[PATCH] D67269: [ARM] Add patterns for VSUB with q and r registers

[oliver cruickshank via Phabricator via llvm-commits]

oliverlars created this revision.
oliverlars added reviewers: dmgreen, SjoerdMeijer, samparker, simon_tatham, t.p.northover.
Herald added subscribers: llvm-commits, hiraditya, kristof.beyls, javed.absar.
Herald added a project: LLVM.

Added patterns for VSUB to support q and r registers, which reduces pressure on q registers.


Repository:
  rL LLVM

https://reviews.llvm.org/D67269

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


Index: llvm/test/CodeGen/Thumb2/mve-vsubqr.ll
===================================================================
--- /dev/null
+++ llvm/test/CodeGen/Thumb2/mve-vsubqr.ll
@@ -0,0 +1,36 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+define arm_aapcs_vfpcc <4 x i32> @vsubqr_v4i32(<4 x i32> %src, i32 %src2, <4 x i32> %a, <4 x i32> %b) {
+; CHECK-LABEL: vsubqr_v4i32:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vsub.i32 q0, q0, r0
+; CHECK-NEXT:    bx lr
+entry:
+  %i = insertelement <4 x i32> undef, i32 %src2, i32 0
+  %sp = shufflevector <4 x i32> %i, <4 x i32> undef, <4 x i32> zeroinitializer
+  %c = sub <4 x i32> %src, %sp
+  ret <4 x i32> %c
+}
+
+define arm_aapcs_vfpcc <8 x i16> @vsubqr_v8i16(<8 x i16> %src, i16 %src2, <8 x i16> %a, <8 x i16> %b) {
+; CHECK-LABEL: vsubqr_v8i16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vsub.i16 q0, q0, r0
+; CHECK-NEXT:    bx lr
+entry:
+  %i = insertelement <8 x i16> undef, i16 %src2, i32 0
+  %sp = shufflevector <8 x i16> %i, <8 x i16> undef, <8 x i32> zeroinitializer
+  %c = sub <8 x i16> %src, %sp
+  ret <8 x i16> %c
+}
+
+define arm_aapcs_vfpcc <16 x i8> @vsubqr_v16i8(<16 x i8> %src, i8 %src2, <16 x i8> %a, <16 x i8> %b) {
+; CHECK-LABEL: vsubqr_v16i8:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vsub.i8 q0, q0, r0
+; CHECK-NEXT:    bx lr
+entry:
+  %i = insertelement <16 x i8> undef, i8 %src2, i32 0
+  %sp = shufflevector <16 x i8> %i, <16 x i8> undef, <16 x i32> zeroinitializer
+  %c = sub <16 x i8> %src, %sp
+  ret <16 x i8> %c
+}
Index: llvm/lib/Target/ARM/ARMInstrMVE.td
===================================================================
--- llvm/lib/Target/ARM/ARMInstrMVE.td
+++ llvm/lib/Target/ARM/ARMInstrMVE.td
@@ -3625,6 +3625,15 @@
             (v4i32 (MVE_VADD_qr_i32 (v4i32 MQPR:$val1), (i32 GPR:$val2)))>;
 }
 
+let Predicates = [HasMVEInt] in {
+  def : Pat<(v16i8 (sub (v16i8 MQPR:$val1), (v16i8 (ARMvdup GPR:$val2)))),
+            (v16i8 (MVE_VSUB_qr_i8 (v16i8 MQPR:$val1), (i32 GPR:$val2)))>;
+  def : Pat<(v8i16 (sub (v8i16 MQPR:$val1), (v8i16 (ARMvdup GPR:$val2)))),
+            (v8i16 (MVE_VSUB_qr_i16 (v8i16 MQPR:$val1), (i32 GPR:$val2)))>;
+  def : Pat<(v4i32 (sub (v4i32 MQPR:$val1), (v4i32 (ARMvdup GPR:$val2)))),
+            (v4i32 (MVE_VSUB_qr_i32 (v4i32 MQPR:$val1), (i32 GPR:$val2)))>;
+}
+
 class MVE_VQDMULL_qr<string iname, string suffix, bit size,
                      bit T, list<dag> pattern=[]>
   : MVE_qDest_rSrc<iname, suffix, pattern> {


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D67269.219079.patch
Type: text/x-patch
Size: 2508 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20190906/b9b009b7/attachment-0001.bin>