Changeset 123 for pypar/README

Timestamp:

Jul 11, 2005, 3:20:03 PM (20 years ago)

Author:

ole

Message:

Files for 64 bit machine + latest Cvs version

File:

• pypar/README (modified) (4 diffs)

pypar/README

@@ -179 +179 @@
   
 HISTORY
+version 1.9.1 (15 Dec 2003)
+  Consistent naming (all lower case, no abbreviations)
+version 1.9 (3 Dec 2003)
+  Obsoleted raw forms of communication as they were confusing.
+  The semantics of send, receive, scatter, gather etc is now that 
+  one can optionally specify a buffer to use if desired.
+  Bypass forms added.
+  
 version 1.8.2 (16 November 2003)
   Fixed scatter, gather and reduce calls to automatically
@@ -235 +243 @@
   
 TODO  
-  Modify scatter, gather and reduce to automatically determine length
-  in order to comply with the other functions: send, receive and bcast 
-  which do.
-  
-  Get rid of 'raw' forms as they are confusing. 
-  Incorporate a keyword argument, buffer = x, instead.
-  This gives the flexibility to specify an existing receive 
-  buffer. The corresponding send command will need a flag, buffer=1,
-  or something like that.
-
   max_tag is set to 32767. This works for Linux/LAM      
   I couldn't use MPI_TAG_UB as it returned 0.
@@ -250 +248 @@
 
 
+  Scatter needs to send buffer specified on all processes
+  even though it is ignored by all non-root processes.
+  How could we overcome that?
+  Similar problem for gather
+  
+  Gather an scatter: One should be able to specify along which axis
+  arrays should be concatenated.
+
+
 KNOWN BUGS
-  In raw_receive one must assign explicitly to buffer variable
-  if it uses (or reverts to) vanilla mode:
-  B = raw_receive(B, ....)  
-    
-  The collective communication have not yet been tested for multidimensional arrays 
-  and the complex datatype
-  
+  Scatter for the moment works only properly when the amount of data is a 
+  multiple of the number of processors (as does the underlying MPI_Scatter).
+  I am working on a more general scatter (and gather) which will
+  distribute data as evenly as possible for all amounts of data.  
+
       
 LICENSE
@@ -323 +328 @@
 
 
+
+
+