significant optimizations to the progs interpreter, changed the runaway loop counter...
authorhavoc <havoc@d7cf8633-e32d-0410-b094-e92efae38249>
Tue, 25 Apr 2006 06:14:44 +0000 (06:14 +0000)
committerhavoc <havoc@d7cf8633-e32d-0410-b094-e92efae38249>
Tue, 25 Apr 2006 06:14:44 +0000 (06:14 +0000)
commit2e45bcb55c7a64d75ffec7e1ad7a4490b754dea6
treef4218cfee0e5c52fddeeb0237d60c5990597fd4f
parent599182488a723028afced5f83d721a9e77f0b7c3
significant optimizations to the progs interpreter, changed the runaway loop counter to only count jump instructions (if/ifnot/goto), and redesigned the profile counting to use pointer math on batches of instructions instead of counting each instruction individually
changed prvm_profile report to show callcount before statements and builtincost, changed the sorting order to use summed callcount+statements+builtincost (and this made it print call counts for builtin functions as well)

git-svn-id: svn://svn.icculus.org/twilight/trunk/darkplaces@6326 d7cf8633-e32d-0410-b094-e92efae38249
prvm_exec.c
prvm_execprogram.h