[LLVMbugs] [Bug 5209] New: regression on "Canonicalize natural loops" time
bugzilla-daemon at cs.uiuc.edu
bugzilla-daemon at cs.uiuc.edu
Fri Oct 16 03:48:06 PDT 2009
http://llvm.org/bugs/show_bug.cgi?id=5209
Summary: regression on "Canonicalize natural loops" time
Product: libraries
Version: trunk
Platform: PC
OS/Version: Linux
Status: NEW
Severity: normal
Priority: P2
Component: Loop Optimizer
AssignedTo: unassignedbugs at nondot.org
ReportedBy: nunoplopes at sapo.pt
CC: llvmbugs at cs.uiuc.edu
Created an attachment (id=3663)
--> (http://llvm.org/bugs/attachment.cgi?id=3663)
test case
There was a regression in the "Canonicalize natural loops" pass. Right now it
is taking a lot of time in the example attached.
$ time clang -c a.c
real 0m6.649s
user 0m6.512s
sys 0m0.080s
$ time clang -c -O1 a.c
real 6m20.150s
user 6m6.407s
sys 0m0.184s
$ time gcc -c -O2 a.c
real 0m11.844s
user 0m10.969s
sys 0m0.176s
Total Execution Time: 364.3308 seconds (379.7165 wall clock)
---User Time--- --System Time-- --User+System-- ---Wall Time--- ---
Name ---
117.0553 ( 32.2%) 0.0360 ( 7.1%) 117.0913 ( 32.1%) 121.9846 ( 32.1%)
Canonicalize natural loops
78.8049 ( 21.7%) 0.0880 ( 17.3%) 78.8929 ( 21.7%) 83.8040 ( 22.1%)
Canonicalize natural loops
47.6430 ( 13.1%) 0.0120 ( 2.4%) 47.6550 ( 13.1%) 49.4836 ( 13.0%) Loop
Strength Reduction
31.8540 ( 8.8%) 0.0040 ( 0.8%) 31.8580 ( 8.7%) 33.1530 ( 8.7%)
Canonicalize natural loops
16.6970 ( 4.6%) 0.0120 ( 2.4%) 16.7090 ( 4.6%) 17.7542 ( 4.7%)
Canonicalize natural loops
13.7689 ( 3.8%) 0.0040 ( 0.8%) 13.7729 ( 3.8%) 14.4182 ( 3.8%)
Canonicalize natural loops
7.6805 ( 2.1%) 0.0040 ( 0.8%) 7.6845 ( 2.1%) 7.7229 ( 2.0%)
Simple Register Coalescing
5.1363 ( 1.4%) 0.3000 ( 59.1%) 5.4363 ( 1.5%) 5.4532 ( 1.4%) X86
DAG->DAG Instruction Selection
3.5322 ( 1.0%) 0.0000 ( 0.0%) 3.5322 ( 1.0%) 3.5983 ( 0.9%)
Loop-Closed SSA Form Pass
3.5122 ( 1.0%) 0.0000 ( 0.0%) 3.5122 ( 1.0%) 3.5776 ( 0.9%)
Loop-Closed SSA Form Pass
--
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.
More information about the llvm-bugs
mailing list