| ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py |
| ; RUN: llc -mtriple=riscv32 -mattr=+experimental-v -riscv-v-vector-bits-min=128 -verify-machineinstrs < %s \ |
| ; RUN: | FileCheck %s --check-prefixes=CHECK,RV32 |
| ; RUN: llc -mtriple=riscv64 -mattr=+experimental-v -riscv-v-vector-bits-min=128 -verify-machineinstrs < %s \ |
| ; RUN: | FileCheck %s --check-prefixes=CHECK,RV64 |
| |
| declare i8 @llvm.vp.reduce.add.v2i8(i8, <2 x i8>, <2 x i1>, i32) |
| |
| define signext i8 @vpreduce_add_v2i8(i8 signext %s, <2 x i8> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_add_v2i8: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e8, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e8, mf8, tu, mu |
| ; CHECK-NEXT: vredsum.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i8 @llvm.vp.reduce.add.v2i8(i8 %s, <2 x i8> %v, <2 x i1> %m, i32 %evl) |
| ret i8 %r |
| } |
| |
| declare i8 @llvm.vp.reduce.umax.v2i8(i8, <2 x i8>, <2 x i1>, i32) |
| |
| define signext i8 @vpreduce_umax_v2i8(i8 signext %s, <2 x i8> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_umax_v2i8: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: andi a0, a0, 255 |
| ; CHECK-NEXT: vsetvli a2, zero, e8, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e8, mf8, tu, mu |
| ; CHECK-NEXT: vredmaxu.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i8 @llvm.vp.reduce.umax.v2i8(i8 %s, <2 x i8> %v, <2 x i1> %m, i32 %evl) |
| ret i8 %r |
| } |
| |
| declare i8 @llvm.vp.reduce.smax.v2i8(i8, <2 x i8>, <2 x i1>, i32) |
| |
| define signext i8 @vpreduce_smax_v2i8(i8 signext %s, <2 x i8> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_smax_v2i8: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e8, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e8, mf8, tu, mu |
| ; CHECK-NEXT: vredmax.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i8 @llvm.vp.reduce.smax.v2i8(i8 %s, <2 x i8> %v, <2 x i1> %m, i32 %evl) |
| ret i8 %r |
| } |
| |
| declare i8 @llvm.vp.reduce.umin.v2i8(i8, <2 x i8>, <2 x i1>, i32) |
| |
| define signext i8 @vpreduce_umin_v2i8(i8 signext %s, <2 x i8> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_umin_v2i8: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: andi a0, a0, 255 |
| ; CHECK-NEXT: vsetvli a2, zero, e8, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e8, mf8, tu, mu |
| ; CHECK-NEXT: vredminu.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i8 @llvm.vp.reduce.umin.v2i8(i8 %s, <2 x i8> %v, <2 x i1> %m, i32 %evl) |
| ret i8 %r |
| } |
| |
| declare i8 @llvm.vp.reduce.smin.v2i8(i8, <2 x i8>, <2 x i1>, i32) |
| |
| define signext i8 @vpreduce_smin_v2i8(i8 signext %s, <2 x i8> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_smin_v2i8: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e8, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e8, mf8, tu, mu |
| ; CHECK-NEXT: vredmin.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i8 @llvm.vp.reduce.smin.v2i8(i8 %s, <2 x i8> %v, <2 x i1> %m, i32 %evl) |
| ret i8 %r |
| } |
| |
| declare i8 @llvm.vp.reduce.and.v2i8(i8, <2 x i8>, <2 x i1>, i32) |
| |
| define signext i8 @vpreduce_and_v2i8(i8 signext %s, <2 x i8> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_and_v2i8: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e8, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e8, mf8, tu, mu |
| ; CHECK-NEXT: vredand.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i8 @llvm.vp.reduce.and.v2i8(i8 %s, <2 x i8> %v, <2 x i1> %m, i32 %evl) |
| ret i8 %r |
| } |
| |
| declare i8 @llvm.vp.reduce.or.v2i8(i8, <2 x i8>, <2 x i1>, i32) |
| |
| define signext i8 @vpreduce_or_v2i8(i8 signext %s, <2 x i8> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_or_v2i8: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e8, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e8, mf8, tu, mu |
| ; CHECK-NEXT: vredor.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i8 @llvm.vp.reduce.or.v2i8(i8 %s, <2 x i8> %v, <2 x i1> %m, i32 %evl) |
| ret i8 %r |
| } |
| |
| declare i8 @llvm.vp.reduce.xor.v2i8(i8, <2 x i8>, <2 x i1>, i32) |
| |
| define signext i8 @vpreduce_xor_v2i8(i8 signext %s, <2 x i8> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_xor_v2i8: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e8, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e8, mf8, tu, mu |
| ; CHECK-NEXT: vredxor.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i8 @llvm.vp.reduce.xor.v2i8(i8 %s, <2 x i8> %v, <2 x i1> %m, i32 %evl) |
| ret i8 %r |
| } |
| |
| declare i8 @llvm.vp.reduce.add.v4i8(i8, <4 x i8>, <4 x i1>, i32) |
| |
| define signext i8 @vpreduce_add_v4i8(i8 signext %s, <4 x i8> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_add_v4i8: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e8, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e8, mf4, tu, mu |
| ; CHECK-NEXT: vredsum.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i8 @llvm.vp.reduce.add.v4i8(i8 %s, <4 x i8> %v, <4 x i1> %m, i32 %evl) |
| ret i8 %r |
| } |
| |
| declare i8 @llvm.vp.reduce.umax.v4i8(i8, <4 x i8>, <4 x i1>, i32) |
| |
| define signext i8 @vpreduce_umax_v4i8(i8 signext %s, <4 x i8> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_umax_v4i8: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: andi a0, a0, 255 |
| ; CHECK-NEXT: vsetvli a2, zero, e8, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e8, mf4, tu, mu |
| ; CHECK-NEXT: vredmaxu.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i8 @llvm.vp.reduce.umax.v4i8(i8 %s, <4 x i8> %v, <4 x i1> %m, i32 %evl) |
| ret i8 %r |
| } |
| |
| declare i8 @llvm.vp.reduce.smax.v4i8(i8, <4 x i8>, <4 x i1>, i32) |
| |
| define signext i8 @vpreduce_smax_v4i8(i8 signext %s, <4 x i8> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_smax_v4i8: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e8, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e8, mf4, tu, mu |
| ; CHECK-NEXT: vredmax.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i8 @llvm.vp.reduce.smax.v4i8(i8 %s, <4 x i8> %v, <4 x i1> %m, i32 %evl) |
| ret i8 %r |
| } |
| |
| declare i8 @llvm.vp.reduce.umin.v4i8(i8, <4 x i8>, <4 x i1>, i32) |
| |
| define signext i8 @vpreduce_umin_v4i8(i8 signext %s, <4 x i8> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_umin_v4i8: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: andi a0, a0, 255 |
| ; CHECK-NEXT: vsetvli a2, zero, e8, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e8, mf4, tu, mu |
| ; CHECK-NEXT: vredminu.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i8 @llvm.vp.reduce.umin.v4i8(i8 %s, <4 x i8> %v, <4 x i1> %m, i32 %evl) |
| ret i8 %r |
| } |
| |
| declare i8 @llvm.vp.reduce.smin.v4i8(i8, <4 x i8>, <4 x i1>, i32) |
| |
| define signext i8 @vpreduce_smin_v4i8(i8 signext %s, <4 x i8> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_smin_v4i8: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e8, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e8, mf4, tu, mu |
| ; CHECK-NEXT: vredmin.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i8 @llvm.vp.reduce.smin.v4i8(i8 %s, <4 x i8> %v, <4 x i1> %m, i32 %evl) |
| ret i8 %r |
| } |
| |
| declare i8 @llvm.vp.reduce.and.v4i8(i8, <4 x i8>, <4 x i1>, i32) |
| |
| define signext i8 @vpreduce_and_v4i8(i8 signext %s, <4 x i8> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_and_v4i8: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e8, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e8, mf4, tu, mu |
| ; CHECK-NEXT: vredand.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i8 @llvm.vp.reduce.and.v4i8(i8 %s, <4 x i8> %v, <4 x i1> %m, i32 %evl) |
| ret i8 %r |
| } |
| |
| declare i8 @llvm.vp.reduce.or.v4i8(i8, <4 x i8>, <4 x i1>, i32) |
| |
| define signext i8 @vpreduce_or_v4i8(i8 signext %s, <4 x i8> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_or_v4i8: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e8, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e8, mf4, tu, mu |
| ; CHECK-NEXT: vredor.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i8 @llvm.vp.reduce.or.v4i8(i8 %s, <4 x i8> %v, <4 x i1> %m, i32 %evl) |
| ret i8 %r |
| } |
| |
| declare i8 @llvm.vp.reduce.xor.v4i8(i8, <4 x i8>, <4 x i1>, i32) |
| |
| define signext i8 @vpreduce_xor_v4i8(i8 signext %s, <4 x i8> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_xor_v4i8: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e8, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e8, mf4, tu, mu |
| ; CHECK-NEXT: vredxor.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i8 @llvm.vp.reduce.xor.v4i8(i8 %s, <4 x i8> %v, <4 x i1> %m, i32 %evl) |
| ret i8 %r |
| } |
| |
| declare i16 @llvm.vp.reduce.add.v2i16(i16, <2 x i16>, <2 x i1>, i32) |
| |
| define signext i16 @vpreduce_add_v2i16(i16 signext %s, <2 x i16> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_add_v2i16: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e16, mf4, tu, mu |
| ; CHECK-NEXT: vredsum.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i16 @llvm.vp.reduce.add.v2i16(i16 %s, <2 x i16> %v, <2 x i1> %m, i32 %evl) |
| ret i16 %r |
| } |
| |
| declare i16 @llvm.vp.reduce.umax.v2i16(i16, <2 x i16>, <2 x i1>, i32) |
| |
| define signext i16 @vpreduce_umax_v2i16(i16 signext %s, <2 x i16> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_umax_v2i16: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: lui a2, 16 |
| ; RV32-NEXT: addi a2, a2, -1 |
| ; RV32-NEXT: and a0, a0, a2 |
| ; RV32-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; RV32-NEXT: vmv.v.x v9, a0 |
| ; RV32-NEXT: vsetvli zero, a1, e16, mf4, tu, mu |
| ; RV32-NEXT: vredmaxu.vs v9, v8, v9, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v9 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_umax_v2i16: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: lui a2, 16 |
| ; RV64-NEXT: addiw a2, a2, -1 |
| ; RV64-NEXT: and a0, a0, a2 |
| ; RV64-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v9, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e16, mf4, tu, mu |
| ; RV64-NEXT: vredmaxu.vs v9, v8, v9, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v9 |
| ; RV64-NEXT: ret |
| %r = call i16 @llvm.vp.reduce.umax.v2i16(i16 %s, <2 x i16> %v, <2 x i1> %m, i32 %evl) |
| ret i16 %r |
| } |
| |
| declare i16 @llvm.vp.reduce.smax.v2i16(i16, <2 x i16>, <2 x i1>, i32) |
| |
| define signext i16 @vpreduce_smax_v2i16(i16 signext %s, <2 x i16> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_smax_v2i16: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e16, mf4, tu, mu |
| ; CHECK-NEXT: vredmax.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i16 @llvm.vp.reduce.smax.v2i16(i16 %s, <2 x i16> %v, <2 x i1> %m, i32 %evl) |
| ret i16 %r |
| } |
| |
| declare i16 @llvm.vp.reduce.umin.v2i16(i16, <2 x i16>, <2 x i1>, i32) |
| |
| define signext i16 @vpreduce_umin_v2i16(i16 signext %s, <2 x i16> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_umin_v2i16: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: lui a2, 16 |
| ; RV32-NEXT: addi a2, a2, -1 |
| ; RV32-NEXT: and a0, a0, a2 |
| ; RV32-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; RV32-NEXT: vmv.v.x v9, a0 |
| ; RV32-NEXT: vsetvli zero, a1, e16, mf4, tu, mu |
| ; RV32-NEXT: vredminu.vs v9, v8, v9, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v9 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_umin_v2i16: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: lui a2, 16 |
| ; RV64-NEXT: addiw a2, a2, -1 |
| ; RV64-NEXT: and a0, a0, a2 |
| ; RV64-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v9, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e16, mf4, tu, mu |
| ; RV64-NEXT: vredminu.vs v9, v8, v9, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v9 |
| ; RV64-NEXT: ret |
| %r = call i16 @llvm.vp.reduce.umin.v2i16(i16 %s, <2 x i16> %v, <2 x i1> %m, i32 %evl) |
| ret i16 %r |
| } |
| |
| declare i16 @llvm.vp.reduce.smin.v2i16(i16, <2 x i16>, <2 x i1>, i32) |
| |
| define signext i16 @vpreduce_smin_v2i16(i16 signext %s, <2 x i16> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_smin_v2i16: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e16, mf4, tu, mu |
| ; CHECK-NEXT: vredmin.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i16 @llvm.vp.reduce.smin.v2i16(i16 %s, <2 x i16> %v, <2 x i1> %m, i32 %evl) |
| ret i16 %r |
| } |
| |
| declare i16 @llvm.vp.reduce.and.v2i16(i16, <2 x i16>, <2 x i1>, i32) |
| |
| define signext i16 @vpreduce_and_v2i16(i16 signext %s, <2 x i16> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_and_v2i16: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e16, mf4, tu, mu |
| ; CHECK-NEXT: vredand.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i16 @llvm.vp.reduce.and.v2i16(i16 %s, <2 x i16> %v, <2 x i1> %m, i32 %evl) |
| ret i16 %r |
| } |
| |
| declare i16 @llvm.vp.reduce.or.v2i16(i16, <2 x i16>, <2 x i1>, i32) |
| |
| define signext i16 @vpreduce_or_v2i16(i16 signext %s, <2 x i16> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_or_v2i16: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e16, mf4, tu, mu |
| ; CHECK-NEXT: vredor.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i16 @llvm.vp.reduce.or.v2i16(i16 %s, <2 x i16> %v, <2 x i1> %m, i32 %evl) |
| ret i16 %r |
| } |
| |
| declare i16 @llvm.vp.reduce.xor.v2i16(i16, <2 x i16>, <2 x i1>, i32) |
| |
| define signext i16 @vpreduce_xor_v2i16(i16 signext %s, <2 x i16> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_xor_v2i16: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e16, mf4, tu, mu |
| ; CHECK-NEXT: vredxor.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i16 @llvm.vp.reduce.xor.v2i16(i16 %s, <2 x i16> %v, <2 x i1> %m, i32 %evl) |
| ret i16 %r |
| } |
| |
| declare i16 @llvm.vp.reduce.add.v4i16(i16, <4 x i16>, <4 x i1>, i32) |
| |
| define signext i16 @vpreduce_add_v4i16(i16 signext %s, <4 x i16> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_add_v4i16: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e16, mf2, tu, mu |
| ; CHECK-NEXT: vredsum.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i16 @llvm.vp.reduce.add.v4i16(i16 %s, <4 x i16> %v, <4 x i1> %m, i32 %evl) |
| ret i16 %r |
| } |
| |
| declare i16 @llvm.vp.reduce.umax.v4i16(i16, <4 x i16>, <4 x i1>, i32) |
| |
| define signext i16 @vpreduce_umax_v4i16(i16 signext %s, <4 x i16> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_umax_v4i16: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: lui a2, 16 |
| ; RV32-NEXT: addi a2, a2, -1 |
| ; RV32-NEXT: and a0, a0, a2 |
| ; RV32-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; RV32-NEXT: vmv.v.x v9, a0 |
| ; RV32-NEXT: vsetvli zero, a1, e16, mf2, tu, mu |
| ; RV32-NEXT: vredmaxu.vs v9, v8, v9, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v9 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_umax_v4i16: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: lui a2, 16 |
| ; RV64-NEXT: addiw a2, a2, -1 |
| ; RV64-NEXT: and a0, a0, a2 |
| ; RV64-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v9, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e16, mf2, tu, mu |
| ; RV64-NEXT: vredmaxu.vs v9, v8, v9, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v9 |
| ; RV64-NEXT: ret |
| %r = call i16 @llvm.vp.reduce.umax.v4i16(i16 %s, <4 x i16> %v, <4 x i1> %m, i32 %evl) |
| ret i16 %r |
| } |
| |
| declare i16 @llvm.vp.reduce.smax.v4i16(i16, <4 x i16>, <4 x i1>, i32) |
| |
| define signext i16 @vpreduce_smax_v4i16(i16 signext %s, <4 x i16> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_smax_v4i16: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e16, mf2, tu, mu |
| ; CHECK-NEXT: vredmax.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i16 @llvm.vp.reduce.smax.v4i16(i16 %s, <4 x i16> %v, <4 x i1> %m, i32 %evl) |
| ret i16 %r |
| } |
| |
| declare i16 @llvm.vp.reduce.umin.v4i16(i16, <4 x i16>, <4 x i1>, i32) |
| |
| define signext i16 @vpreduce_umin_v4i16(i16 signext %s, <4 x i16> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_umin_v4i16: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: lui a2, 16 |
| ; RV32-NEXT: addi a2, a2, -1 |
| ; RV32-NEXT: and a0, a0, a2 |
| ; RV32-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; RV32-NEXT: vmv.v.x v9, a0 |
| ; RV32-NEXT: vsetvli zero, a1, e16, mf2, tu, mu |
| ; RV32-NEXT: vredminu.vs v9, v8, v9, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v9 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_umin_v4i16: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: lui a2, 16 |
| ; RV64-NEXT: addiw a2, a2, -1 |
| ; RV64-NEXT: and a0, a0, a2 |
| ; RV64-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v9, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e16, mf2, tu, mu |
| ; RV64-NEXT: vredminu.vs v9, v8, v9, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v9 |
| ; RV64-NEXT: ret |
| %r = call i16 @llvm.vp.reduce.umin.v4i16(i16 %s, <4 x i16> %v, <4 x i1> %m, i32 %evl) |
| ret i16 %r |
| } |
| |
| declare i16 @llvm.vp.reduce.smin.v4i16(i16, <4 x i16>, <4 x i1>, i32) |
| |
| define signext i16 @vpreduce_smin_v4i16(i16 signext %s, <4 x i16> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_smin_v4i16: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e16, mf2, tu, mu |
| ; CHECK-NEXT: vredmin.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i16 @llvm.vp.reduce.smin.v4i16(i16 %s, <4 x i16> %v, <4 x i1> %m, i32 %evl) |
| ret i16 %r |
| } |
| |
| declare i16 @llvm.vp.reduce.and.v4i16(i16, <4 x i16>, <4 x i1>, i32) |
| |
| define signext i16 @vpreduce_and_v4i16(i16 signext %s, <4 x i16> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_and_v4i16: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e16, mf2, tu, mu |
| ; CHECK-NEXT: vredand.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i16 @llvm.vp.reduce.and.v4i16(i16 %s, <4 x i16> %v, <4 x i1> %m, i32 %evl) |
| ret i16 %r |
| } |
| |
| declare i16 @llvm.vp.reduce.or.v4i16(i16, <4 x i16>, <4 x i1>, i32) |
| |
| define signext i16 @vpreduce_or_v4i16(i16 signext %s, <4 x i16> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_or_v4i16: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e16, mf2, tu, mu |
| ; CHECK-NEXT: vredor.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i16 @llvm.vp.reduce.or.v4i16(i16 %s, <4 x i16> %v, <4 x i1> %m, i32 %evl) |
| ret i16 %r |
| } |
| |
| declare i16 @llvm.vp.reduce.xor.v4i16(i16, <4 x i16>, <4 x i1>, i32) |
| |
| define signext i16 @vpreduce_xor_v4i16(i16 signext %s, <4 x i16> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_xor_v4i16: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e16, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e16, mf2, tu, mu |
