diff options
author | Andrew Trick <atrick@apple.com> | 2011-09-29 01:33:38 +0000 |
---|---|---|
committer | Andrew Trick <atrick@apple.com> | 2011-09-29 01:33:38 +0000 |
commit | 0c01bc385a4c01bee012bda504c8ce0c3d402f2c (patch) | |
tree | 9b57d78a2b2d96043cf7ca24139ac4080e413fee /test/CodeGen/ARM | |
parent | 03b08764d2d1735c33710950be4207edc9b90fb2 (diff) | |
download | llvm-0c01bc385a4c01bee012bda504c8ce0c3d402f2c.tar.gz llvm-0c01bc385a4c01bee012bda504c8ce0c3d402f2c.tar.bz2 llvm-0c01bc385a4c01bee012bda504c8ce0c3d402f2c.tar.xz |
LSR: rewrite inner loops only.
Rewriting the entire loop nest now requires -enable-lsr-nested.
See PR11035 for some performance data.
A few unit tests specifically test nested LSR, and are now under a flag.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@140762 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/CodeGen/ARM')
-rw-r--r-- | test/CodeGen/ARM/lsr-on-unrolled-loops.ll | 5 |
1 files changed, 4 insertions, 1 deletions
diff --git a/test/CodeGen/ARM/lsr-on-unrolled-loops.ll b/test/CodeGen/ARM/lsr-on-unrolled-loops.ll index 140b02cb39..47379016cf 100644 --- a/test/CodeGen/ARM/lsr-on-unrolled-loops.ll +++ b/test/CodeGen/ARM/lsr-on-unrolled-loops.ll @@ -1,4 +1,4 @@ -; RUN: llc -mtriple=thumbv7-apple-darwin10 -mcpu=cortex-a8 < %s | FileCheck %s +; RUN: llc -mtriple=thumbv7-apple-darwin10 -mcpu=cortex-a8 -enable-lsr-nested < %s | FileCheck %s ; LSR should recognize that this is an unrolled loop which can use ; constant offset addressing, so that each of the following stores @@ -8,6 +8,9 @@ ; CHECK: vstr.32 s{{.*}}, [{{(r[0-9]+)|(lr)}}, #64] ; CHECK: vstr.32 s{{.*}}, [{{(r[0-9]+)|(lr)}}, #96] +; We can also save a register in the outer loop, but that requires +; performing LSR on the outer loop. + target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:32-f32:32:32-f64:32:32-v64:64:64-v128:128:128-a0:0:32-n32" %0 = type { %1*, %3*, %6*, i8*, i32, i32, %8*, i32, i32, i32, i32, i32, i32, i32, double, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i8**, i32, i32, i32, i32, i32, [64 x i32]*, [4 x %9*], [4 x %10*], [4 x %10*], i32, %11*, i32, i32, [16 x i8], [16 x i8], [16 x i8], i32, i32, i8, i8, i8, i16, i16, i32, i8, i32, %12*, i32, i32, i32, i32, i8*, i32, [4 x %11*], i32, i32, i32, [10 x i32], i32, i32, i32, i32, i32, %13*, %14*, %15*, %16*, %17*, %18*, %19*, %20*, %21*, %22*, %23* } |