trampoline and return counts
Suresh Jagannathan
suresh@research.nj.nec.com
Fri, 12 Feb 1999 17:50:32 -0500
The following compares recursive inlining (3 rounds)
with the current inliner showing trampoline, return,
and limit-check counts. (I can't explain mandelbrot.)
-- sj
==============================================================
Recursive Inlining (3 rounds)
=============================
0 Trampolines Returns SelfReturns LimitChecks
fft 76.8 904 451 0 13127749
barnes-hut 32.4 21145136 19455812 18359650 40205829
matrix-multiply 29.2 2022 2010 2000 804040
knuth-bendix 24.7 46609246 14797062 1222600 36955278
lexgen 71.3 62923733 31910804 11093377 74765219
count-graphs 53.8 94111012 30589640 9413430 103914097
life 238.1 32114478 16032254 1170282 1180211922
logic 167.6 323245632 141277320 57933510 135459010
mandelbrot 32.6 2 0 0 0
mlyacc 59.2 64791599 40931448 25263339 93986068
nucleic 95.8 13149402 6471400 212200 36324802
simple 58.1 84025450 53255374 17055610 54301155
tsp 72.0 8862576 3939782 983010 987150
=====================================================================
Simple inlining
---------------
0 Trampolines Returns SelfReturns LimitCheck
fft 86.2 20975390 10487694 0 7888027
barnes-hut 71.7 103426210 49053934 2639790 70612146
matrix-multiply 31.0 2062 2030 2000 404040
knuth-bendix 20.6 32588646 13649402 1659590 38889188
lexgen 149.3 224654657 117465176 11093377 76776358
count-graphs 35.6 47907976 22990027 9413430 90931930
life 149.7 26126793 13063414 2605042 1183031432
logic 178.1 282401452 136706120 57933510 193460360
mandelbrot 32.6 2 0 0 0
mlyacc 65.8 78967069 52091511 25827539 95086878
nucleic 78.7 51962802 25871000 212200 41171602
simple 58.6 84474228 56529038 19073110 50332516
tsp 131.3 173113244 87048126 975360 1002447