| ; CHECK-NEXT: vredxor.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i16 @llvm.vp.reduce.xor.v4i16(i16 %s, <4 x i16> %v, <4 x i1> %m, i32 %evl) |
| ret i16 %r |
| } |
| |
| declare i32 @llvm.vp.reduce.add.v2i32(i32, <2 x i32>, <2 x i1>, i32) |
| |
| define signext i32 @vpreduce_add_v2i32(i32 signext %s, <2 x i32> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_add_v2i32: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e32, mf2, tu, mu |
| ; CHECK-NEXT: vredsum.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i32 @llvm.vp.reduce.add.v2i32(i32 %s, <2 x i32> %v, <2 x i1> %m, i32 %evl) |
| ret i32 %r |
| } |
| |
| declare i32 @llvm.vp.reduce.umax.v2i32(i32, <2 x i32>, <2 x i1>, i32) |
| |
| define signext i32 @vpreduce_umax_v2i32(i32 signext %s, <2 x i32> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_umax_v2i32: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; RV32-NEXT: vmv.v.x v9, a0 |
| ; RV32-NEXT: vsetvli zero, a1, e32, mf2, tu, mu |
| ; RV32-NEXT: vredmaxu.vs v9, v8, v9, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v9 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_umax_v2i32: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: slli a0, a0, 32 |
| ; RV64-NEXT: srli a0, a0, 32 |
| ; RV64-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v9, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e32, mf2, tu, mu |
| ; RV64-NEXT: vredmaxu.vs v9, v8, v9, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v9 |
| ; RV64-NEXT: ret |
| %r = call i32 @llvm.vp.reduce.umax.v2i32(i32 %s, <2 x i32> %v, <2 x i1> %m, i32 %evl) |
| ret i32 %r |
| } |
| |
| declare i32 @llvm.vp.reduce.smax.v2i32(i32, <2 x i32>, <2 x i1>, i32) |
| |
| define signext i32 @vpreduce_smax_v2i32(i32 signext %s, <2 x i32> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_smax_v2i32: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e32, mf2, tu, mu |
| ; CHECK-NEXT: vredmax.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i32 @llvm.vp.reduce.smax.v2i32(i32 %s, <2 x i32> %v, <2 x i1> %m, i32 %evl) |
| ret i32 %r |
| } |
| |
| declare i32 @llvm.vp.reduce.umin.v2i32(i32, <2 x i32>, <2 x i1>, i32) |
| |
| define signext i32 @vpreduce_umin_v2i32(i32 signext %s, <2 x i32> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_umin_v2i32: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; RV32-NEXT: vmv.v.x v9, a0 |
| ; RV32-NEXT: vsetvli zero, a1, e32, mf2, tu, mu |
| ; RV32-NEXT: vredminu.vs v9, v8, v9, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v9 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_umin_v2i32: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: slli a0, a0, 32 |
| ; RV64-NEXT: srli a0, a0, 32 |
| ; RV64-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v9, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e32, mf2, tu, mu |
| ; RV64-NEXT: vredminu.vs v9, v8, v9, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v9 |
| ; RV64-NEXT: ret |
| %r = call i32 @llvm.vp.reduce.umin.v2i32(i32 %s, <2 x i32> %v, <2 x i1> %m, i32 %evl) |
| ret i32 %r |
| } |
| |
| declare i32 @llvm.vp.reduce.smin.v2i32(i32, <2 x i32>, <2 x i1>, i32) |
| |
| define signext i32 @vpreduce_smin_v2i32(i32 signext %s, <2 x i32> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_smin_v2i32: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e32, mf2, tu, mu |
| ; CHECK-NEXT: vredmin.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i32 @llvm.vp.reduce.smin.v2i32(i32 %s, <2 x i32> %v, <2 x i1> %m, i32 %evl) |
| ret i32 %r |
| } |
| |
| declare i32 @llvm.vp.reduce.and.v2i32(i32, <2 x i32>, <2 x i1>, i32) |
| |
| define signext i32 @vpreduce_and_v2i32(i32 signext %s, <2 x i32> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_and_v2i32: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e32, mf2, tu, mu |
| ; CHECK-NEXT: vredand.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i32 @llvm.vp.reduce.and.v2i32(i32 %s, <2 x i32> %v, <2 x i1> %m, i32 %evl) |
| ret i32 %r |
| } |
| |
| declare i32 @llvm.vp.reduce.or.v2i32(i32, <2 x i32>, <2 x i1>, i32) |
| |
| define signext i32 @vpreduce_or_v2i32(i32 signext %s, <2 x i32> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_or_v2i32: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e32, mf2, tu, mu |
| ; CHECK-NEXT: vredor.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i32 @llvm.vp.reduce.or.v2i32(i32 %s, <2 x i32> %v, <2 x i1> %m, i32 %evl) |
| ret i32 %r |
| } |
| |
| declare i32 @llvm.vp.reduce.xor.v2i32(i32, <2 x i32>, <2 x i1>, i32) |
| |
| define signext i32 @vpreduce_xor_v2i32(i32 signext %s, <2 x i32> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_xor_v2i32: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e32, mf2, tu, mu |
| ; CHECK-NEXT: vredxor.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i32 @llvm.vp.reduce.xor.v2i32(i32 %s, <2 x i32> %v, <2 x i1> %m, i32 %evl) |
| ret i32 %r |
| } |
| |
| declare i32 @llvm.vp.reduce.add.v4i32(i32, <4 x i32>, <4 x i1>, i32) |
| |
| define signext i32 @vpreduce_add_v4i32(i32 signext %s, <4 x i32> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_add_v4i32: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e32, m1, tu, mu |
| ; CHECK-NEXT: vredsum.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i32 @llvm.vp.reduce.add.v4i32(i32 %s, <4 x i32> %v, <4 x i1> %m, i32 %evl) |
| ret i32 %r |
| } |
| |
| declare i32 @llvm.vp.reduce.umax.v4i32(i32, <4 x i32>, <4 x i1>, i32) |
| |
| define signext i32 @vpreduce_umax_v4i32(i32 signext %s, <4 x i32> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_umax_v4i32: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; RV32-NEXT: vmv.v.x v9, a0 |
| ; RV32-NEXT: vsetvli zero, a1, e32, m1, tu, mu |
| ; RV32-NEXT: vredmaxu.vs v9, v8, v9, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v9 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_umax_v4i32: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: slli a0, a0, 32 |
| ; RV64-NEXT: srli a0, a0, 32 |
| ; RV64-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v9, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e32, m1, tu, mu |
| ; RV64-NEXT: vredmaxu.vs v9, v8, v9, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v9 |
| ; RV64-NEXT: ret |
| %r = call i32 @llvm.vp.reduce.umax.v4i32(i32 %s, <4 x i32> %v, <4 x i1> %m, i32 %evl) |
| ret i32 %r |
| } |
| |
| declare i32 @llvm.vp.reduce.smax.v4i32(i32, <4 x i32>, <4 x i1>, i32) |
| |
| define signext i32 @vpreduce_smax_v4i32(i32 signext %s, <4 x i32> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_smax_v4i32: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e32, m1, tu, mu |
| ; CHECK-NEXT: vredmax.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i32 @llvm.vp.reduce.smax.v4i32(i32 %s, <4 x i32> %v, <4 x i1> %m, i32 %evl) |
| ret i32 %r |
| } |
| |
| declare i32 @llvm.vp.reduce.umin.v4i32(i32, <4 x i32>, <4 x i1>, i32) |
| |
| define signext i32 @vpreduce_umin_v4i32(i32 signext %s, <4 x i32> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_umin_v4i32: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; RV32-NEXT: vmv.v.x v9, a0 |
| ; RV32-NEXT: vsetvli zero, a1, e32, m1, tu, mu |
| ; RV32-NEXT: vredminu.vs v9, v8, v9, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v9 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_umin_v4i32: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: slli a0, a0, 32 |
| ; RV64-NEXT: srli a0, a0, 32 |
| ; RV64-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v9, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e32, m1, tu, mu |
| ; RV64-NEXT: vredminu.vs v9, v8, v9, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v9 |
| ; RV64-NEXT: ret |
| %r = call i32 @llvm.vp.reduce.umin.v4i32(i32 %s, <4 x i32> %v, <4 x i1> %m, i32 %evl) |
| ret i32 %r |
| } |
| |
| declare i32 @llvm.vp.reduce.smin.v4i32(i32, <4 x i32>, <4 x i1>, i32) |
| |
| define signext i32 @vpreduce_smin_v4i32(i32 signext %s, <4 x i32> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_smin_v4i32: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e32, m1, tu, mu |
| ; CHECK-NEXT: vredmin.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i32 @llvm.vp.reduce.smin.v4i32(i32 %s, <4 x i32> %v, <4 x i1> %m, i32 %evl) |
| ret i32 %r |
| } |
| |
| declare i32 @llvm.vp.reduce.and.v4i32(i32, <4 x i32>, <4 x i1>, i32) |
| |
| define signext i32 @vpreduce_and_v4i32(i32 signext %s, <4 x i32> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_and_v4i32: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e32, m1, tu, mu |
| ; CHECK-NEXT: vredand.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i32 @llvm.vp.reduce.and.v4i32(i32 %s, <4 x i32> %v, <4 x i1> %m, i32 %evl) |
| ret i32 %r |
| } |
| |
| declare i32 @llvm.vp.reduce.or.v4i32(i32, <4 x i32>, <4 x i1>, i32) |
| |
| define signext i32 @vpreduce_or_v4i32(i32 signext %s, <4 x i32> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_or_v4i32: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e32, m1, tu, mu |
| ; CHECK-NEXT: vredor.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i32 @llvm.vp.reduce.or.v4i32(i32 %s, <4 x i32> %v, <4 x i1> %m, i32 %evl) |
| ret i32 %r |
| } |
| |
| declare i32 @llvm.vp.reduce.xor.v4i32(i32, <4 x i32>, <4 x i1>, i32) |
| |
| define signext i32 @vpreduce_xor_v4i32(i32 signext %s, <4 x i32> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; CHECK-LABEL: vpreduce_xor_v4i32: |
| ; CHECK: # %bb.0: |
| ; CHECK-NEXT: vsetvli a2, zero, e32, m1, ta, mu |
| ; CHECK-NEXT: vmv.v.x v9, a0 |
| ; CHECK-NEXT: vsetvli zero, a1, e32, m1, tu, mu |
| ; CHECK-NEXT: vredxor.vs v9, v8, v9, v0.t |
| ; CHECK-NEXT: vmv.x.s a0, v9 |
| ; CHECK-NEXT: ret |
| %r = call i32 @llvm.vp.reduce.xor.v4i32(i32 %s, <4 x i32> %v, <4 x i1> %m, i32 %evl) |
| ret i32 %r |
| } |
| |
| declare i64 @llvm.vp.reduce.add.v2i64(i64, <2 x i64>, <2 x i1>, i32) |
| |
| define signext i64 @vpreduce_add_v2i64(i64 signext %s, <2 x i64> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_add_v2i64: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: addi sp, sp, -16 |
| ; RV32-NEXT: .cfi_def_cfa_offset 16 |
| ; RV32-NEXT: sw a1, 12(sp) |
| ; RV32-NEXT: sw a0, 8(sp) |
| ; RV32-NEXT: vsetvli a0, zero, e64, m1, ta, mu |
| ; RV32-NEXT: addi a0, sp, 8 |
| ; RV32-NEXT: vlse64.v v9, (a0), zero |
| ; RV32-NEXT: vsetvli zero, a2, e64, m1, tu, mu |
| ; RV32-NEXT: vredsum.vs v9, v8, v9, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v9 |
| ; RV32-NEXT: li a1, 32 |
| ; RV32-NEXT: vsetivli zero, 1, e64, m1, ta, mu |
| ; RV32-NEXT: vsrl.vx v8, v9, a1 |
| ; RV32-NEXT: vmv.x.s a1, v8 |
| ; RV32-NEXT: addi sp, sp, 16 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_add_v2i64: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: vsetvli a2, zero, e64, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v9, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e64, m1, tu, mu |
| ; RV64-NEXT: vredsum.vs v9, v8, v9, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v9 |
| ; RV64-NEXT: ret |
| %r = call i64 @llvm.vp.reduce.add.v2i64(i64 %s, <2 x i64> %v, <2 x i1> %m, i32 %evl) |
| ret i64 %r |
| } |
| |
| declare i64 @llvm.vp.reduce.umax.v2i64(i64, <2 x i64>, <2 x i1>, i32) |
| |
| define signext i64 @vpreduce_umax_v2i64(i64 signext %s, <2 x i64> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_umax_v2i64: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: addi sp, sp, -16 |
| ; RV32-NEXT: .cfi_def_cfa_offset 16 |
| ; RV32-NEXT: sw a1, 12(sp) |
| ; RV32-NEXT: sw a0, 8(sp) |
| ; RV32-NEXT: vsetvli a0, zero, e64, m1, ta, mu |
| ; RV32-NEXT: addi a0, sp, 8 |
| ; RV32-NEXT: vlse64.v v9, (a0), zero |
| ; RV32-NEXT: vsetvli zero, a2, e64, m1, tu, mu |
| ; RV32-NEXT: vredmaxu.vs v9, v8, v9, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v9 |
| ; RV32-NEXT: li a1, 32 |
| ; RV32-NEXT: vsetivli zero, 1, e64, m1, ta, mu |
| ; RV32-NEXT: vsrl.vx v8, v9, a1 |
| ; RV32-NEXT: vmv.x.s a1, v8 |
| ; RV32-NEXT: addi sp, sp, 16 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_umax_v2i64: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: vsetvli a2, zero, e64, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v9, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e64, m1, tu, mu |
| ; RV64-NEXT: vredmaxu.vs v9, v8, v9, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v9 |
| ; RV64-NEXT: ret |
| %r = call i64 @llvm.vp.reduce.umax.v2i64(i64 %s, <2 x i64> %v, <2 x i1> %m, i32 %evl) |
| ret i64 %r |
| } |
| |
| declare i64 @llvm.vp.reduce.smax.v2i64(i64, <2 x i64>, <2 x i1>, i32) |
| |
| define signext i64 @vpreduce_smax_v2i64(i64 signext %s, <2 x i64> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_smax_v2i64: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: addi sp, sp, -16 |
| ; RV32-NEXT: .cfi_def_cfa_offset 16 |
| ; RV32-NEXT: sw a1, 12(sp) |
| ; RV32-NEXT: sw a0, 8(sp) |
| ; RV32-NEXT: vsetvli a0, zero, e64, m1, ta, mu |
| ; RV32-NEXT: addi a0, sp, 8 |
| ; RV32-NEXT: vlse64.v v9, (a0), zero |
| ; RV32-NEXT: vsetvli zero, a2, e64, m1, tu, mu |
| ; RV32-NEXT: vredmax.vs v9, v8, v9, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v9 |
| ; RV32-NEXT: li a1, 32 |
| ; RV32-NEXT: vsetivli zero, 1, e64, m1, ta, mu |
| ; RV32-NEXT: vsrl.vx v8, v9, a1 |
| ; RV32-NEXT: vmv.x.s a1, v8 |
| ; RV32-NEXT: addi sp, sp, 16 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_smax_v2i64: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: vsetvli a2, zero, e64, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v9, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e64, m1, tu, mu |
| ; RV64-NEXT: vredmax.vs v9, v8, v9, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v9 |
| ; RV64-NEXT: ret |
| %r = call i64 @llvm.vp.reduce.smax.v2i64(i64 %s, <2 x i64> %v, <2 x i1> %m, i32 %evl) |
| ret i64 %r |
| } |
| |
| declare i64 @llvm.vp.reduce.umin.v2i64(i64, <2 x i64>, <2 x i1>, i32) |
| |
| define signext i64 @vpreduce_umin_v2i64(i64 signext %s, <2 x i64> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_umin_v2i64: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: addi sp, sp, -16 |
| ; RV32-NEXT: .cfi_def_cfa_offset 16 |
| ; RV32-NEXT: sw a1, 12(sp) |
| ; RV32-NEXT: sw a0, 8(sp) |
| ; RV32-NEXT: vsetvli a0, zero, e64, m1, ta, mu |
| ; RV32-NEXT: addi a0, sp, 8 |
| ; RV32-NEXT: vlse64.v v9, (a0), zero |
| ; RV32-NEXT: vsetvli zero, a2, e64, m1, tu, mu |
| ; RV32-NEXT: vredminu.vs v9, v8, v9, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v9 |
| ; RV32-NEXT: li a1, 32 |
| ; RV32-NEXT: vsetivli zero, 1, e64, m1, ta, mu |
| ; RV32-NEXT: vsrl.vx v8, v9, a1 |
| ; RV32-NEXT: vmv.x.s a1, v8 |
| ; RV32-NEXT: addi sp, sp, 16 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_umin_v2i64: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: vsetvli a2, zero, e64, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v9, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e64, m1, tu, mu |
| ; RV64-NEXT: vredminu.vs v9, v8, v9, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v9 |
| ; RV64-NEXT: ret |
| %r = call i64 @llvm.vp.reduce.umin.v2i64(i64 %s, <2 x i64> %v, <2 x i1> %m, i32 %evl) |
| ret i64 %r |
| } |
| |
| declare i64 @llvm.vp.reduce.smin.v2i64(i64, <2 x i64>, <2 x i1>, i32) |
| |
| define signext i64 @vpreduce_smin_v2i64(i64 signext %s, <2 x i64> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_smin_v2i64: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: addi sp, sp, -16 |
| ; RV32-NEXT: .cfi_def_cfa_offset 16 |
| ; RV32-NEXT: sw a1, 12(sp) |
| ; RV32-NEXT: sw a0, 8(sp) |
| ; RV32-NEXT: vsetvli a0, zero, e64, m1, ta, mu |
| ; RV32-NEXT: addi a0, sp, 8 |
| ; RV32-NEXT: vlse64.v v9, (a0), zero |
| ; RV32-NEXT: vsetvli zero, a2, e64, m1, tu, mu |
| ; RV32-NEXT: vredmin.vs v9, v8, v9, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v9 |
| ; RV32-NEXT: li a1, 32 |
| ; RV32-NEXT: vsetivli zero, 1, e64, m1, ta, mu |
| ; RV32-NEXT: vsrl.vx v8, v9, a1 |
| ; RV32-NEXT: vmv.x.s a1, v8 |
| ; RV32-NEXT: addi sp, sp, 16 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_smin_v2i64: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: vsetvli a2, zero, e64, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v9, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e64, m1, tu, mu |
| ; RV64-NEXT: vredmin.vs v9, v8, v9, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v9 |
| ; RV64-NEXT: ret |
| %r = call i64 @llvm.vp.reduce.smin.v2i64(i64 %s, <2 x i64> %v, <2 x i1> %m, i32 %evl) |
| ret i64 %r |
| } |
| |
| declare i64 @llvm.vp.reduce.and.v2i64(i64, <2 x i64>, <2 x i1>, i32) |
| |
| define signext i64 @vpreduce_and_v2i64(i64 signext %s, <2 x i64> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_and_v2i64: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: addi sp, sp, -16 |
| ; RV32-NEXT: .cfi_def_cfa_offset 16 |
| ; RV32-NEXT: sw a1, 12(sp) |
| ; RV32-NEXT: sw a0, 8(sp) |
| ; RV32-NEXT: vsetvli a0, zero, e64, m1, ta, mu |
| ; RV32-NEXT: addi a0, sp, 8 |
| ; RV32-NEXT: vlse64.v v9, (a0), zero |
| ; RV32-NEXT: vsetvli zero, a2, e64, m1, tu, mu |
| ; RV32-NEXT: vredand.vs v9, v8, v9, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v9 |
| ; RV32-NEXT: li a1, 32 |
| ; RV32-NEXT: vsetivli zero, 1, e64, m1, ta, mu |
| ; RV32-NEXT: vsrl.vx v8, v9, a1 |
| ; RV32-NEXT: vmv.x.s a1, v8 |
| ; RV32-NEXT: addi sp, sp, 16 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_and_v2i64: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: vsetvli a2, zero, e64, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v9, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e64, m1, tu, mu |
| ; RV64-NEXT: vredand.vs v9, v8, v9, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v9 |
| ; RV64-NEXT: ret |
| %r = call i64 @llvm.vp.reduce.and.v2i64(i64 %s, <2 x i64> %v, <2 x i1> %m, i32 %evl) |
| ret i64 %r |
| } |
| |
| declare i64 @llvm.vp.reduce.or.v2i64(i64, <2 x i64>, <2 x i1>, i32) |
| |
| define signext i64 @vpreduce_or_v2i64(i64 signext %s, <2 x i64> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_or_v2i64: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: addi sp, sp, -16 |
| ; RV32-NEXT: .cfi_def_cfa_offset 16 |
| ; RV32-NEXT: sw a1, 12(sp) |
| ; RV32-NEXT: sw a0, 8(sp) |
| ; RV32-NEXT: vsetvli a0, zero, e64, m1, ta, mu |
| ; RV32-NEXT: addi a0, sp, 8 |
| ; RV32-NEXT: vlse64.v v9, (a0), zero |
| ; RV32-NEXT: vsetvli zero, a2, e64, m1, tu, mu |
| ; RV32-NEXT: vredor.vs v9, v8, v9, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v9 |
| ; RV32-NEXT: li a1, 32 |
| ; RV32-NEXT: vsetivli zero, 1, e64, m1, ta, mu |
| ; RV32-NEXT: vsrl.vx v8, v9, a1 |
| ; RV32-NEXT: vmv.x.s a1, v8 |
