summaryrefslogtreecommitdiffstats
path: root/llvm/test/Bitcode/thinlto-alias.ll
diff options
context:
space:
mode:
authorPiotr Padlewski <piotr.padlewski@gmail.com>2016-09-26 20:37:32 +0000
committerPiotr Padlewski <piotr.padlewski@gmail.com>2016-09-26 20:37:32 +0000
commitd9830eb79fdc42368d370abeab9a3b56c08e3963 (patch)
tree20fab6b2cd9ae3a74f212b7d4bd4814727a91717 /llvm/test/Bitcode/thinlto-alias.ll
parent2cd84c905df4fe9a6005320a45e98cd4b8ade045 (diff)
downloadbcm5719-llvm-d9830eb79fdc42368d370abeab9a3b56c08e3963.tar.gz
bcm5719-llvm-d9830eb79fdc42368d370abeab9a3b56c08e3963.zip
[thinlto] Basic thinlto fdo heuristic
Summary: This patch improves thinlto importer by importing 3x larger functions that are called from hot block. I compared performance with the trunk on spec, and there were about 2% on povray and 3.33% on milc. These results seems to be consistant and match the results Teresa got with her simple heuristic. Some benchmarks got slower but I think they are just noisy (mcf, xalancbmki, omnetpp)- running the benchmarks again with more iterations to confirm. Geomean of all benchmarks including the noisy ones were about +0.02%. I see much better improvement on google branch with Easwaran patch for pgo callsite inlining (the inliner actually inline those big functions) Over all I see +0.5% improvement, and I get +8.65% on povray. So I guess we will see much bigger change when Easwaran patch will land (it depends on new pass manager), but it is still worth putting this to trunk before it. Implementation details changes: - Removed CallsiteCount. - ProfileCount got replaced by Hotness - hot-import-multiplier is set to 3.0 for now, didn't have time to tune it up, but I see that we get most of the interesting functions with 3, so there is no much performance difference with higher, and binary size doesn't grow as much as with 10.0. Reviewers: eraman, mehdi_amini, tejohnson Subscribers: mehdi_amini, llvm-commits Differential Revision: https://reviews.llvm.org/D24638 llvm-svn: 282437
Diffstat (limited to 'llvm/test/Bitcode/thinlto-alias.ll')
-rw-r--r--llvm/test/Bitcode/thinlto-alias.ll4
1 files changed, 2 insertions, 2 deletions
diff --git a/llvm/test/Bitcode/thinlto-alias.ll b/llvm/test/Bitcode/thinlto-alias.ll
index 58411ca6523..cfdf8f7b0bd 100644
--- a/llvm/test/Bitcode/thinlto-alias.ll
+++ b/llvm/test/Bitcode/thinlto-alias.ll
@@ -9,7 +9,7 @@
; CHECK-NEXT: <VERSION
; See if the call to func is registered, using the expected callsite count
; and value id matching the subsequent value symbol table.
-; CHECK-NEXT: <PERMODULE {{.*}} op4=[[FUNCID:[0-9]+]] op5=1/>
+; CHECK-NEXT: <PERMODULE {{.*}} op4=[[FUNCID:[0-9]+]]/>
; CHECK-NEXT: </GLOBALVAL_SUMMARY_BLOCK>
; CHECK-NEXT: <VALUE_SYMTAB
; CHECK-NEXT: <FNENTRY {{.*}} record string = 'main'
@@ -21,7 +21,7 @@
; COMBINED-NEXT: <VERSION
; See if the call to analias is registered, using the expected callsite count
; and value id matching the subsequent value symbol table.
-; COMBINED-NEXT: <COMBINED {{.*}} op5=[[ALIASID:[0-9]+]] op6=1/>
+; COMBINED-NEXT: <COMBINED {{.*}} op5=[[ALIASID:[0-9]+]]/>
; Followed by the alias and aliasee
; COMBINED-NEXT: <COMBINED {{.*}}
; COMBINED-NEXT: <COMBINED_ALIAS {{.*}} op3=[[ALIASEEID:[0-9]+]]
OpenPOWER on IntegriCloud