[llvm] 8689f98 - [ARM] Fix MVE VCMPr f16 pattern

[David Green via llvm-commits]

Author: David Green
Date: 2020-03-27T11:18:24Z
New Revision: 8689f98e9ba5abb1942ad23dc7d08eaf6e0bfad9

URL: https://github.com/llvm/llvm-project/commit/8689f98e9ba5abb1942ad23dc7d08eaf6e0bfad9
DIFF: https://github.com/llvm/llvm-project/commit/8689f98e9ba5abb1942ad23dc7d08eaf6e0bfad9.diff

LOG: [ARM] Fix MVE VCMPr f16 pattern

This patterns seemed to be using the f32 instruction, not f16. Fix it to
use the correct one.

Differential Revision: https://reviews.llvm.org/D76841

Added: 
    

Modified: 
    llvm/lib/Target/ARM/ARMInstrMVE.td
    llvm/test/CodeGen/Thumb2/mve-vcmpfz.ll

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Target/ARM/ARMInstrMVE.td b/llvm/lib/Target/ARM/ARMInstrMVE.td
index fcd3b5775869..06c9a4289274 100644
--- a/llvm/lib/Target/ARM/ARMInstrMVE.td
+++ b/llvm/lib/Target/ARM/ARMInstrMVE.td
@@ -4101,7 +4101,7 @@ multiclass unpred_vcmpf_z<PatLeaf fc> {
                 (v4i1 (MVE_VCMPf32r (v4f32 MQPR:$v1), ZR, fc))>;
 
   def : Pat<(v8i1 (and (v8i1 VCCR:$p1), (v8i1 (ARMvcmpz (v8f16 MQPR:$v1), fc)))),
-            (v8i1 (MVE_VCMPf32r (v8f16 MQPR:$v1), ZR, fc, ARMVCCThen, VCCR:$p1))>;
+            (v8i1 (MVE_VCMPf16r (v8f16 MQPR:$v1), ZR, fc, ARMVCCThen, VCCR:$p1))>;
   def : Pat<(v4i1 (and (v4i1 VCCR:$p1), (v4i1 (ARMvcmpz (v4f32 MQPR:$v1), fc)))),
             (v4i1 (MVE_VCMPf32r (v4f32 MQPR:$v1), ZR, fc, ARMVCCThen, VCCR:$p1))>;
 }