| ; RV32-NEXT: addi sp, sp, 16 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_or_v2i64: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: vsetvli a2, zero, e64, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v9, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e64, m1, tu, mu |
| ; RV64-NEXT: vredor.vs v9, v8, v9, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v9 |
| ; RV64-NEXT: ret |
| %r = call i64 @llvm.vp.reduce.or.v2i64(i64 %s, <2 x i64> %v, <2 x i1> %m, i32 %evl) |
| ret i64 %r |
| } |
| |
| declare i64 @llvm.vp.reduce.xor.v2i64(i64, <2 x i64>, <2 x i1>, i32) |
| |
| define signext i64 @vpreduce_xor_v2i64(i64 signext %s, <2 x i64> %v, <2 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_xor_v2i64: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: addi sp, sp, -16 |
| ; RV32-NEXT: .cfi_def_cfa_offset 16 |
| ; RV32-NEXT: sw a1, 12(sp) |
| ; RV32-NEXT: sw a0, 8(sp) |
| ; RV32-NEXT: vsetvli a0, zero, e64, m1, ta, mu |
| ; RV32-NEXT: addi a0, sp, 8 |
| ; RV32-NEXT: vlse64.v v9, (a0), zero |
| ; RV32-NEXT: vsetvli zero, a2, e64, m1, tu, mu |
| ; RV32-NEXT: vredxor.vs v9, v8, v9, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v9 |
| ; RV32-NEXT: li a1, 32 |
| ; RV32-NEXT: vsetivli zero, 1, e64, m1, ta, mu |
| ; RV32-NEXT: vsrl.vx v8, v9, a1 |
| ; RV32-NEXT: vmv.x.s a1, v8 |
| ; RV32-NEXT: addi sp, sp, 16 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_xor_v2i64: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: vsetvli a2, zero, e64, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v9, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e64, m1, tu, mu |
| ; RV64-NEXT: vredxor.vs v9, v8, v9, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v9 |
| ; RV64-NEXT: ret |
| %r = call i64 @llvm.vp.reduce.xor.v2i64(i64 %s, <2 x i64> %v, <2 x i1> %m, i32 %evl) |
| ret i64 %r |
| } |
| |
| declare i64 @llvm.vp.reduce.add.v4i64(i64, <4 x i64>, <4 x i1>, i32) |
| |
| define signext i64 @vpreduce_add_v4i64(i64 signext %s, <4 x i64> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_add_v4i64: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: addi sp, sp, -16 |
| ; RV32-NEXT: .cfi_def_cfa_offset 16 |
| ; RV32-NEXT: sw a1, 12(sp) |
| ; RV32-NEXT: sw a0, 8(sp) |
| ; RV32-NEXT: vsetvli a0, zero, e64, m1, ta, mu |
| ; RV32-NEXT: addi a0, sp, 8 |
| ; RV32-NEXT: vlse64.v v10, (a0), zero |
| ; RV32-NEXT: vsetvli zero, a2, e64, m2, tu, mu |
| ; RV32-NEXT: vredsum.vs v10, v8, v10, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v10 |
| ; RV32-NEXT: li a1, 32 |
| ; RV32-NEXT: vsetivli zero, 1, e64, m1, ta, mu |
| ; RV32-NEXT: vsrl.vx v8, v10, a1 |
| ; RV32-NEXT: vmv.x.s a1, v8 |
| ; RV32-NEXT: addi sp, sp, 16 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_add_v4i64: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: vsetvli a2, zero, e64, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v10, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e64, m2, tu, mu |
| ; RV64-NEXT: vredsum.vs v10, v8, v10, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v10 |
| ; RV64-NEXT: ret |
| %r = call i64 @llvm.vp.reduce.add.v4i64(i64 %s, <4 x i64> %v, <4 x i1> %m, i32 %evl) |
| ret i64 %r |
| } |
| |
| declare i64 @llvm.vp.reduce.umax.v4i64(i64, <4 x i64>, <4 x i1>, i32) |
| |
| define signext i64 @vpreduce_umax_v4i64(i64 signext %s, <4 x i64> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_umax_v4i64: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: addi sp, sp, -16 |
| ; RV32-NEXT: .cfi_def_cfa_offset 16 |
| ; RV32-NEXT: sw a1, 12(sp) |
| ; RV32-NEXT: sw a0, 8(sp) |
| ; RV32-NEXT: vsetvli a0, zero, e64, m1, ta, mu |
| ; RV32-NEXT: addi a0, sp, 8 |
| ; RV32-NEXT: vlse64.v v10, (a0), zero |
| ; RV32-NEXT: vsetvli zero, a2, e64, m2, tu, mu |
| ; RV32-NEXT: vredmaxu.vs v10, v8, v10, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v10 |
| ; RV32-NEXT: li a1, 32 |
| ; RV32-NEXT: vsetivli zero, 1, e64, m1, ta, mu |
| ; RV32-NEXT: vsrl.vx v8, v10, a1 |
| ; RV32-NEXT: vmv.x.s a1, v8 |
| ; RV32-NEXT: addi sp, sp, 16 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_umax_v4i64: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: vsetvli a2, zero, e64, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v10, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e64, m2, tu, mu |
| ; RV64-NEXT: vredmaxu.vs v10, v8, v10, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v10 |
| ; RV64-NEXT: ret |
| %r = call i64 @llvm.vp.reduce.umax.v4i64(i64 %s, <4 x i64> %v, <4 x i1> %m, i32 %evl) |
| ret i64 %r |
| } |
| |
| declare i64 @llvm.vp.reduce.smax.v4i64(i64, <4 x i64>, <4 x i1>, i32) |
| |
| define signext i64 @vpreduce_smax_v4i64(i64 signext %s, <4 x i64> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_smax_v4i64: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: addi sp, sp, -16 |
| ; RV32-NEXT: .cfi_def_cfa_offset 16 |
| ; RV32-NEXT: sw a1, 12(sp) |
| ; RV32-NEXT: sw a0, 8(sp) |
| ; RV32-NEXT: vsetvli a0, zero, e64, m1, ta, mu |
| ; RV32-NEXT: addi a0, sp, 8 |
| ; RV32-NEXT: vlse64.v v10, (a0), zero |
| ; RV32-NEXT: vsetvli zero, a2, e64, m2, tu, mu |
| ; RV32-NEXT: vredmax.vs v10, v8, v10, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v10 |
| ; RV32-NEXT: li a1, 32 |
| ; RV32-NEXT: vsetivli zero, 1, e64, m1, ta, mu |
| ; RV32-NEXT: vsrl.vx v8, v10, a1 |
| ; RV32-NEXT: vmv.x.s a1, v8 |
| ; RV32-NEXT: addi sp, sp, 16 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_smax_v4i64: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: vsetvli a2, zero, e64, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v10, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e64, m2, tu, mu |
| ; RV64-NEXT: vredmax.vs v10, v8, v10, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v10 |
| ; RV64-NEXT: ret |
| %r = call i64 @llvm.vp.reduce.smax.v4i64(i64 %s, <4 x i64> %v, <4 x i1> %m, i32 %evl) |
| ret i64 %r |
| } |
| |
| declare i64 @llvm.vp.reduce.umin.v4i64(i64, <4 x i64>, <4 x i1>, i32) |
| |
| define signext i64 @vpreduce_umin_v4i64(i64 signext %s, <4 x i64> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_umin_v4i64: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: addi sp, sp, -16 |
| ; RV32-NEXT: .cfi_def_cfa_offset 16 |
| ; RV32-NEXT: sw a1, 12(sp) |
| ; RV32-NEXT: sw a0, 8(sp) |
