; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=amdgcn -mcpu=gfx1200 -mattr=+wavefrontsize64 -verify-machineinstrs < %s | FileCheck %s --check-prefix=GFX12

define amdgpu_ps void @test_wmma_f32_16x16x16_f16(<4 x half> %A, <4 x half> %B, <4 x float> %C, ptr addrspace(1) %out) {
; GFX12-LABEL: test_wmma_f32_16x16x16_f16:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_wmma_f32_16x16x16_f16 v[4:7], v[0:1], v[2:3], v[4:7]
; GFX12-NEXT:    global_store_b128 v[8:9], v[4:7], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x float>@llvm.amdgcn.wmma.f32.16x16x16.f16.v4f32.v4f16(<4 x half> %A, <4 x half> %B, <4 x float> %C)
  store <4 x float> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_wmma_f32_16x16x16_bf16(<4 x i16> %A, <4 x i16> %B, <4 x float> %C, ptr addrspace(1) %out) {
; GFX12-LABEL: test_wmma_f32_16x16x16_bf16:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_wmma_f32_16x16x16_bf16 v[4:7], v[0:1], v[2:3], v[4:7]
; GFX12-NEXT:    global_store_b128 v[8:9], v[4:7], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf16.v4f32.v4i16(<4 x i16> %A, <4 x i16> %B, <4 x float> %C)
  store <4 x float> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_wmma_f16_16x16x16_f16(<4 x half> %A, <4 x half> %B, <4 x half> %C, ptr addrspace(1) %out) {
; GFX12-LABEL: test_wmma_f16_16x16x16_f16:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_wmma_f16_16x16x16_f16 v[4:5], v[0:1], v[2:3], v[4:5]
; GFX12-NEXT:    global_store_b64 v[6:7], v[4:5], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x half> @llvm.amdgcn.wmma.f16.16x16x16.f16.v4f16.v4f16(<4 x half> %A, <4 x half> %B, <4 x half> %C, i1 0)
  store <4 x half> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_wmma_bf16_16x16x16_bf16(<4 x i16> %A, <4 x i16> %B, <4 x i16> %C, ptr addrspace(1) %out) {
; GFX12-LABEL: test_wmma_bf16_16x16x16_bf16:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_wmma_bf16_16x16x16_bf16 v[4:5], v[0:1], v[2:3], v[4:5]
; GFX12-NEXT:    global_store_b64 v[6:7], v[4:5], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x i16> @llvm.amdgcn.wmma.bf16.16x16x16.bf16.v4i16.v4i16(<4 x i16> %A, <4 x i16> %B, <4 x i16> %C, i1 0)
  store <4 x i16> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_wmma_i32_16x16x16_iu8(i32 %A, i32 %B, <4 x i32> %C, ptr addrspace(1) %out) {
; GFX12-LABEL: test_wmma_i32_16x16x16_iu8:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_wmma_i32_16x16x16_iu8 v[2:5], v0, v1, v[2:5]
; GFX12-NEXT:    global_store_b128 v[6:7], v[2:5], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x i32> @llvm.amdgcn.wmma.i32.16x16x16.iu8.v4i32.i32(i1 0, i32 %A, i1 0, i32 %B, <4 x i32> %C, i1 0)
  store <4 x i32> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_wmma_i32_16x16x16_iu4(i32 %A, i32 %B, <4 x i32> %C, ptr addrspace(1) %out) {
; GFX12-LABEL: test_wmma_i32_16x16x16_iu4:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_wmma_i32_16x16x16_iu4 v[2:5], v0, v1, v[2:5]
; GFX12-NEXT:    global_store_b128 v[6:7], v[2:5], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x i32> @llvm.amdgcn.wmma.i32.16x16x16.iu4.v4i32.i32(i1 0, i32 %A, i1 0, i32 %B, <4 x i32> %C, i1 0)
  store <4 x i32> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_wmma_f32_16x16x16_fp8_fp8(i32 %A, i32 %B, <4 x float> %C, ptr addrspace(1) %out) {
; GFX12-LABEL: test_wmma_f32_16x16x16_fp8_fp8:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_wmma_f32_16x16x16_fp8_fp8 v[2:5], v0, v1, v[2:5]
; GFX12-NEXT:    global_store_b128 v[6:7], v[2:5], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.fp8.fp8.v4f32.i32(i32 %A, i32 %B, <4 x float> %C)
  store <4 x float> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_wmma_f32_16x16x16_bf8_fp8(i32 %A, i32 %B, <4 x float> %C, ptr addrspace(1) %out) {
; GFX12-LABEL: test_wmma_f32_16x16x16_bf8_fp8:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_wmma_f32_16x16x16_bf8_fp8 v[2:5], v0, v1, v[2:5]
; GFX12-NEXT:    global_store_b128 v[6:7], v[2:5], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf8.fp8.v4f32.i32(i32 %A, i32 %B, <4 x float> %C)
  store <4 x float> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_wmma_f32_16x16x16_fp8_bf8(i32 %A, i32 %B, <4 x float> %C, ptr addrspace(1) %out) {
; GFX12-LABEL: test_wmma_f32_16x16x16_fp8_bf8:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_wmma_f32_16x16x16_fp8_bf8 v[2:5], v0, v1, v[2:5]
; GFX12-NEXT:    global_store_b128 v[6:7], v[2:5], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.fp8.bf8.v4f32.i32(i32 %A, i32 %B, <4 x float> %C)
  store <4 x float> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_wmma_f32_16x16x16_bf8_bf8(i32 %A, i32 %B, <4 x float> %C, ptr addrspace(1) %out) {
; GFX12-LABEL: test_wmma_f32_16x16x16_bf8_bf8:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_wmma_f32_16x16x16_bf8_bf8 v[2:5], v0, v1, v[2:5]
; GFX12-NEXT:    global_store_b128 v[6:7], v[2:5], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf8.bf8.v4f32.i32(i32 %A, i32 %B, <4 x float> %C)
  store <4 x float> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_wmma_i32_16x16x32_iu4(i32 %A, i32 %B, <4 x i32> %C, ptr addrspace(1) %out) {
; GFX12-LABEL: test_wmma_i32_16x16x32_iu4:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_wmma_i32_16x16x32_iu4 v[2:5], v0, v1, v[2:5]
; GFX12-NEXT:    global_store_b128 v[6:7], v[2:5], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x i32> @llvm.amdgcn.wmma.i32.16x16x32.iu4.v4i32.i32(i1 0, i32 %A, i1 0, i32 %B, <4 x i32> %C, i1 0)
  store <4 x i32> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_swmmac_f32_16x16x32_f16(<4 x half> %A, <8 x half> %B, <4 x float> %C, i8 %Index, ptr addrspace(1) %out) {
; GFX12-LABEL: test_swmmac_f32_16x16x32_f16:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_swmmac_f32_16x16x32_f16 v[6:9], v[0:1], v[2:5], v10
; GFX12-NEXT:    global_store_b128 v[11:12], v[6:9], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.f16.v4f32.v4f16.v8f16.i8(<4 x half> %A, <8 x half> %B, <4 x float> %C, i8 %Index)
  store <4 x float> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_swmmac_f32_16x16x32_bf16(<4 x i16> %A, <8 x i16> %B, <4 x float> %C, i8 %Index, ptr addrspace(1) %out) {
; GFX12-LABEL: test_swmmac_f32_16x16x32_bf16:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_swmmac_f32_16x16x32_bf16 v[6:9], v[0:1], v[2:5], v10
; GFX12-NEXT:    global_store_b128 v[11:12], v[6:9], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.bf16.v4f32.v4i16.v8i16.i8(<4 x i16> %A, <8 x i16> %B, <4 x float> %C, i8 %Index)
  store <4 x float> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_swmmac_f16_16x16x32_f16(<4 x half> %A, <8 x half> %B, <4 x half> %C, i8 %Index, ptr addrspace(1) %out) {
; GFX12-LABEL: test_swmmac_f16_16x16x32_f16:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_swmmac_f16_16x16x32_f16 v[6:7], v[0:1], v[2:5], v8
; GFX12-NEXT:    global_store_b64 v[9:10], v[6:7], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x half> @llvm.amdgcn.swmmac.f16.16x16x32.f16.v4f16.v4f16.v8f16.i8(<4 x half> %A, <8 x half> %B, <4 x half> %C, i8 %Index)