diff  --git a/llvm/test/CodeGen/Thumb2/mve-vcmpfz.ll b/llvm/test/CodeGen/Thumb2/mve-vcmpfz.ll
index 82ef5df349aa..b6c28c3e1016 100644
--- a/llvm/test/CodeGen/Thumb2/mve-vcmpfz.ll
+++ b/llvm/test/CodeGen/Thumb2/mve-vcmpfz.ll
@@ -1010,7 +1010,7 @@ define arm_aapcs_vfpcc <8 x half> @vcmp_one_v8f16(<8 x half> %src, <8 x half> %a
 ; CHECK-MVEFP-LABEL: vcmp_one_v8f16:
 ; CHECK-MVEFP:       @ %bb.0: @ %entry
 ; CHECK-MVEFP-NEXT:    vpt.f16 ge, q0, zr
-; CHECK-MVEFP-NEXT:    vcmpt.f32 le, q0, zr
+; CHECK-MVEFP-NEXT:    vcmpt.f16 le, q0, zr
 ; CHECK-MVEFP-NEXT:    vpnot
 ; CHECK-MVEFP-NEXT:    vpsel q0, q1, q2
 ; CHECK-MVEFP-NEXT:    bx lr
@@ -1628,7 +1628,7 @@ define arm_aapcs_vfpcc <8 x half> @vcmp_ueq_v8f16(<8 x half> %src, <8 x half> %a
 ; CHECK-MVEFP-LABEL: vcmp_ueq_v8f16:
 ; CHECK-MVEFP:       @ %bb.0: @ %entry
 ; CHECK-MVEFP-NEXT:    vpt.f16 ge, q0, zr
-; CHECK-MVEFP-NEXT:    vcmpt.f32 le, q0, zr
+; CHECK-MVEFP-NEXT:    vcmpt.f16 le, q0, zr
 ; CHECK-MVEFP-NEXT:    vpsel q0, q1, q2
 ; CHECK-MVEFP-NEXT:    bx lr
 entry:
@@ -2353,7 +2353,7 @@ define arm_aapcs_vfpcc <8 x half> @vcmp_ord_v8f16(<8 x half> %src, <8 x half> %a
 ; CHECK-MVEFP-LABEL: vcmp_ord_v8f16:
 ; CHECK-MVEFP:       @ %bb.0: @ %entry
 ; CHECK-MVEFP-NEXT:    vpt.f16 ge, q0, zr
-; CHECK-MVEFP-NEXT:    vcmpt.f32 lt, q0, zr
+; CHECK-MVEFP-NEXT:    vcmpt.f16 lt, q0, zr
 ; CHECK-MVEFP-NEXT:    vpnot
 ; CHECK-MVEFP-NEXT:    vpsel q0, q1, q2
 ; CHECK-MVEFP-NEXT:    bx lr
@@ -2475,7 +2475,7 @@ define arm_aapcs_vfpcc <8 x half> @vcmp_uno_v8f16(<8 x half> %src, <8 x half> %a
 ; CHECK-MVEFP-LABEL: vcmp_uno_v8f16:
 ; CHECK-MVEFP:       @ %bb.0: @ %entry
 ; CHECK-MVEFP-NEXT:    vpt.f16 ge, q0, zr
-; CHECK-MVEFP-NEXT:    vcmpt.f32 lt, q0, zr
+; CHECK-MVEFP-NEXT:    vcmpt.f16 lt, q0, zr
 ; CHECK-MVEFP-NEXT:    vpsel q0, q1, q2
 ; CHECK-MVEFP-NEXT:    bx lr
 entry:
@@ -3495,7 +3495,7 @@ define arm_aapcs_vfpcc <8 x half> @vcmp_r_one_v8f16(<8 x half> %src, <8 x half>
 ; CHECK-MVEFP-LABEL: vcmp_r_one_v8f16:
 ; CHECK-MVEFP:       @ %bb.0: @ %entry
 ; CHECK-MVEFP-NEXT:    vpt.f16 le, q0, zr
-; CHECK-MVEFP-NEXT:    vcmpt.f32 ge, q0, zr
+; CHECK-MVEFP-NEXT:    vcmpt.f16 ge, q0, zr
 ; CHECK-MVEFP-NEXT:    vpnot
 ; CHECK-MVEFP-NEXT:    vpsel q0, q1, q2
 ; CHECK-MVEFP-NEXT:    bx lr
@@ -4113,7 +4113,7 @@ define arm_aapcs_vfpcc <8 x half> @vcmp_r_ueq_v8f16(<8 x half> %src, <8 x half>
 ; CHECK-MVEFP-LABEL: vcmp_r_ueq_v8f16:
 ; CHECK-MVEFP:       @ %bb.0: @ %entry
 ; CHECK-MVEFP-NEXT:    vpt.f16 le, q0, zr
-; CHECK-MVEFP-NEXT:    vcmpt.f32 ge, q0, zr
+; CHECK-MVEFP-NEXT:    vcmpt.f16 ge, q0, zr
 ; CHECK-MVEFP-NEXT:    vpsel q0, q1, q2
 ; CHECK-MVEFP-NEXT:    bx lr
 entry:
@@ -4838,7 +4838,7 @@ define arm_aapcs_vfpcc <8 x half> @vcmp_r_ord_v8f16(<8 x half> %src, <8 x half>
 ; CHECK-MVEFP-LABEL: vcmp_r_ord_v8f16:
 ; CHECK-MVEFP:       @ %bb.0: @ %entry
 ; CHECK-MVEFP-NEXT:    vpt.f16 le, q0, zr
-; CHECK-MVEFP-NEXT:    vcmpt.f32 gt, q0, zr
+; CHECK-MVEFP-NEXT:    vcmpt.f16 gt, q0, zr
 ; CHECK-MVEFP-NEXT:    vpnot
 ; CHECK-MVEFP-NEXT:    vpsel q0, q1, q2
 ; CHECK-MVEFP-NEXT:    bx lr
@@ -4960,7 +4960,7 @@ define arm_aapcs_vfpcc <8 x half> @vcmp_r_uno_v8f16(<8 x half> %src, <8 x half>
 ; CHECK-MVEFP-LABEL: vcmp_r_uno_v8f16:
 ; CHECK-MVEFP:       @ %bb.0: @ %entry
 ; CHECK-MVEFP-NEXT:    vpt.f16 le, q0, zr
-; CHECK-MVEFP-NEXT:    vcmpt.f32 gt, q0, zr
+; CHECK-MVEFP-NEXT:    vcmpt.f16 gt, q0, zr
 ; CHECK-MVEFP-NEXT:    vpsel q0, q1, q2
 ; CHECK-MVEFP-NEXT:    bx lr
 entry: