diff options
| author | Andrea Di Biagio <Andrea_DiBiagio@sn.scee.net> | 2014-05-06 17:09:03 +0000 |
|---|---|---|
| committer | Andrea Di Biagio <Andrea_DiBiagio@sn.scee.net> | 2014-05-06 17:09:03 +0000 |
| commit | 8a712ba229f506ff2d25d75cf2e2fcd980ee0834 (patch) | |
| tree | 6f546e75b9366105ee6827fa0907857c0c2ec368 /test/CodeGen/X86/ret-mmx.ll | |
| parent | 22f779d1fd0d7a7e878a3c28bc5d9910ae655df3 (diff) | |
| download | llvm-8a712ba229f506ff2d25d75cf2e2fcd980ee0834.tar.gz llvm-8a712ba229f506ff2d25d75cf2e2fcd980ee0834.tar.bz2 llvm-8a712ba229f506ff2d25d75cf2e2fcd980ee0834.tar.xz | |
[X86] Improve the lowering of BITCAST dag nodes from type f64 to type v2i32 (and vice versa).
Before this patch, the backend always emitted a store+load sequence to
bitconvert from f64 to i64 the input operand of a ISD::BITCAST dag node that
performed a bitconvert from type MVT::f64 to type MVT::v2i32. The resulting
i64 node was then used to build a v2i32 vector.
With this patch, the backend now produces a cheaper SCALAR_TO_VECTOR from
MVT::f64 to MVT::v2f64. That SCALAR_TO_VECTOR is then followed by a "free"
bitcast to type MVT::v4i32. The elements of the resulting
v4i32 are then extracted to build a v2i32 vector (which is illegal and
therefore promoted to MVT::v2i64).
This is in general cheaper than emitting a stack store+load sequence
to bitconvert the operand from type f64 to type i64.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208107 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/CodeGen/X86/ret-mmx.ll')
| -rw-r--r-- | test/CodeGen/X86/ret-mmx.ll | 5 |
1 files changed, 3 insertions, 2 deletions
diff --git a/test/CodeGen/X86/ret-mmx.ll b/test/CodeGen/X86/ret-mmx.ll index 091fd53984..7adf307ce0 100644 --- a/test/CodeGen/X86/ret-mmx.ll +++ b/test/CodeGen/X86/ret-mmx.ll @@ -33,7 +33,8 @@ define <2 x i32> @t3() nounwind { define double @t4() nounwind { ret double bitcast (<2 x i32> <i32 1, i32 0> to double) ; CHECK-LABEL: t4: -; CHECK: movl $1 -; CHECK: movd {{.*}}, %xmm0 +; CHECK-NOT: movl $1 +; CHECK-NOT: pshufd +; CHECK: movsd {{.*}}, %xmm0 } |