diff options
| author | Andrew Trick <atrick@apple.com> | 2011-04-13 00:38:32 +0000 |
|---|---|---|
| committer | Andrew Trick <atrick@apple.com> | 2011-04-13 00:38:32 +0000 |
| commit | 87896d9368e08d93493427ce7bf8272d1e5cca35 (patch) | |
| tree | d3183747b2917bf4e1254c7a331ff86fd3352a2f /test/CodeGen/ARM/vfp.ll | |
| parent | f93f7b2446bec3febc30b7136e18704664bd98cc (diff) | |
Recommit r129383. PreRA scheduler heuristic fixes: VRegCycle, TokenFactor latency.
Additional fixes:
Do something reasonable for subtargets with generic
itineraries by handle node latency the same as for an empty
itinerary. Now nodes default to unit latency unless an itinerary
explicitly specifies a zero cycle stage or it is a TokenFactor chain.
Original fixes:
UnitsSharePred was a source of randomness in the scheduler: node
priority depended on the queue data structure. I rewrote the recent
VRegCycle heuristics to completely replace the old heuristic without
any randomness. To make the ndoe latency adjustments work, I also
needed to do something a little more reasonable with TokenFactor. I
gave it zero latency to its consumers and always schedule it as low as
possible.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@129421 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/CodeGen/ARM/vfp.ll')
| -rw-r--r-- | test/CodeGen/ARM/vfp.ll | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/test/CodeGen/ARM/vfp.ll b/test/CodeGen/ARM/vfp.ll index 390457fc21b..49a69827bc0 100644 --- a/test/CodeGen/ARM/vfp.ll +++ b/test/CodeGen/ARM/vfp.ll @@ -40,8 +40,8 @@ define void @test_add(float* %P, double* %D) { define void @test_ext_round(float* %P, double* %D) { ;CHECK: test_ext_round: %a = load float* %P ; <float> [#uses=1] -;CHECK: vcvt.f32.f64 ;CHECK: vcvt.f64.f32 +;CHECK: vcvt.f32.f64 %b = fpext float %a to double ; <double> [#uses=1] %A = load double* %D ; <double> [#uses=1] %B = fptrunc double %A to float ; <float> [#uses=1] |