; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=amdgcn -mcpu=gfx1200 -amdgpu-loop-prefetch < %s | FileCheck --check-prefix=GFX12 %s
; RUN: llc -mtriple=amdgcn -mcpu=gfx1200 -amdgpu-loop-prefetch -mattr=+safe-smem-prefetch < %s | FileCheck --check-prefix=GFX12-SPREFETCH %s

define amdgpu_kernel void @copy_flat(ptr nocapture %d, ptr nocapture readonly %s, i32 %n) {
; GFX12-LABEL: copy_flat:
; GFX12:       ; %bb.0: ; %entry
; GFX12-NEXT:    s_load_b32 s6, s[4:5], 0x34
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_cmp_eq_u32 s6, 0
; GFX12-NEXT:    s_cbranch_scc1 .LBB0_3
; GFX12-NEXT:  ; %bb.1: ; %for.body.preheader
; GFX12-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_add_nc_u64 s[2:3], s[2:3], 0xb0
; GFX12-NEXT:  .LBB0_2: ; %for.body
; GFX12-NEXT:    ; =>This Inner Loop Header: Depth=1
; GFX12-NEXT:    s_wait_alu 0xfffe
; GFX12-NEXT:    v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3
; GFX12-NEXT:    v_dual_mov_b32 v5, s1 :: v_dual_mov_b32 v4, s0
; GFX12-NEXT:    s_add_co_i32 s6, s6, -1
; GFX12-NEXT:    s_add_nc_u64 s[2:3], s[2:3], 16
; GFX12-NEXT:    flat_load_b128 v[0:3], v[0:1] offset:-176
; GFX12-NEXT:    s_cmp_lg_u32 s6, 0
; GFX12-NEXT:    s_add_nc_u64 s[0:1], s[0:1], 16
; GFX12-NEXT:    s_wait_loadcnt_dscnt 0x0
; GFX12-NEXT:    flat_store_b128 v[4:5], v[0:3]
; GFX12-NEXT:    s_cbranch_scc1 .LBB0_2
; GFX12-NEXT:  .LBB0_3: ; %for.end
; GFX12-NEXT:    s_endpgm
;
; GFX12-SPREFETCH-LABEL: copy_flat:
; GFX12-SPREFETCH:       ; %bb.0: ; %entry
; GFX12-SPREFETCH-NEXT:    s_load_b32 s6, s[4:5], 0x34
; GFX12-SPREFETCH-NEXT:    s_wait_kmcnt 0x0
; GFX12-SPREFETCH-NEXT:    s_cmp_eq_u32 s6, 0
; GFX12-SPREFETCH-NEXT:    s_cbranch_scc1 .LBB0_3
; GFX12-SPREFETCH-NEXT:  ; %bb.1: ; %for.body.preheader
; GFX12-SPREFETCH-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX12-SPREFETCH-NEXT:    s_wait_kmcnt 0x0
; GFX12-SPREFETCH-NEXT:    s_add_nc_u64 s[2:3], s[2:3], 0xb0
; GFX12-SPREFETCH-NEXT:  .LBB0_2: ; %for.body
; GFX12-SPREFETCH-NEXT:    ; =>This Inner Loop Header: Depth=1
; GFX12-SPREFETCH-NEXT:    s_wait_alu 0xfffe
; GFX12-SPREFETCH-NEXT:    v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3
; GFX12-SPREFETCH-NEXT:    s_prefetch_data s[2:3], 0x0, null, 0
; GFX12-SPREFETCH-NEXT:    v_dual_mov_b32 v5, s1 :: v_dual_mov_b32 v4, s0
; GFX12-SPREFETCH-NEXT:    s_add_co_i32 s6, s6, -1
; GFX12-SPREFETCH-NEXT:    flat_load_b128 v[0:3], v[0:1] offset:-176
; GFX12-SPREFETCH-NEXT:    s_add_nc_u64 s[2:3], s[2:3], 16
; GFX12-SPREFETCH-NEXT:    s_cmp_lg_u32 s6, 0
; GFX12-SPREFETCH-NEXT:    s_add_nc_u64 s[0:1], s[0:1], 16
; GFX12-SPREFETCH-NEXT:    s_wait_loadcnt_dscnt 0x0
; GFX12-SPREFETCH-NEXT:    flat_store_b128 v[4:5], v[0:3]
; GFX12-SPREFETCH-NEXT:    s_cbranch_scc1 .LBB0_2
; GFX12-SPREFETCH-NEXT:  .LBB0_3: ; %for.end
; GFX12-SPREFETCH-NEXT:    s_endpgm
entry:
  %cmp6.not = icmp eq i32 %n, 0
  br i1 %cmp6.not, label %for.end, label %for.body

