Skip to content

Version 3.0.0 Release#9

Merged
ShatrovOA merged 10 commits intomasterfrom
development
Dec 16, 2025
Merged

Version 3.0.0 Release#9
ShatrovOA merged 10 commits intomasterfrom
development

Conversation

@ShatrovOA
Copy link
Owner

Breaking Changes

  • dtfft_transpose now accepts an optional aux buffer that can be utilized when the underlying backend is pipelined
  • dtfft_get_pencil now requires the new enum dtfft_layout_t instead of the dim attribute to retrieve the appropriate layout
  • Size of the aux buffer for dtfft_execute should be obtained from dtfft_get_aux_size instead of dtfft_get_alloc_size
  • Removed n_configs_to_test and enable_kernel_optimization from dtfft_config_t and their corresponding environment variables
  • Renamed force_kernel_optimization to enable_kernel_autotune in dtfft_config_t
  • Renamed several error codes for consistency

New Features

  • Added brick decomposition support for both 2D and 3D plans on Host and CUDA platforms

  • Added new functions to perform data reshaping between bricks and pencils layouts:

    • dtfft_reshape - synchronous reshape
    • dtfft_reshape_start - start asynchronous reshape
    • dtfft_reshape_end - finalize asynchronous reshape

    All reshape functions accept an optional aux argument for pipelined backends, similar to dtfft_transpose

  • Added new effort level: DTFFT_EXHAUSTIVE

    • Enables kernel parameter autotuning for Host/CUDA kernels
    • Enables reshape backend autotuning
  • Added new API functions:

    • dtfft_get_reshape_backend - returns the backend used for reshape operations
    • dtfft_get_backend_pipelined - checks if a backend is pipelined
    • dtfft_get_aux_size - returns auxiliary buffer size in elements
    • dtfft_get_aux_bytes - returns auxiliary buffer size in bytes
  • R2C plans are now part of the public API even when building without FFT support

    • Transpose-only R2C plans can be created
    • execute is not permitted on transpose-only R2C plans
    • Only reshape and transpose operations are allowed
  • Updated documentation with detailed usage guide covering:

    • Brick decomposition workflows
    • Reshape strategies
  • Added enable_fourier_reshape option to dtfft_config_t to control reshape execution in Fourier space

@codecov
Copy link

codecov bot commented Dec 16, 2025

Codecov Report

❌ Patch coverage is 85.07183% with 239 lines in your changes missing coverage. Please review.
✅ Project coverage is 89.14%. Comparing base (0a1abc3) to head (a701cef).
⚠️ Report is 1 commits behind head on master.

Additional details and impacted files
@@            Coverage Diff             @@
##           master       #9      +/-   ##
==========================================
- Coverage   89.85%   89.14%   -0.71%     
==========================================
  Files          24       26       +2     
  Lines        3814     4790     +976     
  Branches      299      410     +111     
==========================================
+ Hits         3427     4270     +843     
- Misses        387      520     +133     
🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

@ShatrovOA ShatrovOA merged commit 9649608 into master Dec 16, 2025
14 checks passed
@ShatrovOA ShatrovOA deleted the development branch January 6, 2026 23:29
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant