Sergei Lebedev 273cb27047 compat.tensor __*__ methods no longer do implicit broadcasting ...

This change makes it simpler to lower binary operations to Triton IR
bypassing Triton Python bindings.

PiperOrigin-RevId: 601796719

2024-01-26 10:13:51 -08:00

..

__init__.py

Generate Python bindings for the Triton MLIR dialect

2024-01-02 11:55:05 -08:00

_triton_ext.cc

Pallas now uses MLIR Python builders to lower to Triton IR

2024-01-11 13:33:26 -08:00

_triton_ext.pyi

Pallas now uses MLIR Python builders to lower to Triton IR

2024-01-11 13:33:26 -08:00

BUILD

Pallas now uses MLIR Python builders to lower to Triton IR

2024-01-11 13:33:26 -08:00

compat.py

compat.tensor __*__ methods no longer do implicit broadcasting

2024-01-26 10:13:51 -08:00

dialect.py

Pallas now uses MLIR Python builders to lower to Triton IR

2024-01-11 13:33:26 -08:00

triton_dialect_capi.cc

Pallas now uses MLIR Python builders to lower to Triton IR

2024-01-11 13:33:26 -08:00

triton_dialect_capi.h

Pallas now uses MLIR Python builders to lower to Triton IR

2024-01-11 13:33:26 -08:00

triton.td

Generate Python bindings for the Triton MLIR dialect

2024-01-02 11:55:05 -08:00