for.body:                                         ; preds = %entry, %for.body
  %i.07 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
  %idxprom = zext i32 %i.07 to i64
  %arrayidx = getelementptr inbounds <4 x i32>, ptr %s, i64 %idxprom
  %ld = load <4 x i32>, ptr %arrayidx, align 4
  %arrayidx2 = getelementptr inbounds <4 x i32>, ptr %d, i64 %idxprom
  store <4 x i32> %ld, ptr %arrayidx2, align 4
  %inc = add nuw i32 %i.07, 1
  %exitcond.not = icmp eq i32 %inc, %n
  br i1 %exitcond.not, label %for.end, label %for.body

for.end:                                          ; preds = %for.body, %entry
  ret void
}

define amdgpu_kernel void @copy_global(ptr addrspace(1) nocapture %d, ptr addrspace(1) nocapture readonly %s, i32 %n) {
; GFX12-LABEL: copy_global:
; GFX12:       ; %bb.0: ; %entry
; GFX12-NEXT:    s_load_b32 s6, s[4:5], 0x34
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_cmp_eq_u32 s6, 0
; GFX12-NEXT:    s_cbranch_scc1 .LBB1_3
; GFX12-NEXT:  ; %bb.1: ; %for.body.preheader
; GFX12-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX12-NEXT:    v_mov_b32_e32 v0, 0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_add_nc_u64 s[2:3], s[2:3], 0xb0
; GFX12-NEXT:  .LBB1_2: ; %for.body
; GFX12-NEXT:    ; =>This Inner Loop Header: Depth=1
; GFX12-NEXT:    global_load_b128 v[1:4], v0, s[2:3] offset:-176
; GFX12-NEXT:    s_add_co_i32 s6, s6, -1
; GFX12-NEXT:    s_add_nc_u64 s[2:3], s[2:3], 16
; GFX12-NEXT:    s_cmp_lg_u32 s6, 0
; GFX12-NEXT:    s_wait_loadcnt 0x0
; GFX12-NEXT:    global_store_b128 v0, v[1:4], s[0:1]
; GFX12-NEXT:    s_add_nc_u64 s[0:1], s[0:1], 16
; GFX12-NEXT:    s_cbranch_scc1 .LBB1_2
; GFX12-NEXT:  .LBB1_3: ; %for.end
; GFX12-NEXT:    s_endpgm
;
; GFX12-SPREFETCH-LABEL: copy_global:
; GFX12-SPREFETCH:       ; %bb.0: ; %entry
; GFX12-SPREFETCH-NEXT:    s_load_b32 s6, s[4:5], 0x34
; GFX12-SPREFETCH-NEXT:    s_wait_kmcnt 0x0
; GFX12-SPREFETCH-NEXT:    s_cmp_eq_u32 s6, 0
; GFX12-SPREFETCH-NEXT:    s_cbranch_scc1 .LBB1_3
; GFX12-SPREFETCH-NEXT:  ; %bb.1: ; %for.body.preheader
; GFX12-SPREFETCH-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX12-SPREFETCH-NEXT:    v_mov_b32_e32 v0, 0
; GFX12-SPREFETCH-NEXT:    s_wait_kmcnt 0x0
; GFX12-SPREFETCH-NEXT:    s_add_nc_u64 s[2:3], s[2:3], 0xb0
; GFX12-SPREFETCH-NEXT:  .LBB1_2: ; %for.body
; GFX12-SPREFETCH-NEXT:    ; =>This Inner Loop Header: Depth=1
; GFX12-SPREFETCH-NEXT:    global_load_b128 v[1:4], v0, s[2:3] offset:-176
; GFX12-SPREFETCH-NEXT:    s_prefetch_data s[2:3], 0x0, null, 0
; GFX12-SPREFETCH-NEXT:    s_add_co_i32 s6, s6, -1
; GFX12-SPREFETCH-NEXT:    s_add_nc_u64 s[2:3], s[2:3], 16
; GFX12-SPREFETCH-NEXT:    s_cmp_lg_u32 s6, 0
; GFX12-SPREFETCH-NEXT:    s_wait_loadcnt 0x0
; GFX12-SPREFETCH-NEXT:    global_store_b128 v0, v[1:4], s[0:1]
; GFX12-SPREFETCH-NEXT:    s_add_nc_u64 s[0:1], s[0:1], 16
; GFX12-SPREFETCH-NEXT:    s_cbranch_scc1 .LBB1_2
; GFX12-SPREFETCH-NEXT:  .LBB1_3: ; %for.end
; GFX12-SPREFETCH-NEXT:    s_endpgm
entry:
  %cmp6.not = icmp eq i32 %n, 0
  br i1 %cmp6.not, label %for.end, label %for.body