  store <4 x half> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_swmmac_bf16_16x16x32_bf16(<4 x i16> %A, <8 x i16> %B, <4 x i16> %C, i8 %Index, ptr addrspace(1) %out) {
; GFX12-LABEL: test_swmmac_bf16_16x16x32_bf16:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_swmmac_bf16_16x16x32_bf16 v[6:7], v[0:1], v[2:5], v8
; GFX12-NEXT:    global_store_b64 v[9:10], v[6:7], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x i16> @llvm.amdgcn.swmmac.bf16.16x16x32.bf16.v4i16.v4i16.v8i16.i8(<4 x i16> %A, <8 x i16> %B, <4 x i16> %C, i8 %Index)
  store <4 x i16> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_swmmac_i32_16x16x32_iu8(i32 %A, <2 x i32> %B, <4 x i32> %C, i8 %Index, ptr addrspace(1) %out) {
; GFX12-LABEL: test_swmmac_i32_16x16x32_iu8:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_swmmac_i32_16x16x32_iu8 v[3:6], v0, v[1:2], v7
; GFX12-NEXT:    global_store_b128 v[8:9], v[3:6], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x i32> @llvm.amdgcn.swmmac.i32.16x16x32.iu8.v4i32.i32.v2i32.i8(i1 0, i32 %A, i1 0, <2 x i32> %B, <4 x i32> %C, i8 %Index, i1 0)
  store <4 x i32> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_swmmac_i32_16x16x32_iu4(i32 %A, i32 %B, <4 x i32> %C, i16 %Index, ptr addrspace(1) %out) {
; GFX12-LABEL: test_swmmac_i32_16x16x32_iu4:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_swmmac_i32_16x16x32_iu4 v[2:5], v0, v1, v6
; GFX12-NEXT:    global_store_b128 v[7:8], v[2:5], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x i32> @llvm.amdgcn.swmmac.i32.16x16x32.iu4.v4i32.i32.i32.i16(i1 0, i32 %A, i1 0, i32 %B, <4 x i32> %C, i16 %Index, i1 0)
  store <4 x i32> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_swmmac_i32_16x16x64_iu4(i32 %A, <2 x i32> %B, <4 x i32> %C, i16 %Index, ptr addrspace(1) %out) {
; GFX12-LABEL: test_swmmac_i32_16x16x64_iu4:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_swmmac_i32_16x16x64_iu4 v[3:6], v0, v[1:2], v7
; GFX12-NEXT:    global_store_b128 v[8:9], v[3:6], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x i32> @llvm.amdgcn.swmmac.i32.16x16x64.iu4.v4i32.i32.v2i32.i16(i1 0, i32 %A, i1 0, <2 x i32> %B, <4 x i32> %C, i16 %Index, i1 0)
  store <4 x i32> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_swmmac_f32_16x16x32_fp8_fp8(i32 %A, <2 x i32> %B, <4 x float> %C, i8 %Index, ptr addrspace(1) %out) {
; GFX12-LABEL: test_swmmac_f32_16x16x32_fp8_fp8:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_swmmac_f32_16x16x32_fp8_fp8 v[3:6], v0, v[1:2], v7
; GFX12-NEXT:    global_store_b128 v[8:9], v[3:6], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.fp8.fp8.v4f32.i32.v2i32.i8(i32 %A, <2 x i32> %B, <4 x float> %C, i8 %Index)
  store <4 x float> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_swmmac_f32_16x16x32_fp8_bf8(i32 %A, <2 x i32> %B, <4 x float> %C, i8 %Index, ptr addrspace(1) %out) {
; GFX12-LABEL: test_swmmac_f32_16x16x32_fp8_bf8:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_swmmac_f32_16x16x32_fp8_bf8 v[3:6], v0, v[1:2], v7
; GFX12-NEXT:    global_store_b128 v[8:9], v[3:6], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.fp8.bf8.v4f32.i32.v2i32.i8(i32 %A, <2 x i32> %B, <4 x float> %C, i8 %Index)
  store <4 x float> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_swmmac_f32_16x16x32_bf8_fp8(i32 %A, <2 x i32> %B, <4 x float> %C, i8 %Index, ptr addrspace(1) %out) {
; GFX12-LABEL: test_swmmac_f32_16x16x32_bf8_fp8:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_swmmac_f32_16x16x32_bf8_fp8 v[3:6], v0, v[1:2], v7
