; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
; RUN: llc -mtriple=r600-- -mcpu=cypress < %s | FileCheck -check-prefix=EG %s

;
; kernel void combine_vloads(global char8 addrspace(5)* src, global char8 addrspace(5)* result) {
;   for (int i = 0; i < 1024; ++i)
;     result[i] = src[0] + src[1] + src[2] + src[3];
; }
;


; 128-bit loads instead of many 8-bit
define amdgpu_kernel void @combine_vloads(ptr addrspace(1) nocapture %src, ptr addrspace(1) nocapture %result) nounwind {
; EG-LABEL: combine_vloads:
; EG:       ; %bb.0: ; %entry
; EG-NEXT:    ALU 3, @16, KC0[CB0:0-32], KC1[]
; EG-NEXT:    LOOP_START_DX10 @10
; EG-NEXT:    TEX 1 @12
; EG-NEXT:    ALU 86, @20, KC0[], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T14.XY, T15.X, 0
; EG-NEXT:    ALU_PUSH_BEFORE 4, @107, KC0[], KC1[]
; EG-NEXT:    JUMP @9 POP:1
; EG-NEXT:    LOOP_BREAK @9
; EG-NEXT:    POP @9 POP:1
; EG-NEXT:    END_LOOP @2
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    Fetch clause starting at 12:
; EG-NEXT:     VTX_READ_128 T14.XYZW, T13.X, 0, #1
; EG-NEXT:     VTX_READ_128 T15.XYZW, T13.X, 16, #1
; EG-NEXT:    ALU clause starting at 16:
; EG-NEXT:     MOV T13.X, KC0[2].Y,
; EG-NEXT:     MOV T0.W, KC0[2].Z,
; EG-NEXT:     MOV * T1.W, literal.x,
; EG-NEXT:    0(0.000000e+00), 0(0.000000e+00)
; EG-NEXT:    ALU clause starting at 20:
; EG-NEXT:     LSHR T2.W, T14.Y, literal.x,
; EG-NEXT:     LSHR * T3.W, T14.W, literal.x,
; EG-NEXT:    24(3.363116e-44), 0(0.000000e+00)
; EG-NEXT:     ADD_INT T2.W, PV.W, PS,
; EG-NEXT:     LSHR * T3.W, T15.Y, literal.x,
; EG-NEXT:    24(3.363116e-44), 0(0.000000e+00)
; EG-NEXT:     LSHR T0.Y, T14.Y, literal.x,
; EG-NEXT:     LSHR T0.Z, T14.W, literal.x,
; EG-NEXT:     ADD_INT T2.W, PV.W, PS,
; EG-NEXT:     LSHR * T3.W, T15.W, literal.y,
; EG-NEXT:    8(1.121039e-44), 24(3.363116e-44)
; EG-NEXT:     ADD_INT T16.X, PV.W, PS,
; EG-NEXT:     ADD_INT T0.Y, PV.Y, PV.Z,
; EG-NEXT:     LSHR T0.Z, T15.Y, literal.x,
; EG-NEXT:     LSHR T2.W, T14.X, literal.y,
; EG-NEXT:     LSHR * T3.W, T14.Z, literal.y,
; EG-NEXT:    8(1.121039e-44), 24(3.363116e-44)
; EG-NEXT:     ADD_INT T17.X, PV.W, PS,
; EG-NEXT:     ADD_INT T0.Y, PV.Y, PV.Z,
; EG-NEXT:     LSHR T0.Z, T15.W, literal.x,
; EG-NEXT:     LSHR T2.W, T14.Y, literal.y,
; EG-NEXT:     LSHR * T3.W, T14.W, literal.y,
; EG-NEXT:    8(1.121039e-44), 16(2.242078e-44)
; EG-NEXT:     LSHR T18.X, T15.X, literal.x,
; EG-NEXT:     LSHR T1.Y, T14.X, literal.y, BS:VEC_120/SCL_212
; EG-NEXT:     ADD_INT T1.Z, PV.W, PS,
; EG-NEXT:     LSHR T2.W, T15.Y, literal.z,
; EG-NEXT:     ADD_INT * T3.W, PV.Y, PV.Z,
; EG-NEXT:    24(3.363116e-44), 8(1.121039e-44)
; EG-NEXT:    16(2.242078e-44), 0(0.000000e+00)
; EG-NEXT:     LSHR T19.X, T14.Z, literal.x,
; EG-NEXT:     ADD_INT T0.Y, T14.Y, T14.W,
; EG-NEXT:     AND_INT T0.Z, PS, literal.y,
; EG-NEXT:     ADD_INT T2.W, PV.Z, PV.W,
; EG-NEXT:     LSHR * T3.W, T15.W, literal.z,
; EG-NEXT:    8(1.121039e-44), 255(3.573311e-43)
; EG-NEXT:    16(2.242078e-44), 0(0.000000e+00)
; EG-NEXT:     ADD_INT T20.X, PV.W, PS,
; EG-NEXT:     LSHL T2.Y, PV.Z, literal.x,
; EG-NEXT:     ADD_INT T0.Z, PV.Y, T15.Y,
; EG-NEXT:     ADD_INT T2.W, T1.Y, PV.X,
; EG-NEXT:     LSHR * T3.W, T15.X, literal.x,
; EG-NEXT:    8(1.121039e-44), 0(0.000000e+00)
; EG-NEXT:     ADD_INT T19.X, T14.X, T14.Z,
; EG-NEXT:     ADD_INT T0.Y, PV.W, PS,
; EG-NEXT:     LSHR T1.Z, T15.Z, literal.x,
; EG-NEXT:     LSHR T2.W, T14.X, literal.y,
; EG-NEXT:     LSHR * T3.W, T14.Z, literal.y,
; EG-NEXT:    8(1.121039e-44), 16(2.242078e-44)
; EG-NEXT:     ADD_INT T14.X, PV.W, PS,
; EG-NEXT:     LSHR T1.Y, T15.X, literal.x,
; EG-NEXT:     ADD_INT T1.Z, PV.Y, PV.Z,
; EG-NEXT:     ADD_INT T2.W, PV.X, T15.X,
; EG-NEXT:     ADD_INT * T3.W, T0.Z, T15.W,
; EG-NEXT:    16(2.242078e-44), 0(0.000000e+00)
; EG-NEXT:     AND_INT T15.X, PS, literal.x,
; EG-NEXT:     ADD_INT T0.Y, PV.W, T15.Z,
; EG-NEXT:     AND_INT T0.Z, PV.Z, literal.x,
; EG-NEXT:     ADD_INT T2.W, PV.X, PV.Y,
; EG-NEXT:     LSHR * T3.W, T15.Z, literal.y,
; EG-NEXT:    255(3.573311e-43), 16(2.242078e-44)
; EG-NEXT:     ADD_INT T14.X, PV.W, PS,
; EG-NEXT:     LSHL T1.Y, PV.Z, literal.x,
; EG-NEXT:     AND_INT T0.Z, PV.Y, literal.y,
; EG-NEXT:     OR_INT T2.W, PV.X, T2.Y,
; EG-NEXT:     LSHL * T3.W, T20.X, literal.z,
; EG-NEXT:    8(1.121039e-44), 255(3.573311e-43)
; EG-NEXT:    16(2.242078e-44), 0(0.000000e+00)
; EG-NEXT:     OR_INT T15.X, PV.W, PS,
; EG-NEXT:     OR_INT T0.Y, PV.Z, PV.Y,
; EG-NEXT:     LSHL T0.Z, PV.X, literal.x,
; EG-NEXT:     ADD_INT T2.W, T17.X, T18.X,
; EG-NEXT:     LSHR * T3.W, T15.Z, literal.y,
; EG-NEXT:    16(2.242078e-44), 24(3.363116e-44)
; EG-NEXT:     ADD_INT T1.Y, PV.W, PS,
; EG-NEXT:     OR_INT T0.Z, PV.Y, PV.Z,
; EG-NEXT:     AND_INT T2.W, PV.X, literal.x,
; EG-NEXT:     LSHL * T3.W, T16.X, literal.y,
; EG-NEXT:    16777215(2.350989e-38), 24(3.363116e-44)
; EG-NEXT:     OR_INT T14.Y, PV.W, PS,
; EG-NEXT:     AND_INT T2.W, PV.Z, literal.x,
; EG-NEXT:     LSHL * T3.W, PV.Y, literal.y,
; EG-NEXT:    16777215(2.350989e-38), 24(3.363116e-44)
; EG-NEXT:     OR_INT T14.X, PV.W, PS,
; EG-NEXT:     ADD_INT * T2.W, T0.W, T1.W,
; EG-NEXT:     LSHR * T15.X, PV.W, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
; EG-NEXT:    ALU clause starting at 107:
; EG-NEXT:     ADD_INT * T1.W, T1.W, literal.x,
; EG-NEXT:    8(1.121039e-44), 0(0.000000e+00)
; EG-NEXT:     SETE_INT * T2.W, PV.W, literal.x,
; EG-NEXT:    8192(1.147944e-41), 0(0.000000e+00)
; EG-NEXT:     PRED_SETNE_INT * ExecMask,PredicateBit (MASKED), PV.W, 0.0,
entry:
  br label %for.body