for.body:                                         ; preds = %entry, %for.body
  %i.07 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
  %idxprom = zext i32 %i.07 to i64
  %arrayidx = getelementptr inbounds <4 x i32>, ptr addrspace(1) %s, i64 %idxprom
  %ld = load <4 x i32>, ptr addrspace(1) %arrayidx, align 4
  %arrayidx2 = getelementptr inbounds <4 x i32>, ptr addrspace(1) %d, i64 %idxprom
  store <4 x i32> %ld, ptr addrspace(1) %arrayidx2, align 4
  %inc = add nuw i32 %i.07, 1
  %exitcond.not = icmp eq i32 %inc, %n
  br i1 %exitcond.not, label %for.end, label %for.body

for.end:                                          ; preds = %for.body, %entry
  ret void
}

define amdgpu_kernel void @copy_constant(ptr addrspace(1) nocapture %d, ptr addrspace(4) nocapture readonly %s, i32 %n) {
; GFX12-LABEL: copy_constant:
; GFX12:       ; %bb.0: ; %entry
; GFX12-NEXT:    s_load_b32 s6, s[4:5], 0x34
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_cmp_eq_u32 s6, 0
; GFX12-NEXT:    s_cbranch_scc1 .LBB2_3
; GFX12-NEXT:  ; %bb.1: ; %for.body.preheader
; GFX12-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX12-NEXT:    v_mov_b32_e32 v0, 0
; GFX12-NEXT:  .LBB2_2: ; %for.body
; GFX12-NEXT:    ; =>This Inner Loop Header: Depth=1
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_load_b128 s[8:11], s[2:3], 0x0
; GFX12-NEXT:    s_add_co_i32 s6, s6, -1
; GFX12-NEXT:    s_add_nc_u64 s[2:3], s[2:3], 16
; GFX12-NEXT:    s_cmp_lg_u32 s6, 0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    v_dual_mov_b32 v1, s8 :: v_dual_mov_b32 v2, s9
; GFX12-NEXT:    v_dual_mov_b32 v3, s10 :: v_dual_mov_b32 v4, s11
; GFX12-NEXT:    global_store_b128 v0, v[1:4], s[0:1]
; GFX12-NEXT:    s_add_nc_u64 s[0:1], s[0:1], 16
; GFX12-NEXT:    s_cbranch_scc1 .LBB2_2
; GFX12-NEXT:  .LBB2_3: ; %for.end
; GFX12-NEXT:    s_endpgm
;
; GFX12-SPREFETCH-LABEL: copy_constant:
; GFX12-SPREFETCH:       ; %bb.0: ; %entry
; GFX12-SPREFETCH-NEXT:    s_load_b32 s6, s[4:5], 0x34
