Debian Codegraph @

.cargo/vendor/rustfft/CHANGELOG.md

raw
1 ## [6.2]
2 Released 22nd January 2024
3 ### Minimum Rustc Version
4 - The MSRV for RustFFT is now 1.61.0
5 ### Added
6 - Implemented a code path for SIMD-optimized FFTs on WASM targets (Thanks to @pr1metine) (#120)
7 ### Fixed
8 - Fixed pointer aliasing causing unsoundness and miri check failures (#113)
9 - Fixed computation of size-1 FFTs (#119)
10 - Fixed readme type (#121)
11
12 ## [6.1]
13 Released 7th Novemeber 2022
14 ### Added
15 - Implemented a code path for Neon-optimized FFTs on AArch64 (Thanks to Henrik Enquist!) (#84 and #78)
16 ### Changed
17 - Improved performance of power-of-3 FFTs when not using SIMD-accelerated code paths (#80)
18 - Reduced memory usage for some FFT sizes (#81)
19
20 ## [6.0.1]
21 Released 10 May 2021
22 ### Fixed
23 - Fixed a compile-time divide by zero error on nightly Rust in `stdarch\crates\core_arch\src\macros.rs` (#75)
24 - Increased the minimum version of `strength_reduce` to 0.2.3
25
26
27 ## [6.0.0]
28 Released 16 April 2021
29 ### Breaking Changes
30 - Increased the version of the num-complex dependency to 0.4.
31 - This is a breaking change because we have a public dependency on num-complex.
32 - See the [num-complex changelog](https://github.com/rust-num/num-complex/blob/master/RELEASES.md) for a list of breaking changes in num-complex 0.4
33 - As a high-level summary, most users will not need to do anything to upgrade to RustFFT 6.0: num-complex 0.4 re-exports a newer version of `rand`, and that's num-complex's only documented breaking change.
34
35 ## [5.1.1]
36 Released 10 May 2021
37 ### Fixed
38 - Fixed a compile-time divide by zero error on nightly Rust in `stdarch\crates\core_arch\src\macros.rs` (Backported from v6.0.1)
39 - Increased the minimum version of `strength_reduce` to 0.2.3 (Backported from v6.0.1)
40
41 ## [5.1.0]
42 Released 16 April 2021
43 ### Added
44 - Implemented a code path for SSE-optimized FFTs (Thanks to Henrik Enquist!) (#60)
45 - Plan a FFT using the `FftPlanner` (or the new `FftPlannerSse`) on a machine that supports SSE4.1 (but not AVX) and you'll see a 2-3x performance improvement over the default scalar code.
46 ### Fixed
47 - Fixed underflow when planning an AVX FFT of size zero (#56)
48 - Fixed the FFT planner not being Send, due to internal use of Rc<> (#55)
49 - Fixed typo in documentation (#54)
50 - Slightly improved numerical precision of Rader's Algorithm and Bluestein's Algorithm (#66, #68)
51 - Minor optimizations to Rader's Algorithm and Bluestein's Algorithm (#59)
52 - Minor optimizations to MixedRadix setup time (#57)
53 - Optimized performance of Radix4 (#65)
54
55 ## [5.0.1]
56 Released 8 January 2021
57 ### Fixed
58 - Fixed the FFT planner not choosing an obviously faster plan in some rare cases (#46)
59 - Documentation fixes and clarificarions (#47, #48, #51)
60
61 ## [5.0.0]
62 Released 4 January 2021
63 ### Breaking Changes
64 - Several breaking changes. See the [Upgrade Guide](/UpgradeGuide4to5.md) for details.
65
66 ### Added
67 - Added support for the `Avx` instruction set. Plan a FFT with the `FftPlanner` on a machine that supports AVX, and you'll get a 5x-10x speedup in FFT performance.
68
69 ### Changed
70 - Even though the main focus of this release is on AVX, most users should see moderate performance improvements due to a new internal architecture that reduces the amount of internal copies required when computing a FFT.
71
72 ## [4.1.0]
73 Released 24 December 2020
74 ### Added
75 - Added a blanket impl of `FftNum` to any type that implements the required traits (#7)
76 - Added butterflies for many prime sizes, up to 31, and optimized the size-3, size-5, and size-7 buitterflies (#10)
77 - Added an implementation of Bluestein's Algorithm (#6)
78
79 ### Changed
80 - Improved the performance of GoodThomasAlgorithm re-indexing (#20)
81
82 ## [4.0.0]
83 Released 8 October 2020
84
85 This release moved the home repository of RustFFT from https://github.com/awelkie/RustFFT to https://github.com/ejmahler/RustFFT
86
87 ### Breaking Changes
88 - Increased the version of the num-complex dependency to 0.3. This is a breaking change because we have a public dependency on num-complex.
89 See the [num-complex changelog](https://github.com/rust-num/num-complex/blob/master/RELEASES.md) for a list of breaking changes in num-complex 0.3.
90 - Increased the minimum required Rust version from 1.26 to 1.31. This was required by the upgrade to num-complex 0.3.
91
92
93 ## [3.0.1]
94 Released 27 December 2019
95 ### Fixed
96 - Fixed warnings regarding "dyn trait", and warnings regarding inclusive ranges
97 - Several documentation improvements
98
99 ## [3.0.0]
100 Released 4 January 2019
101 ### Changed
102 - Reduced the setup time and memory usage of GoodThomasAlgorithm
103 - Reduced the setup time and memory usage of RadersAlgorithm
104
105 ### Breaking Changes
106 - Documented the minimum rustsc version. Before, none was specified. now, it's 1.26. Further increases to minimum version will be a breaking change.
107 - Increased the version of the num-complex dependency to 0.2. This is a breaking change because we have a public dependency on num-complex.
108 See the [num-complex changelog](https://github.com/rust-num/num-complex/blob/master/RELEASES.md) for a list of breaking changes in num-complex 0.2
109
110 ## [2.1.0]
111 Released 30 July 2018
112 ### Added
113 - Added a specialized implementation of Good Thomas Algorithm for when both inner FFTs are butterflies
114
115 ### Changed
116 - Documentation typo fixes
117 - Increased minimum version of num_traits and num_complex. Notably, Complex<T> is now guaranteed to be repr(C)
118 - Significantly improved the performance of the Radix4 algorithm
119 - Reduced memory usage of prime-sized FFTs
120 - Incorporated the Good-Thomas Double Butterfly algorithm into the planner, improving performance for most composite and prime FFTs
121
122 ## [2.0.0]
123 Released 22 May 2017
124 ### Added
125 - Added implementation of Good Thomas algorithm.
126 - Added implementation of Raders algorithm.
127 - Added implementation of Radix algorithm for power-of-two lengths.
128 - Added `FFTPlanner` to choose the fastest algorithm for a given size.
129
130 ### Changed
131 - Changed API to take the "signal" as mutable and use it for scratch space.
132
133 ## [1.0.1]
134 Released 15 January 2016
135 ### Changed
136 - Relicensed to dual MIT/Apache-2.0.
137
138 ## [1.0.0]
139 Released 4 October 2015
140 ### Added
141 - Added initial implementation of Cooley-Tukey.