TCOLSUM¶

Tile Operation Diagram¶

Introduction¶

Reduce each column by summing across rows.

Math Interpretation¶

Let R = src.GetValidRow() and C = src.GetValidCol(). For 0 <= j < C:

isBinary selects the implementation path (binary-tree accumulation vs. sequential accumulation).

Assembly Syntax¶

Synchronous form:

%dst = tcolsum %src {isBinary = false} : !pto.tile<...> -> !pto.tile<...>

Lowering may introduce internal scratch tiles; the C++ intrinsic requires an explicit tmp operand.

AS Level 1 (SSA)¶

%dst = pto.tcolsum %src : !pto.tile<...> -> !pto.tile<...>
%dst = pto.tcolsum %src, %tmp {isBinary = false} : (!pto.tile<...>, !pto.tile<...>) -> !pto.tile<...>

AS Level 2 (DPS)¶

pto.tcolsum ins(%src : !pto.tile_buf<...>) outs(%dst : !pto.tile_buf<...>)
pto.tcolsum ins(%src, %tmp {isBinary = false} : !pto.tile_buf<...>, !pto.tile_buf<...>) outs(%dst : !pto.tile_buf<...>)

C++ Intrinsic¶

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

template <typename TileDataOut, typename TileDataIn, typename... WaitEvents>
PTO_INST RecordEvent TCOLSUM(TileDataOut &dst, TileDataIn &src, WaitEvents &... events);

template <typename TileDataOut, typename TileDataIn, typename TileDataTmp, typename... WaitEvents>
PTO_INST RecordEvent TCOLSUM(TileDataOut &dst, TileDataIn &src, TileDataTmp &tmp, bool isBinary, WaitEvents &... events);

Constraints¶

General constraints / checks¶

• dst and src must be TileType::Vec.
• dst and src must use standard ND layout: row-major and non-fractal (BLayout::RowMajor, SLayout::NoneBox).
• dst and src must use the same element type.
• Runtime checks:
  • src.GetValidCol() == dst.GetValidCol()
  • src.GetValidRow() != 0 (implementation silently returns early when zero, no computation performed)
  • src.GetValidCol() != 0 (implementation silently returns early when zero, no computation performed)
  • src.GetValidCol() <= tmp row stride measured in src elements
• isBinary selects the checked backend path:
  • true: binary-tree accumulation using tmp
  • false: sequential accumulation into dst

A2A3 implementation checks¶

• Supported element types: half, float, int16_t, int32_t.
• tmp must be TileType::Vec and use standard ND layout: row-major and non-fractal (BLayout::RowMajor, SLayout::NoneBox).
• tmp must use the same element type as src and dst.
• If src.GetValidRow() == 0 or src.GetValidCol() == 0, the implementation returns early.

A5 implementation checks¶

• Shared A5 column-reduce checks allow half, float, int8_t, uint8_t, int16_t, uint16_t, int32_t, uint32_t, bfloat16_t.
• The checked A5 TCOLSUM path still takes tmp only for the binary accumulation path; no extra compile-time tmp type/layout assertions are explicitly enforced in TCOLSUM_IMPL.

Temporary Space¶

Without tmp (2-argument overload: TCOLSUM(dst, src))¶

No tmp is required. Both A2A3 and A5 use sequential accumulation directly into dst.

With tmp and isBinary (4-argument overload: TCOLSUM(dst, src, tmp, isBinary))¶

A2A3¶

• When isBinary = true: tmp is used for binary-tree accumulation. Adjacent row pairs from src are summed into tmp, then tmp is recursively halved until a single row remains.
• tmp must have the same element type as src/dst.
• tmp must be TileType::Vec, row-major, non-fractal.
• tmp.GetValidCol() >= src.GetValidCol() (in element count, accounting for tmp stride).
• tmp needs at least ceil(src.GetValidRow() / 2) rows.
• When isBinary = false: tmp is accepted but the implementation uses sequential accumulation into dst; tmp is not actively used.

A5¶

• When isBinary = true: tmp is used for binary-tree accumulation in vector registers with UB storage.
• tmp must have the same element type as src/dst.
• tmp.GetValidCol() >= src.GetValidCol() (in element count, accounting for tmp stride).
• tmp needs at least ceil(src.GetValidRow() / 2) rows.
• When isBinary = false: tmp is not actively used; the implementation uses sequential reduction via TColReduceInstr.

Examples¶

Auto¶

#include <pto/pto-inst.hpp>

using namespace pto;

void example_auto() {
  using SrcT = Tile<TileType::Vec, float, 16, 16>;
  using DstT = Tile<TileType::Vec, float, 1, 16>;
  using TmpT = Tile<TileType::Vec, float, 16, 16>;
  SrcT src;
  DstT dst;
  TmpT tmp;
  TCOLSUM(dst, src, tmp, /*isBinary=*/false);
}

Manual¶

#include <pto/pto-inst.hpp>

using namespace pto;

void example_manual() {
  using SrcT = Tile<TileType::Vec, float, 16, 16>;
  using DstT = Tile<TileType::Vec, float, 1, 16>;
  using TmpT = Tile<TileType::Vec, float, 16, 16>;
  SrcT src;
  DstT dst;
  TmpT tmp;
  TASSIGN(src, 0x1000);
  TASSIGN(dst, 0x2000);
  TASSIGN(tmp, 0x3000);
  TCOLSUM(dst, src, tmp, /*isBinary=*/false);
}

ASM Form Examples¶

Auto Mode¶

# Auto mode: compiler/runtime-managed placement and scheduling.
%dst = pto.tcolsum %src : !pto.tile<...> -> !pto.tile<...>

Manual Mode¶

# Manual mode: resources must be bound explicitly before issuing the instruction.
# Optional for tile operands:
# pto.tassign %arg0, @tile(0x1000)
# pto.tassign %arg1, @tile(0x2000)
%dst = pto.tcolsum %src : !pto.tile<...> -> !pto.tile<...>

PTO Assembly Form¶

%dst = tcolsum %src {isBinary = false} : !pto.tile<...> -> !pto.tile<...>
# AS Level 2 (DPS)
pto.tcolsum ins(%src : !pto.tile_buf<...>) outs(%dst : !pto.tile_buf<...>)