6.1.0

• It is now possible to perform multidimensional DFT via the C API.
• Documentation updated.
• [Internal] Updated to the latest CxxDox for enhanced documentation features.
• The Dims template parameter in dft_plan_md and dft_plan_md_real now defaults to dynamic_shape.
• A workaround has been applied to mitigate the MSVC Internal Compiler Error.
• Additional tests added.

Full Changelog: 6.0.4...6.1.0

6.0.4

This release includes fixes to ensure compatibility with Clang 19. No new features have been added.

6.0.3

KFR 6.0.3 fixes and improvements

• More documentation (thanks to @Jalmenara)
• Stateless parameter fixed
• Fix build with non-MSVC on Win32 (thanks to @jcelerier)
• Android fixes
• Removed deprecated atomic initialization

6.0.2

What's new in KFR 6

• DFT performance has been improved up to 40% (backported to KFR 5.2.0 branch)
• C API for non x86 architectures
• DSP refactoring with easier initialization
• Multiarchitecture for resampling, FIR and IIR filters
• matrix_transpose: optimized matrix transpose (square/non-square, inplace/out-of-place, real/complex, scalar/vectors)
• CMake config file generation (find_package(KFR CONFIG) support, see installation)
• .npy format support (reading/writing, v1/v2, c/fortran order, real/complex, bigendian/littleendian)
• Multidimensional DFT: real/complex
• inline_vector
• Windows arm64 support
• Emscripten (wasm/wasm64) support

Other changes

• CMake minimum version is 3.12
• Multidimensional reference DFT
• Easier cross compilation to ARM64 on x86_64 macOS
• Automated tests using GitHub Actions (previously Azure Pipelines)
• GCC 7 and 8: emulate missing avx-512 instrinsics
• read_group and write_group
• [❗breaking change] reshape_may_copy and flatten_may_copy in tensor<> allows copying by default
• shape<>::transpose function
• tensor<>::transpose function
• convert_endianess
• DFT, DSP and IO sources have been moved to src/ directory
• Multiarchitecture is enabled by default
• KFR_DFT_NO_NPo2 has been removed (assumed always enabled)
• Tests refactoring
• Some tests moved to tests/internal/
• [❗breaking change] Scalars are now passed by value in expressions (this fixes dangling references in some cases)
• Expression functions should return expression_make_function instead of expression_function
• KFR_WITH_CLANG
• KFR_VERSION CMake variable
• Functions to get module versions (library_version_dft, library_version_dsp etc)
• Exceptions are no longer enforced in MSVC
• kfr::complex removed (use std::complex instead). KFR_STD_COMPLEX cmake variable removed too
• strides_for_shape for fortran order
• AARCH and ARM emulation refactoring (dynamic libraries are now supported)
• call_with_temp
• maximum_dims is now 16 (was 8)
• to_fmt/from_fmt supports inplace
• shape refactoring: rotate_left, rotate_right, remove_back, remove_front
• temp argument can be nullptr for DFT (temporary buffer will be allocated on stack or heap)
• dft_plan and similar classes have now default and move constructors
• -DCMAKE_POSITION_INDEPENDENT_CODE=ON is required for building C API
• ci/run.sh can now build in a directory outside source tree
• [❗breaking change]graphics/color.hpp and graphics/geometry.hpp have been removed
• Simpler CMT_CVAL macro
• /Zc:lambda is now required for building KFR in MSVC
• println for string_view
• MSVC internal compiler error fixed
• Complex vector operators fixed

6.0.0 RC

What's new in KFR 6

• DFT performance has been improved up to 40% (backported to KFR 5.2.0 branch)
• C API for non x86 architectures
• DSP refactoring with easier initialization
• Multiarchitecture for resampling, FIR and IIR filters
• matrix_transpose: optimized matrix transpose (square/non-square, inplace/out-of-place, real/complex, scalar/vectors)
• CMake config file generation (find_package(KFR CONFIG) support, see installation)
• .npy format support (reading/writing, v1/v2, c/fortran order, real/complex, bigendian/littleendian)
• Multidimensional DFT: real/complex
• inline_vector

