx/ DNA vs protein
+pv4comptfx/y protein vs DNA
+
+Programs to summarize the effectiveness of a search (require
+super-family-labeled databases):
+
+ps4compfa protein vs protein, DNA vs DNA
+ps4compsw protein vs protein, DNA vs DNA
+ps4compfx/ DNA vs protein
+ps4comptfx/y protein vs DNA
+
+Programs to report the scores and alignments of the highest scoring
+unrelated sequence (require super-family-labeled databases). These
+programs are used to evaluate the super-family labeling.
+
+pu4compfa protein vs protein, DNA vs DNA
+pu4compsw protein vs protein, DNA vs DNA
+pucompfx/ DNA vs protein
+pu4comptfx/y protein vs DNA
+
+================
+Release notes:
+
+--> Aug. 4, 2000
+
+Compiled and tested mp4compfa/mp4compsw programs.
+
+--> July 22, 2000
+
+First release of restructured p2_complib.c/p2_workcomp.c, which use
+the manager program to read both sequence databases and send the
+"reference database" to the workers.
+