| ; RV32-NEXT: vsetvli a0, zero, e64, m1, ta, mu |
| ; RV32-NEXT: addi a0, sp, 8 |
| ; RV32-NEXT: vlse64.v v10, (a0), zero |
| ; RV32-NEXT: vsetvli zero, a2, e64, m2, tu, mu |
| ; RV32-NEXT: vredminu.vs v10, v8, v10, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v10 |
| ; RV32-NEXT: li a1, 32 |
| ; RV32-NEXT: vsetivli zero, 1, e64, m1, ta, mu |
| ; RV32-NEXT: vsrl.vx v8, v10, a1 |
| ; RV32-NEXT: vmv.x.s a1, v8 |
| ; RV32-NEXT: addi sp, sp, 16 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_umin_v4i64: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: vsetvli a2, zero, e64, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v10, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e64, m2, tu, mu |
| ; RV64-NEXT: vredminu.vs v10, v8, v10, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v10 |
| ; RV64-NEXT: ret |
| %r = call i64 @llvm.vp.reduce.umin.v4i64(i64 %s, <4 x i64> %v, <4 x i1> %m, i32 %evl) |
| ret i64 %r |
| } |
| |
| declare i64 @llvm.vp.reduce.smin.v4i64(i64, <4 x i64>, <4 x i1>, i32) |
| |
| define signext i64 @vpreduce_smin_v4i64(i64 signext %s, <4 x i64> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_smin_v4i64: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: addi sp, sp, -16 |
| ; RV32-NEXT: .cfi_def_cfa_offset 16 |
| ; RV32-NEXT: sw a1, 12(sp) |
| ; RV32-NEXT: sw a0, 8(sp) |
| ; RV32-NEXT: vsetvli a0, zero, e64, m1, ta, mu |
| ; RV32-NEXT: addi a0, sp, 8 |
| ; RV32-NEXT: vlse64.v v10, (a0), zero |
| ; RV32-NEXT: vsetvli zero, a2, e64, m2, tu, mu |
| ; RV32-NEXT: vredmin.vs v10, v8, v10, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v10 |
| ; RV32-NEXT: li a1, 32 |
| ; RV32-NEXT: vsetivli zero, 1, e64, m1, ta, mu |
| ; RV32-NEXT: vsrl.vx v8, v10, a1 |
| ; RV32-NEXT: vmv.x.s a1, v8 |
| ; RV32-NEXT: addi sp, sp, 16 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_smin_v4i64: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: vsetvli a2, zero, e64, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v10, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e64, m2, tu, mu |
| ; RV64-NEXT: vredmin.vs v10, v8, v10, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v10 |
| ; RV64-NEXT: ret |
| %r = call i64 @llvm.vp.reduce.smin.v4i64(i64 %s, <4 x i64> %v, <4 x i1> %m, i32 %evl) |
| ret i64 %r |
| } |
| |
| declare i64 @llvm.vp.reduce.and.v4i64(i64, <4 x i64>, <4 x i1>, i32) |
| |
| define signext i64 @vpreduce_and_v4i64(i64 signext %s, <4 x i64> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_and_v4i64: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: addi sp, sp, -16 |
| ; RV32-NEXT: .cfi_def_cfa_offset 16 |
| ; RV32-NEXT: sw a1, 12(sp) |
| ; RV32-NEXT: sw a0, 8(sp) |
| ; RV32-NEXT: vsetvli a0, zero, e64, m1, ta, mu |
| ; RV32-NEXT: addi a0, sp, 8 |
| ; RV32-NEXT: vlse64.v v10, (a0), zero |
| ; RV32-NEXT: vsetvli zero, a2, e64, m2, tu, mu |
| ; RV32-NEXT: vredand.vs v10, v8, v10, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v10 |
| ; RV32-NEXT: li a1, 32 |
| ; RV32-NEXT: vsetivli zero, 1, e64, m1, ta, mu |
| ; RV32-NEXT: vsrl.vx v8, v10, a1 |
| ; RV32-NEXT: vmv.x.s a1, v8 |
| ; RV32-NEXT: addi sp, sp, 16 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_and_v4i64: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: vsetvli a2, zero, e64, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v10, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e64, m2, tu, mu |
| ; RV64-NEXT: vredand.vs v10, v8, v10, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v10 |
| ; RV64-NEXT: ret |
| %r = call i64 @llvm.vp.reduce.and.v4i64(i64 %s, <4 x i64> %v, <4 x i1> %m, i32 %evl) |
| ret i64 %r |
| } |
| |
| declare i64 @llvm.vp.reduce.or.v4i64(i64, <4 x i64>, <4 x i1>, i32) |
| |
| define signext i64 @vpreduce_or_v4i64(i64 signext %s, <4 x i64> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_or_v4i64: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: addi sp, sp, -16 |
| ; RV32-NEXT: .cfi_def_cfa_offset 16 |
| ; RV32-NEXT: sw a1, 12(sp) |
| ; RV32-NEXT: sw a0, 8(sp) |
| ; RV32-NEXT: vsetvli a0, zero, e64, m1, ta, mu |
| ; RV32-NEXT: addi a0, sp, 8 |
| ; RV32-NEXT: vlse64.v v10, (a0), zero |
| ; RV32-NEXT: vsetvli zero, a2, e64, m2, tu, mu |
| ; RV32-NEXT: vredor.vs v10, v8, v10, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v10 |
| ; RV32-NEXT: li a1, 32 |
| ; RV32-NEXT: vsetivli zero, 1, e64, m1, ta, mu |
| ; RV32-NEXT: vsrl.vx v8, v10, a1 |
| ; RV32-NEXT: vmv.x.s a1, v8 |
| ; RV32-NEXT: addi sp, sp, 16 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_or_v4i64: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: vsetvli a2, zero, e64, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v10, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e64, m2, tu, mu |
| ; RV64-NEXT: vredor.vs v10, v8, v10, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v10 |
| ; RV64-NEXT: ret |
| %r = call i64 @llvm.vp.reduce.or.v4i64(i64 %s, <4 x i64> %v, <4 x i1> %m, i32 %evl) |
| ret i64 %r |
| } |
| |
| declare i64 @llvm.vp.reduce.xor.v4i64(i64, <4 x i64>, <4 x i1>, i32) |
| |
| define signext i64 @vpreduce_xor_v4i64(i64 signext %s, <4 x i64> %v, <4 x i1> %m, i32 zeroext %evl) { |
| ; RV32-LABEL: vpreduce_xor_v4i64: |
| ; RV32: # %bb.0: |
| ; RV32-NEXT: addi sp, sp, -16 |
| ; RV32-NEXT: .cfi_def_cfa_offset 16 |
| ; RV32-NEXT: sw a1, 12(sp) |
| ; RV32-NEXT: sw a0, 8(sp) |
| ; RV32-NEXT: vsetvli a0, zero, e64, m1, ta, mu |
| ; RV32-NEXT: addi a0, sp, 8 |
| ; RV32-NEXT: vlse64.v v10, (a0), zero |
| ; RV32-NEXT: vsetvli zero, a2, e64, m2, tu, mu |
| ; RV32-NEXT: vredxor.vs v10, v8, v10, v0.t |
| ; RV32-NEXT: vmv.x.s a0, v10 |
| ; RV32-NEXT: li a1, 32 |
| ; RV32-NEXT: vsetivli zero, 1, e64, m1, ta, mu |
| ; RV32-NEXT: vsrl.vx v8, v10, a1 |
| ; RV32-NEXT: vmv.x.s a1, v8 |
| ; RV32-NEXT: addi sp, sp, 16 |
| ; RV32-NEXT: ret |
| ; |
| ; RV64-LABEL: vpreduce_xor_v4i64: |
| ; RV64: # %bb.0: |
| ; RV64-NEXT: vsetvli a2, zero, e64, m1, ta, mu |
| ; RV64-NEXT: vmv.v.x v10, a0 |
| ; RV64-NEXT: vsetvli zero, a1, e64, m2, tu, mu |
| ; RV64-NEXT: vredxor.vs v10, v8, v10, v0.t |
| ; RV64-NEXT: vmv.x.s a0, v10 |
| ; RV64-NEXT: ret |
| %r = call i64 @llvm.vp.reduce.xor.v4i64(i64 %s, <4 x i64> %v, <4 x i1> %m, i32 %evl) |
| ret i64 %r |
| } |