Other changes

• CMake minimum version is 3.12
• Multidimensional reference DFT
• Easier cross compilation to ARM64 on x86_64 macOS
• Automated tests using GitHub Actions (previously Azure Pipelines)
• GCC 7 and 8: emulate missing avx-512 instrinsics
• read_group and write_group
• [❗breaking change] reshape_may_copy and flatten_may_copy in tensor<> allows copying by default
• shape<>::transpose function
• tensor<>::transpose function
• convert_endianess
• DFT, DSP and IO sources have been moved to src/ directory
• Multiarchitecture is enabled by default
• KFR_DFT_NO_NPo2 has been removed (assumed always enabled)
• Tests refactoring
• Some tests moved to tests/internal/
• [❗breaking change] Scalars are now passed by value in expressions (this fixes dangling references in some cases)
• Expression functions should return expression_make_function instead of expression_function
• KFR_WITH_CLANG
• KFR_VERSION CMake variable
• Functions to get module versions (library_version_dft, library_version_dsp etc)
• Exceptions are no longer enforced in MSVC
• kfr::complex removed (use std::complex instead). KFR_STD_COMPLEX cmake variable removed too
• strides_for_shape for fortran order
• AARCH and ARM emulation refactoring (dynamic libraries are now supported)
• call_with_temp
• maximum_dims is now 16 (was 8)
• to_fmt/from_fmt supports inplace
• shape refactoring: rotate_left, rotate_right, remove_back, remove_front
• temp argument can be nullptr for DFT (temporary buffer will be allocated on stack or heap)
• dft_plan and similar classes have now default and move constructors
• -DCMAKE_POSITION_INDEPENDENT_CODE=ON is required for building C API
• ci/run.sh can now build in a directory outside source tree
• [❗breaking change]graphics/color.hpp and graphics/geometry.hpp have been removed
• Simpler CMT_CVAL macro
• /Zc:lambda is now required for building KFR in MSVC
• println for string_view
• MSVC internal compiler error fixed
• Complex vector operators fixed

5.2.0

This release mainly focuses on DFT performance (which has been increased up to 40% on x86 and x86_64) and bug fixes.

Changed

• The performance of DFT has been increased up to 40% compared to KFR 5 on x86 and x86_64
  in single and double precision, inplace and out of place processing.

Added

• KFR_NO_PERF_TESTS define can now disable performance tests
• CMT_CVAL for extracting constexpr-enabled value from cval_t
• fft_algorithm_selection to select FFT algorithm for given FFT size.

Fixed

• Warnings in Clang 10 #198
• std::is_pov is peprecated in C++20 #190
• DFT sizes 0 and 1 were not processed correctly #195
• Internal compiler error in Visual Studio Compiler 19.37 #194
• Goertzel issue #121
• Bug in nearest_real_or_complex #137
• Fixed operators in KFR_STD_COMPLEX mode
• Testo library: typo in epsilon_scope
• Fix ambiguities with std::identity (C++20)
• Force linking in correct order for multi-architecture binaries

5.1.0

Added

• Tukey window function
• Subscript operator for vec<>
• Unary operators for vec<>

Fixed

• Inverse DCT has been fixed
• Allow C API to be built on non-x86 systems
• Tensor iteration range has been fixed
• transpose(vec<vec<>>) has been fixed
• vec<bit<>> bug on GCC and MSVC
• Internal constant is_pod has been removed
• root and cbrt have been fixed for negative values
• Fixed numerous warnings in MSVC

5.0.3

Added

• Planck-taper window function

Fixed

• Sample rate conversion: automatic zero padding
• DFT: incorrect result of real dft when input size != 4N #141
• Add options for installing libraries and headers #182

5.0.2

• ARM/ARM64 performance has been improved up to 2 times in various usage scenarios, including DFT.
• Fix sine and other functions not accepting scalar references
• Fix possible name conflict with fmt

5.0.1

• New tensor<T, dims> class for multidimensional data (like numpy's nparray)
• Histogram computation
• Normal (gaussian) distribution for random number generator
• Documentation updates
• All builtin expressions support multiple dimensions
• Exception support (may be configured to call user-supplied function or std::abort)
• [changes required] CMake variables now have KFR_ prefix
• Template parameter deduction for vec, so vec{1, 2} is the same as vec<int, 2>{1, 2}
• [changes required] random_state is now architecture-agnostic and defined in kfr namespace
• All expression classes have been moved from kfr::CMT_ARCH_NAME::internal to kfr::CMT_ARCH_NAME namespace
• expression_traits<T> introduced to support interpreting any object as kfr expression
• [changes required] User-defined expressions should be rewritten to be used in KFR5
• Out-of-class assign operators for all input & output expressions
• round.hpp, clamp.hpp, select.hpp, sort.hpp, saturation.hpp, min_max.hpp, logical.hpp, abs.hpp headers have been moved to simd module
• state_holder.hpp has been moved to base module
• All code related to expressions have been moved to base module
• vec<T, N>::front() and vec<T, N>::front() are now writable
• set_elements functions for output expressions like get_elements for input expressions