Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog.

v0.7.1 (2024-07-01)

Changed

XLA_TARGET to default to a matching target when CUDA installation is detected (#88)

v0.7.0 (2024-05-21)

Changed

Bumped XLA version (#83)
Renamed the recognised XLA_TARGET "cuda120" to "cuda12" (#84)

Removed

Dropped support for CUDA 11.8+, now 12.1+ is required (#84)

v0.6.0 (2023-11-10)

Changed

Bumped XLA version (#62)

v0.5.1 (2023-09-14)

Changed

Bumped the version requirement for CUDA 12 to CUDA 12.1 and cuDNN 8.9 (#54)

v0.5.0 (2023-08-13)

Added

Support for custom http headers (#44)
Support for CUDA 12

Changed

Migrated to OpenXLA source code (#45)

Removed

Dropped precompiled binary for CUDA 11.1 and CUDA 11.4
Dropped precompiled binary for Linux musl

v0.4.4 (2023-02-17)

Added

Sorting library (#39)

v0.4.3 (2022-12-15)

Fixed

Building with XLA_BUILD (regression from v0.4.2) (#33)

v0.4.2 (2022-12-15)

Added

Precompiled binaries for Linux musl (#31)

Fixed

Partially fixed building for ROCm, see notes (#30)

v0.4.1 (2022-12-08)

Added

Precompiled binaries for CUDA 11.4+ (cuDNN 8.2+) and CUDA 11.8+ (cuDNN 8.6+) (#27)

Changed

Precompiled binaries to assume glibc 31+ (#27)

v0.4.0 (2022-11-20)

Changed

Bumped XLA (Tensorflow) version to 2.11.0 (#25)

v0.3.0 (2022-02-17)

Changed

Bumped XLA (Tensorflow) version to 2.8.0 (#15)

Removed

Dropped support for CUDA 10.2 and 11.0, now 11.1+ is required (#17)

v0.2.0 (2021-09-23)

Added

Added support for Apple Silicon (#9)

Changed

Bumped XLA (Tensorflow) version to 2.6.0 (#9)

v0.1.1 (2021-09-16)

Changed

Build for older glibc versions (#3)

v0.1.0 (2021-09-16)

Initial release.