; GFX12-SPREFETCH-NEXT:    s_wait_kmcnt 0x0
; GFX12-SPREFETCH-NEXT:    s_cmp_eq_u32 s6, 0
; GFX12-SPREFETCH-NEXT:    s_cbranch_scc1 .LBB2_3
; GFX12-SPREFETCH-NEXT:  ; %bb.1: ; %for.body.preheader
; GFX12-SPREFETCH-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX12-SPREFETCH-NEXT:    v_mov_b32_e32 v0, 0
; GFX12-SPREFETCH-NEXT:  .LBB2_2: ; %for.body
; GFX12-SPREFETCH-NEXT:    ; =>This Inner Loop Header: Depth=1
; GFX12-SPREFETCH-NEXT:    s_wait_kmcnt 0x0
; GFX12-SPREFETCH-NEXT:    s_load_b128 s[8:11], s[2:3], 0x0
; GFX12-SPREFETCH-NEXT:    s_prefetch_data s[2:3], 0xb0, null, 0
; GFX12-SPREFETCH-NEXT:    s_add_co_i32 s6, s6, -1
; GFX12-SPREFETCH-NEXT:    s_add_nc_u64 s[2:3], s[2:3], 16
; GFX12-SPREFETCH-NEXT:    s_cmp_lg_u32 s6, 0
; GFX12-SPREFETCH-NEXT:    s_wait_kmcnt 0x0
; GFX12-SPREFETCH-NEXT:    v_dual_mov_b32 v1, s8 :: v_dual_mov_b32 v2, s9
; GFX12-SPREFETCH-NEXT:    v_dual_mov_b32 v3, s10 :: v_dual_mov_b32 v4, s11
; GFX12-SPREFETCH-NEXT:    global_store_b128 v0, v[1:4], s[0:1]
; GFX12-SPREFETCH-NEXT:    s_add_nc_u64 s[0:1], s[0:1], 16
; GFX12-SPREFETCH-NEXT:    s_cbranch_scc1 .LBB2_2
; GFX12-SPREFETCH-NEXT:  .LBB2_3: ; %for.end
; GFX12-SPREFETCH-NEXT:    s_endpgm
entry:
  %cmp6.not = icmp eq i32 %n, 0
  br i1 %cmp6.not, label %for.end, label %for.body

for.body:                                         ; preds = %entry, %for.body
  %i.07 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
  %idxprom = zext i32 %i.07 to i64
  %arrayidx = getelementptr inbounds <4 x i32>, ptr addrspace(4) %s, i64 %idxprom
  %ld = load <4 x i32>, ptr addrspace(4) %arrayidx, align 4
  %arrayidx2 = getelementptr inbounds <4 x i32>, ptr addrspace(1) %d, i64 %idxprom
  store <4 x i32> %ld, ptr addrspace(1) %arrayidx2, align 4
  %inc = add nuw i32 %i.07, 1
  %exitcond.not = icmp eq i32 %inc, %n
  br i1 %exitcond.not, label %for.end, label %for.body

for.end:                                          ; preds = %for.body, %entry
  ret void
}

