diff options
| author | Xavier Leroy <xavier.leroy@college-de-france.fr> | 2019-05-31 19:15:19 +0200 |
|---|---|---|
| committer | Xavier Leroy <xavierleroy@users.noreply.github.com> | 2019-06-06 10:58:13 +0200 |
| commit | 8e3a73448c5ddfa4be3871d7f4fd80281a7549f4 (patch) | |
| tree | 1d2faaa30db1d4c8d7e236f7ddfbd145d778e6e3 /test/regression/Results | |
| parent | 0025c4db576dbc174946b7adb83d4e0b81ce4b5f (diff) | |
| download | compcert-kvx-8e3a73448c5ddfa4be3871d7f4fd80281a7549f4.tar.gz compcert-kvx-8e3a73448c5ddfa4be3871d7f4fd80281a7549f4.zip | |
If-conversion optimization
Extends the instruction selection pass with an if-conversion optimization:
some if/then/else statements are converted into "select" operations,
which in turn can be compiled down to branchless instruction sequences
if the target architecture supports them.
The statements that are converted are of the form
if (cond) { x = a1; } else { x = a2; }
if (cond) { x = a1; }
if (cond) { /*skip*/; } else { x = a2; }
where a1, a2 are "safe" expressions, containing no operations that can
fail at run-time, such as memory loads or integer divisions.
A heuristic in backend/Selectionaux.ml controls when the optimization occurs,
depending on command-line flags and the complexity of the "then" and "else"
branches.
Diffstat (limited to 'test/regression/Results')
| -rw-r--r-- | test/regression/Results/ifconv | 26 |
1 files changed, 26 insertions, 0 deletions
diff --git a/test/regression/Results/ifconv b/test/regression/Results/ifconv new file mode 100644 index 00000000..38019fe6 --- /dev/null +++ b/test/regression/Results/ifconv @@ -0,0 +1,26 @@ +test1(0,1,12,34) = 12 +test1(1,0,45,67) = 67 +test2(0,1,12,34) = 12 +test2(1,0,45,67) = 67 +test3(0,1,12,34) = 12 +test3(1,0,45,67) = 67 +test4(0,1,12,34) = 12 +test4(1,0,45,67) = 67 +test5(0,1,12) = 13 +test5(1,0,45) = 44 +test6(NULL) = 0 +test6(&i) = 1244 +test7(1,0) = -1 +test7(-100,4) = -25 +test8(0) = 0 +test8(1) = -72 +ltest1(-1, 0, 123LL, 456LL) = 124 +ltest1(1, 0, 123LL, 456LL) = 114 +dmax(0.0, 3.14) = 3.140000 +dmax(1.0, -2.718) = 1.000000 +dabs(1.0) = 1.000000 +dabs(-2.718) = 2.718000 +smin(0.0, 3.14) = 0.000000 +smin(1.0, -2.718) = -2.718000 +sdoz(1.0, 0.5) = 0.500000 +sdoz(0.0, 3.14) = 0.000000 |