log.mlir

// -----// IR Dump After CSE (cse) //----- //
func.func @__builtin_fill_i64(%arg0: !util.buffer, %arg1: !util.buffer, %arg2: !util.list<!util.buffer>, %arg3: i32, %arg4: i32, %arg5: i32, %arg6: i32, %arg7: i32, %arg8: i32, %arg9: i32, %arg10: i32, %arg11: i32) attributes {translation_info = #iree_codegen.translation_info<pipeline = VMVXDefault>} {
  %c12 = arith.constant 12 : index
  %c8 = arith.constant 8 : index
  %c4 = arith.constant 4 : index
  %c0 = arith.constant 0 : index
  %c32_i64 = arith.constant 32 : i64
  %c1 = arith.constant 1 : index
  %buffer_size = util.buffer.size %arg1 : !util.buffer
  %0 = util.buffer.load %arg1[%c0 for %c4] : !util.buffer{%buffer_size} -> i32
  %1 = util.buffer.load %arg1[%c4 for %c4] : !util.buffer{%buffer_size} -> i32
  %2 = util.buffer.load %arg1[%c8 for %c4] : !util.buffer{%buffer_size} -> i32
  %3 = util.buffer.load %arg1[%c12 for %c4] : !util.buffer{%buffer_size} -> i32
  %4 = arith.extui %0 : i32 to i64
  %5 = arith.extui %1 : i32 to i64
  %6 = arith.shli %5, %c32_i64 : i64
  %7 = arith.ori %4, %6 : i64
  %8 = arith.index_castui %7 : i64 to index
  %9 = arith.extui %2 : i32 to i64
  %10 = arith.extui %3 : i32 to i64
  %11 = arith.shli %10, %c32_i64 : i64
  %12 = arith.ori %9, %11 : i64
  %13 = util.list.get %arg2[%c0] : !util.list<!util.buffer>
  %14 = arith.index_cast %arg3 : i32 to index
  %15 = arith.index_cast %arg9 : i32 to index
  %16 = affine.apply affine_map<()[s0] -> (s0 * 64)>()[%14]
  %17 = affine.apply affine_map<()[s0] -> (s0 * 64)>()[%15]
  scf.for %arg12 = %16 to %c1 step %17 {
    %18 = affine.apply affine_map<()[s0, s1] -> (s0 + s1 floordiv 8)>()[%arg12, %8]
    %buffer_size_0 = util.buffer.size %13 : !util.buffer
    %19 = arith.muli %18, %c8 : index
    util.buffer.store %12, %13[%19 for %c8] : i64 -> !util.buffer{%buffer_size_0}
  }
  return
}

// -----// IR Dump After LowerAffinePass (lower-affine) //----- //
func.func @__builtin_fill_i64(%arg0: !util.buffer, %arg1: !util.buffer, %arg2: !util.list<!util.buffer>, %arg3: i32, %arg4: i32, %arg5: i32, %arg6: i32, %arg7: i32, %arg8: i32, %arg9: i32, %arg10: i32, %arg11: i32) attributes {translation_info = #iree_codegen.translation_info<pipeline = VMVXDefault>} {
  %c12 = arith.constant 12 : index
  %c8 = arith.constant 8 : index
  %c4 = arith.constant 4 : index
  %c0 = arith.constant 0 : index
  %c32_i64 = arith.constant 32 : i64
  %c1 = arith.constant 1 : index
  %buffer_size = util.buffer.size %arg1 : !util.buffer
  %0 = util.buffer.load %arg1[%c0 for %c4] : !util.buffer{%buffer_size} -> i32
  %1 = util.buffer.load %arg1[%c4 for %c4] : !util.buffer{%buffer_size} -> i32
  %2 = util.buffer.load %arg1[%c8 for %c4] : !util.buffer{%buffer_size} -> i32
  %3 = util.buffer.load %arg1[%c12 for %c4] : !util.buffer{%buffer_size} -> i32
  %4 = arith.extui %0 : i32 to i64
  %5 = arith.extui %1 : i32 to i64
  %6 = arith.shli %5, %c32_i64 : i64
  %7 = arith.ori %4, %6 : i64
  %8 = arith.index_castui %7 : i64 to index
  %9 = arith.extui %2 : i32 to i64
  %10 = arith.extui %3 : i32 to i64
  %11 = arith.shli %10, %c32_i64 : i64
  %12 = arith.ori %9, %11 : i64
  %13 = util.list.get %arg2[%c0] : !util.list<!util.buffer>
  %14 = arith.index_cast %arg3 : i32 to index
  %15 = arith.index_cast %arg9 : i32 to index
  %c64 = arith.constant 64 : index
  %16 = arith.muli %14, %c64 overflow<nsw> : index
  %c64_0 = arith.constant 64 : index
  %17 = arith.muli %15, %c64_0 overflow<nsw> : index
  scf.for %arg12 = %16 to %c1 step %17 {
    %c8_1 = arith.constant 8 : index
    %c0_2 = arith.constant 0 : index
    %c-1 = arith.constant -1 : index
    %18 = arith.cmpi slt, %8, %c0_2 : index
    %19 = arith.subi %c-1, %8 : index
    %20 = arith.select %18, %19, %8 : index
    %21 = arith.divsi %20, %c8_1 : index
    %22 = arith.subi %c-1, %21 : index
    %23 = arith.select %18, %22, %21 : index
    %24 = arith.addi %arg12, %23 : index
    %buffer_size_3 = util.buffer.size %13 : !util.buffer
    %25 = arith.muli %24, %c8 : index
    util.buffer.store %12, %13[%25 for %c8] : i64 -> !util.buffer{%buffer_size_3}
  }
  return
}