Huge performance regression with strings Just two examples, with strcat and
Reported by François VOGEL
-- Bug description --
Huge performance regression with strings
Just two examples, with strcat and part:
function perfoA
tic;
cont = strcat(string(1:100000),',');
t=toc();mprintf("Performance strcat: %f\n",t);
// disp(length(cont))
tic;
for i=1:10000
p = part(cont,100:150);
end
t=toc();mprintf("Performance part: %f\n",t);
endfunction
Scilab 5:
-->perfoA
Performance strcat: 17.359000
Performance part: 168.047000
Scilab 4.1.2:
-->perfoA
Performance strcat: 0.281000
Performance part: 0.047000
Performance ratio on strcat:
-->17.813/0.297
ans =
59.976431
Performance ratio on part:
-->168.047/0.047
ans =
3575.4681
Francois
-- Scilab error message --
-- How to reproduce the bug --