; GFX12-NEXT:    global_store_b128 v[8:9], v[3:6], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.bf8.fp8.v4f32.i32.v2i32.i8(i32 %A, <2 x i32> %B, <4 x float> %C, i8 %Index)
  store <4 x float> %res, ptr addrspace(1) %out
  ret void
}

define amdgpu_ps void @test_swmmac_f32_16x16x32_bf8_bf8(i32 %A, <2 x i32> %B, <4 x float> %C, i8 %Index, ptr addrspace(1) %out) {
; GFX12-LABEL: test_swmmac_f32_16x16x32_bf8_bf8:
; GFX12:       ; %bb.0: ; %bb
; GFX12-NEXT:    v_swmmac_f32_16x16x32_bf8_bf8 v[3:6], v0, v[1:2], v7
; GFX12-NEXT:    global_store_b128 v[8:9], v[3:6], off
; GFX12-NEXT:    s_endpgm
bb:
  %res = call <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.bf8.bf8.v4f32.i32.v2i32.i8(i32 %A, <2 x i32> %B, <4 x float> %C, i8 %Index)
  store <4 x float> %res, ptr addrspace(1) %out
  ret void
}

declare <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.f16.v4f32.v4f16(<4 x half>, <4 x half>, <4 x float>)
declare <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf16.v4f32.v4i16(<4 x i16>, <4 x i16>, <4 x float>)
declare <4 x half> @llvm.amdgcn.wmma.f16.16x16x16.f16.v4f16.v4f16(<4 x half>, <4 x half>, <4 x half>, i1 immarg)
declare <4 x i16> @llvm.amdgcn.wmma.bf16.16x16x16.bf16.v4i16.v4i16(<4 x i16>, <4 x i16>, <4 x i16>, i1 immarg)
declare <4 x i32> @llvm.amdgcn.wmma.i32.16x16x16.iu8.v4i32.i32(i1 immarg, i32, i1 immarg, i32, <4 x i32>, i1 immarg)
declare <4 x i32> @llvm.amdgcn.wmma.i32.16x16x16.iu4.v4i32.i32(i1 immarg, i32, i1 immarg, i32, <4 x i32>, i1 immarg)
declare <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.fp8.fp8.v4f32.i32(i32, i32, <4 x float>)
declare <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.fp8.bf8.v4f32.i32(i32, i32, <4 x float>)
declare <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf8.fp8.v4f32.i32.(i32, i32, <4 x float>)
declare <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf8.bf8.v4f32.i32(i32, i32, <4 x float>)
declare <4 x i32> @llvm.amdgcn.wmma.i32.16x16x32.iu4.v4i32.i32(i1 immarg, i32, i1 immarg, i32, <4 x i32>, i1 immarg)
declare <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.f16.v4f32.v4f16.v8f16.i8(<4 x half>, <8 x half>, <4 x float>, i8)
declare <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.bf16.v4f32.v4i16.v8i16.i8(<4 x i16>, <8 x i16>, <4 x float>, i8)
declare <4 x half> @llvm.amdgcn.swmmac.f16.16x16x32.f16.v4f16.v4f16.v8f16.i8(<4 x half>, <8 x half>, <4 x half>, i8)
declare <4 x i16> @llvm.amdgcn.swmmac.bf16.16x16x32.bf16.v4i16.v4i16.v8i16.i8(<4 x i16>, <8 x i16>, <4 x i16>, i8)
declare <4 x i32> @llvm.amdgcn.swmmac.i32.16x16x32.iu8.v4i32.i32.v2i32.i8(i1 immarg, i32, i1 immarg, <2 x i32>, <4 x i32>, i8 %Index, i1 immarg)
declare <4 x i32> @llvm.amdgcn.swmmac.i32.16x16x32.iu4.v4i32.i32.i32.i16(i1 immarg, i32, i1 immarg, i32, <4 x i32>, i16 %Index, i1 immarg)
declare <4 x i32> @llvm.amdgcn.swmmac.i32.16x16x64.iu4.v4i32.i32.v2i32.i16(i1 immarg, i32, i1 immarg, <2 x i32>, <4 x i32>, i16 %Index, i1 immarg)
declare <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.fp8.fp8.v4f32.i32.v2i32.i8(i32, <2 x i32>, <4 x float>, i8)
declare <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.fp8.bf8.v4f32.i32.v2i32.i8(i32, <2 x i32>, <4 x float>, i8)
declare <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.bf8.fp8.v4f32.i32.v2i32.i8(i32, <2 x i32>, <4 x float>, i8)
declare <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.bf8.bf8.v4f32.i32.v2i32.i8(i32, <2 x i32>, <4 x float>, i8)