TREMS¶
Tile Operation Diagram¶
Introduction¶
Elementwise remainder with a scalar: %.
Math Interpretation¶
For each element (i, j) in the valid region:
\[\mathrm{dst}_{i,j} = \mathrm{src}_{i,j} \bmod \mathrm{scalar}\]
Assembly Syntax¶
PTO-AS form: see PTO-AS Specification.
Synchronous form:
%dst = trems %src, %scalar : !pto.tile<...>, f32
AS Level 1 (SSA)¶
%dst = pto.trems %src, %scalar : (!pto.tile<...>, dtype) -> !pto.tile<...>
AS Level 2 (DPS)¶
pto.trems ins(%src, %scalar : !pto.tile_buf<...>, dtype) outs(%dst : !pto.tile_buf<...>)
C++ Intrinsic¶
Declared in include/pto/common/pto_instr.hpp:
template <typename TileDataDst, typename TileDataSrc, typename TileDataTmp, typename... WaitEvents>
PTO_INST RecordEvent TREMS(TileDataDst &dst, TileDataSrc &src, typename TileDataSrc::DType scalar,
TileDataTmp &tmp, WaitEvents &... events);
Constraints¶
- Implementation Checks (A2A3):
dstandsrcmust use the same element type.- Supported element types:
floatandint32_t. dstandsrcmust be vector tiles.dstandsrcmust be row-major.- Runtime:
dst.GetValidRow() == src.GetValidRow() > 0anddst.GetValidCol() == src.GetValidCol() > 0. - tmp Buffer Requirements:
tmp.GetValidCol() >= dst.GetValidCol()(at least as many columns as dst)tmp.GetValidRow() >= 1(at least 1 row)- Data type must match
TileDataDst::DType.
- Implementation Checks (A5):
dstandsrcmust use the same element type.- Supported element types:
float,int32_t,uint32_t,half,int16_t, anduint16_t. dstandsrcmust be vector tiles.- Static valid bounds:
ValidRow <= RowsandValidCol <= Colsfor both tiles. - Runtime:
dst.GetValidRow() == src.GetValidRow()anddst.GetValidCol() == src.GetValidCol(). - Note: tmp parameter is accepted but not validated or used on A5.
- Division by Zero:
- Behavior is target-defined; the CPU simulator asserts in debug builds.
- Valid Region:
- The op uses
dst.GetValidRow()/dst.GetValidCol()as the iteration domain.
- The op uses
- For
int32_tInputs (A2A3 Only): Bothsrcelements andscalarmust be in the range[-2^24, 2^24](i.e.,[-16777216, 16777216]) to ensure exact conversion to float32 during computation.
Examples¶
#include <pto/pto-inst.hpp>
using namespace pto;
void example() {
using TileT = Tile<TileType::Vec, float, 16, 16>;
TileT x, out;
Tile<TileType::Vec, float, 16, 16> tmp;
TREMS(out, x, 3.0f, tmp);
}
ASM Form Examples¶
Auto Mode¶
# Auto mode: compiler/runtime-managed placement and scheduling.
%dst = pto.trems %src, %scalar : (!pto.tile<...>, dtype) -> !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.trems %src, %scalar : (!pto.tile<...>, dtype) -> !pto.tile<...>
PTO Assembly Form¶
%dst = trems %src, %scalar : !pto.tile<...>, f32
# AS Level 2 (DPS)
pto.trems ins(%src, %scalar : !pto.tile_buf<...>, dtype) outs(%dst : !pto.tile_buf<...>)