define amdgpu_kernel void @copy_local(ptr addrspace(3) nocapture %d, ptr addrspace(3) nocapture readonly %s, i32 %n) {
; GFX12-LABEL: copy_local:
; GFX12:       ; %bb.0: ; %entry
; GFX12-NEXT:    s_load_b96 s[0:2], s[4:5], 0x24
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_cmp_eq_u32 s2, 0
; GFX12-NEXT:    s_cbranch_scc1 .LBB3_2
; GFX12-NEXT:  .LBB3_1: ; %for.body
; GFX12-NEXT:    ; =>This Inner Loop Header: Depth=1
; GFX12-NEXT:    s_wait_alu 0xfffe
; GFX12-NEXT:    v_mov_b32_e32 v2, s1
; GFX12-NEXT:    v_mov_b32_e32 v4, s0
; GFX12-NEXT:    s_add_co_i32 s2, s2, -1
; GFX12-NEXT:    s_add_co_i32 s0, s0, 16
; GFX12-NEXT:    s_add_co_i32 s1, s1, 16
; GFX12-NEXT:    ds_load_2addr_b32 v[0:1], v2 offset0:2 offset1:3
; GFX12-NEXT:    ds_load_2addr_b32 v[2:3], v2 offset1:1
; GFX12-NEXT:    s_cmp_lg_u32 s2, 0
; GFX12-NEXT:    s_wait_dscnt 0x1
; GFX12-NEXT:    ds_store_2addr_b32 v4, v0, v1 offset0:2 offset1:3
; GFX12-NEXT:    s_wait_dscnt 0x1
; GFX12-NEXT:    ds_store_2addr_b32 v4, v2, v3 offset1:1
; GFX12-NEXT:    s_cbranch_scc1 .LBB3_1
; GFX12-NEXT:  .LBB3_2: ; %for.end
; GFX12-NEXT:    s_endpgm
;
; GFX12-SPREFETCH-LABEL: copy_local:
; GFX12-SPREFETCH:       ; %bb.0: ; %entry
; GFX12-SPREFETCH-NEXT:    s_load_b96 s[0:2], s[4:5], 0x24
; GFX12-SPREFETCH-NEXT:    s_wait_kmcnt 0x0
; GFX12-SPREFETCH-NEXT:    s_cmp_eq_u32 s2, 0
; GFX12-SPREFETCH-NEXT:    s_cbranch_scc1 .LBB3_2
; GFX12-SPREFETCH-NEXT:  .LBB3_1: ; %for.body
; GFX12-SPREFETCH-NEXT:    ; =>This Inner Loop Header: Depth=1
; GFX12-SPREFETCH-NEXT:    s_wait_alu 0xfffe
; GFX12-SPREFETCH-NEXT:    v_mov_b32_e32 v2, s1
; GFX12-SPREFETCH-NEXT:    v_mov_b32_e32 v4, s0
; GFX12-SPREFETCH-NEXT:    s_add_co_i32 s2, s2, -1
; GFX12-SPREFETCH-NEXT:    s_add_co_i32 s0, s0, 16
; GFX12-SPREFETCH-NEXT:    s_add_co_i32 s1, s1, 16
; GFX12-SPREFETCH-NEXT:    ds_load_2addr_b32 v[0:1], v2 offset0:2 offset1:3
; GFX12-SPREFETCH-NEXT:    ds_load_2addr_b32 v[2:3], v2 offset1:1
; GFX12-SPREFETCH-NEXT:    s_cmp_lg_u32 s2, 0
; GFX12-SPREFETCH-NEXT:    s_wait_dscnt 0x1
; GFX12-SPREFETCH-NEXT:    ds_store_2addr_b32 v4, v0, v1 offset0:2 offset1:3
; GFX12-SPREFETCH-NEXT:    s_wait_dscnt 0x1
; GFX12-SPREFETCH-NEXT:    ds_store_2addr_b32 v4, v2, v3 offset1:1
; GFX12-SPREFETCH-NEXT:    s_cbranch_scc1 .LBB3_1
; GFX12-SPREFETCH-NEXT:  .LBB3_2: ; %for.end
; GFX12-SPREFETCH-NEXT:    s_endpgm
entry:
  %cmp6.not = icmp eq i32 %n, 0
  br i1 %cmp6.not, label %for.end, label %for.body

for.body:                                         ; preds = %entry, %for.body
  %i.07 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
  %idxprom = zext i32 %i.07 to i64
  %arrayidx = getelementptr inbounds <4 x i32>, ptr addrspace(3) %s, i64 %idxprom
  %ld = load <4 x i32>, ptr addrspace(3) %arrayidx, align 4
  %arrayidx2 = getelementptr inbounds <4 x i32>, ptr addrspace(3) %d, i64 %idxprom
  store <4 x i32> %ld, ptr addrspace(3) %arrayidx2, align 4
  %inc = add nuw i32 %i.07, 1
  %exitcond.not = icmp eq i32 %inc, %n
  br i1 %exitcond.not, label %for.end, label %for.body

for.end:                                          ; preds = %for.body, %entry
  ret void
}