Skip to content

Latest commit

 

History

History
56 lines (31 loc) · 1.32 KB

CHANGELOG.md

File metadata and controls

56 lines (31 loc) · 1.32 KB
Raw

Changelog

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

Unreleased

[v0.5.4] - 2024-09-25

Changed

  • Removed explicit GPU synchronisation barriers (using KA.synchronize) by default. This can now be re-enabled by passing synchronise = true as a plan argument. Enabling synchronisation is useful for getting accurate timings (in p.timer) but may result in decreased performance.

[v0.5.3] - 2024-09-24

Changed

  • Faster spatial sorting of non-uniform points (CPU and GPU).

  • Tune GPU parameters: kernel workgroupsize; block size for spatial sorting.

  • Plans: block_size argument can now be a tuple (block size along each separate dimension).

[v0.5.2] - 2024-09-23

Changed

  • Avoid recompilation of GPU kernels when number of non-uniform points changes.

[v0.5.1] - 2024-09-20

Fixed

  • Fix transforms of real non-uniform data on CUDA.jl.

[v0.5.0] - 2024-09-20

Added

  • Add preliminary GPU support.

[v0.4.1] - 2024-09-14

Fixed

  • AbstractNFFTs interface: fix 1D transforms.

[v0.4.0] - 2024-09-13

Added

  • Implement AbstractNFFTs interface for easier comparison with other NUFFT packages.