maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

Enable the new coalescer algorithm by default. 

The new coalescer is better at merging values into unused vector lanes,
improving NEON code.

llvm-svn: 164794
This commit is contained in:
Jakob Stoklund Olesen 2012-09-27 21:06:02 +00:00
parent 4976d0df41
commit 8919aa508d
2 changed files with 1 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
llvm/lib/CodeGen/RegisterCoalescer.cpp
View File

@ -70,7 +70,7 @@ VerifyCoalescing("verify-coalescing",
// Temporary option for testing new coalescer algo. // Temporary option for testing new coalescer algo.
static cl::opt<bool> static cl::opt<bool>
NewCoalescer("new-coalescer", cl::Hidden, NewCoalescer("new-coalescer", cl::Hidden, cl::init(true),
cl::desc("Use new coalescer algorithm")); cl::desc("Use new coalescer algorithm"));
namespace { namespace {

1
llvm/test/CodeGen/ARM/reg_sequence.ll
View File

@ -124,7 +124,6 @@ return1:
return2: return2:
; CHECK: %return2 ; CHECK: %return2
; CHECK: vadd.i32 ; CHECK: vadd.i32
; CHECK: vorr {{q[0-9]+}}, {{q[0-9]+}}
; CHECK-NOT: vmov ; CHECK-NOT: vmov
; CHECK: vst2.32 {d{{[0-9]+}}, d{{[0-9]+}}, d{{[0-9]+}}, d{{[0-9]+}}} ; CHECK: vst2.32 {d{{[0-9]+}}, d{{[0-9]+}}, d{{[0-9]+}}, d{{[0-9]+}}}
%tmp100 = extractvalue %struct.__neon_int32x4x2_t %tmp2, 0 ; <<4 x i32>> [#uses=1] %tmp100 = extractvalue %struct.__neon_int32x4x2_t %tmp2, 0 ; <<4 x i32>> [#uses=1]