for.exit:                                         ; preds = %for.body
  ret void

for.body:                                         ; preds = %for.body, %entry
  %i.01 = phi i32 [ 0, %entry ], [ %tmp19, %for.body ]
  %vecload2 = load <8 x i32>, ptr addrspace(1) %src, align 32
  %0 = bitcast <8 x i32> %vecload2 to <32 x i8>
  %tmp5 = shufflevector <32 x i8> %0, <32 x i8> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
  %tmp8 = shufflevector <32 x i8> %0, <32 x i8> poison, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
  %tmp9 = add nsw <8 x i8> %tmp5, %tmp8
  %tmp12 = shufflevector <32 x i8> %0, <32 x i8> poison, <8 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
  %tmp13 = add nsw <8 x i8> %tmp9, %tmp12
  %tmp16 = shufflevector <32 x i8> %0, <32 x i8> poison, <8 x i32> <i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
  %tmp17 = add nsw <8 x i8> %tmp13, %tmp16
  %scevgep = getelementptr <8 x i8>, ptr addrspace(1) %result, i32 %i.01
  %1 = bitcast <8 x i8> %tmp17 to <2 x i32>
  store <2 x i32> %1, ptr addrspace(1) %scevgep, align 8
  %tmp19 = add nsw i32 %i.01, 1
  %exitcond = icmp eq i32 %tmp19, 1024
  br i1 %exitcond, label %for.exit, label %for.body
}