[PATCH] D63257: [ARM] Select MVE fp add and sub

[Dave Green via Phabricator via llvm-commits]

dmgreen updated this revision to Diff 206164.
dmgreen added a reviewer: SjoerdMeijer.
dmgreen added a comment.

Rebased for new patterns and moved next to instruction definition.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D63257/new/

https://reviews.llvm.org/D63257

Files:
  llvm/lib/Target/ARM/ARMInstrMVE.td
  llvm/test/CodeGen/Thumb2/mve-simple-arith.ll


Index: llvm/test/CodeGen/Thumb2/mve-simple-arith.ll
===================================================================
--- llvm/test/CodeGen/Thumb2/mve-simple-arith.ll
+++ llvm/test/CodeGen/Thumb2/mve-simple-arith.ll
@@ -31,6 +31,26 @@
   ret <4 x i32> %0
 }
 
+define arm_aapcs_vfpcc <8 x half> @add_float16_t(<8 x half> %src1, <8 x half> %src2) {
+; CHECK-LABEL: add_float16_t:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vadd.f16 q0, q1, q0
+; CHECK-NEXT:    bx lr
+entry:
+  %0 = fadd nnan ninf nsz <8 x half> %src2, %src1
+  ret <8 x half> %0
+}
+
+define arm_aapcs_vfpcc <4 x float> @add_float32_t(<4 x float> %src1, <4 x float> %src2) {
+; CHECK-LABEL: add_float32_t:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vadd.f32 q0, q1, q0
+; CHECK-NEXT:    bx lr
+entry:
+  %0 = fadd nnan ninf nsz <4 x float> %src2, %src1
+  ret <4 x float> %0
+}
+
 
 define arm_aapcs_vfpcc <16 x i8> @sub_int8_t(<16 x i8> %src1, <16 x i8> %src2) {
 ; CHECK-LABEL: sub_int8_t:
@@ -62,3 +82,23 @@
   ret <4 x i32> %0
 }
 
+define arm_aapcs_vfpcc <8 x half> @sub_float16_t(<8 x half> %src1, <8 x half> %src2) {
+; CHECK-LABEL: sub_float16_t:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vsub.f16 q0, q1, q0
+; CHECK-NEXT:    bx lr
+entry:
+  %0 = fsub nnan ninf nsz <8 x half> %src2, %src1
+  ret <8 x half> %0
+}
+
+define arm_aapcs_vfpcc <4 x float> @sub_float32_t(<4 x float> %src1, <4 x float> %src2) {
+; CHECK-LABEL: sub_float32_t:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vsub.f32 q0, q1, q0
+; CHECK-NEXT:    bx lr
+entry:
+  %0 = fsub nnan ninf nsz <4 x float> %src2, %src1
+  ret <4 x float> %0
+}
+
Index: llvm/lib/Target/ARM/ARMInstrMVE.td
===================================================================
--- llvm/lib/Target/ARM/ARMInstrMVE.td
+++ llvm/lib/Target/ARM/ARMInstrMVE.td
@@ -2060,9 +2060,23 @@
 def MVE_VADDf32 : MVE_VADDSUBFMA_fp<"vadd", "f32", 0b0, 0b0, 0b1, 0b0>;
 def MVE_VADDf16 : MVE_VADDSUBFMA_fp<"vadd", "f16", 0b1, 0b0, 0b1, 0b0>;
 
+let Predicates = [HasMVEFloat] in {
+  def : Pat<(v4f32 (fadd (v4f32 MQPR:$val1), (v4f32 MQPR:$val2))),
+            (v4f32 (MVE_VADDf32 (v4f32 MQPR:$val1), (v4f32 MQPR:$val2)))>;
+  def : Pat<(v8f16 (fadd (v8f16 MQPR:$val1), (v8f16 MQPR:$val2))),
+            (v8f16 (MVE_VADDf16 (v8f16 MQPR:$val1), (v8f16 MQPR:$val2)))>;
+}
+
 def MVE_VSUBf32 : MVE_VADDSUBFMA_fp<"vsub", "f32", 0b0, 0b0, 0b1, 0b1>;
 def MVE_VSUBf16 : MVE_VADDSUBFMA_fp<"vsub", "f16", 0b1, 0b0, 0b1, 0b1>;
 
+let Predicates = [HasMVEFloat] in {
+  def : Pat<(v4f32 (fsub (v4f32 MQPR:$val1), (v4f32 MQPR:$val2))),
+            (v4f32 (MVE_VSUBf32 (v4f32 MQPR:$val1), (v4f32 MQPR:$val2)))>;
+  def : Pat<(v8f16 (fsub (v8f16 MQPR:$val1), (v8f16 MQPR:$val2))),
+            (v8f16 (MVE_VSUBf16 (v8f16 MQPR:$val1), (v8f16 MQPR:$val2)))>;
+}
+
 class MVE_VCADD<string suffix, bit size, list<dag> pattern=[]>
   : MVEFloatArithNeon<"vcadd", suffix, size, (outs MQPR:$Qd),
                          (ins MQPR:$Qn, MQPR:$Qm, complexrotateopodd:$rot),


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D63257.206164.patch
Type: text/x-patch
Size: 2999 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20190624/2c2a8df2/attachment.bin>