[Ecls-list] ECL behavior on some array related code

[Juan Jose Garcia-Ripoll]

ECL chooses integer types that are large enough to fit the user type
declarations. It leaves the C compiler the task of choosing the appropriate
internal representation and it seems to do it just fine.

Using int8_t

real time : 0.475 secs
run time  : 0.463 secs
gc count  : 2 times
consed    : 4000105 bytes
#P"/Users/jjgarcia/build/ecl/faa.lsp"

Using cl_fixnum = int on a 32 bits machine

real time : 0.475 secs
run time  : 0.462 secs
gc count  : 1 times
consed    : 4000088 bytes
#P"/Users/jjgarcia/build/ecl/faa.lsp"

This is with a slightly larger number of executions than Jeronimo programmed
(200 instead of 50)

For this larger an more meaningful number we reach SBCL:

Evaluation took:
  0.485 seconds of real time
  0.480578 seconds of total run time (0.473391 user, 0.007187 system)
  99.18% CPU
  967,508,892 processor cycles
  4,000,008 bytes consed

BTW, I just uploded the code. But please expect no improvement on stability
-- I am currently using OpenAxiom and other large pieces of software as a
consistency check and it takes about five hours to build it all and make
sure I did not break any hidden assumption.

Juanjo

-- 
Instituto de Física Fundamental, CSIC
c/ Serrano, 113b, Madrid 28006 (Spain)
http://tream.dreamhosters.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.common-lisp.net/pipermail/ecl-devel/attachments/20100519/9a761147/attachment.html>