TSUBSC¶

Tile Operation Diagram¶

TSUBSC tile operation

Introduction¶

Elementwise fused op: src0 - scalar + src1.

Math Interpretation¶

For each element (i, j) in the valid region:

\[ \mathrm{dst}_{i,j} = \mathrm{src0}_{i,j} - \mathrm{scalar} + \mathrm{src1}_{i,j} \]

Assembly Syntax¶

PTO-AS form: see PTO-AS Specification.

Synchronous form:

%dst = tsubsc %src0, %scalar, %src1 : !pto.tile<...>, f32, !pto.tile<...>

IR Level 1 (SSA)¶

%dst = pto.tsubsc %src0, %scalar, %src1 : (!pto.tile<...>, dtype, !pto.tile<...>) -> !pto.tile<...>

IR Level 2 (DPS)¶

pto.tsubsc ins(%src0, %scalar, %src1 : !pto.tile_buf<...>, dtype, !pto.tile_buf<...>) outs(%dst : !pto.tile_buf<...>)

C++ Intrinsic¶

Declared in include/pto/common/pto_instr.hpp:

template <typename TileData, typename... WaitEvents>
PTO_INST RecordEvent TSUBSC(TileData& dst, TileData& src0, typename TileData::DType scalar, TileData& src1,
                            WaitEvents&... events);

Constraints¶

The op iterates over dst.GetValidRow() / dst.GetValidCol().

Examples¶

#include <pto/pto-inst.hpp>

using namespace pto;

void example() {
  using TileT = Tile<TileType::Vec, float, 16, 16>;
  TileT a, b, out;
  TSUBSC(out, a, 2.0f, b);
}

ASM Form Examples¶

Auto Mode¶

# Auto mode: compiler/runtime-managed placement and scheduling.
%dst = pto.tsubsc %src0, %scalar, %src1 : (!pto.tile<...>, dtype, !pto.tile<...>) -> !pto.tile<...>

Manual Mode¶

# Manual mode: bind resources explicitly before issuing the instruction.
# Optional for tile operands:
# pto.tassign %arg0, @tile(0x1000)
# pto.tassign %arg1, @tile(0x2000)
%dst = pto.tsubsc %src0, %scalar, %src1 : (!pto.tile<...>, dtype, !pto.tile<...>) -> !pto.tile<...>

PTO Assembly Form¶

%dst = tsubsc %src0, %scalar, %src1 : !pto.tile<...>, f32, !pto.tile<...>
# IR Level 2 (DPS)
pto.tsubsc ins(%src0, %scalar, %src1 : !pto.tile_buf<...>, dtype, !pto.tile_buf<...>) outs(%dst : !pto.tile_buf<...>)