[PATCH] D36775: Increase tail dup threshold for -O3 from 3 to 4

[Richard Smith - zygoloid via Phabricator via llvm-commits]

rsmith created this revision.
Herald added subscribers: sdardis, sanjoy.

We see a modest performance improvement from this slightly higher tail dup threshold (~0.1% geomean across the suite).


Repository:
  rL LLVM

https://reviews.llvm.org/D36775

Files:
  lib/CodeGen/MachineBlockPlacement.cpp
  test/CodeGen/Mips/brconge.ll
  test/CodeGen/Mips/brconle.ll
  test/CodeGen/X86/tail-dup-repeat.ll


Index: test/CodeGen/X86/tail-dup-repeat.ll
===================================================================
--- test/CodeGen/X86/tail-dup-repeat.ll
+++ test/CodeGen/X86/tail-dup-repeat.ll
@@ -1,4 +1,4 @@
-; RUN: llc -O3 -tail-dup-placement-threshold=4 -o - %s | FileCheck %s
+; RUN: llc -O3 -o - %s | FileCheck %s
 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64-unknown-linux-gnu"
 
Index: test/CodeGen/Mips/brconle.ll
===================================================================
--- test/CodeGen/Mips/brconle.ll
+++ test/CodeGen/Mips/brconle.ll
@@ -1,4 +1,4 @@
-; RUN: llc  -march=mipsel -mattr=mips16 -relocation-model=pic -O3 < %s | FileCheck %s -check-prefix=16
+; RUN: llc  -march=mipsel -mattr=mips16 -relocation-model=pic -O2 < %s | FileCheck %s -check-prefix=16
 
 @i = global i32 -5, align 4
 @j = global i32 10, align 4
Index: test/CodeGen/Mips/brconge.ll
===================================================================
--- test/CodeGen/Mips/brconge.ll
+++ test/CodeGen/Mips/brconge.ll
@@ -1,4 +1,4 @@
-; RUN: llc  -march=mipsel -mattr=mips16 -relocation-model=pic -O3 < %s | FileCheck %s -check-prefix=16
+; RUN: llc  -march=mipsel -mattr=mips16 -relocation-model=pic -O2 < %s | FileCheck %s -check-prefix=16
 
 @i = global i32 5, align 4
 @j = global i32 10, align 4
Index: lib/CodeGen/MachineBlockPlacement.cpp
===================================================================
--- lib/CodeGen/MachineBlockPlacement.cpp
+++ lib/CodeGen/MachineBlockPlacement.cpp
@@ -143,7 +143,7 @@
     "tail-dup-placement-aggressive-threshold",
     cl::desc("Instruction cutoff for aggressive tail duplication during "
              "layout. Used at -O3. Tail merging during layout is forced to "
-             "have a threshold that won't conflict."), cl::init(3),
+             "have a threshold that won't conflict."), cl::init(4),
     cl::Hidden);
 
 // Heuristic for tail duplication.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D36775.111283.patch
Type: text/x-patch
Size: 1943 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20170815/11298d7c/attachment.bin>