[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

computational efficiency lisping



Greetings,

I am starting to work on an fairly complex computational module...In the spirit 
of Jim Hurd's recent remark about in-lineing svref and aref calls, Can you sug-
gest methods of programming style and object design for the highest efficiency? 
I'm going to have many nested loops over arrays with multiplies etc. 

1. Is there a way to bring in-line, say a multiply within these loops?. Should
row-major-aref in-line? (Perhaps in 2.0 final?) Is that the norm for traversing
large matrices?

2. Does it make sense, for example, in the case of a 2-d array to allocate every-
thing as a simple-vectors? i.e., a simple-vector to hold each row (column) as sim-
ple-vectors?


On this note, I would recommend the topic of efficiency (at least as far as is 
addressed in CLtL2 regarding Steels so-called "pervasive" compiler directives), 
to be included in the 2.1 goodie list...




 Mike Meehan
 Simulation Optimization Technology Group
 EEsof Inc., R&D Department
 5601 Lindero Canyon Road
 Westlake Village CA. 91362
 FAX/VOICE: 818.879.6374

 mmeehan@eesof.com