simd-kernels 0.2.1

Lightning Fast, Arrow-Compatible Compute Kernels
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75

[package]
name = "simd-kernels"
description = "Lightning Fast, Arrow-Compatible Compute Kernels"
version = "0.2.1"
edition = "2024"
authors = ['Peter Bower']
license-file = "LICENSE"
keywords = [
    "simd",
    "kernels",
    "arrow",
    "polars",
    "data",
]
[dependencies]
minarrow = { workspace = true, default-features = false, optional = false, features = ["views"] }
num-traits = "0.2.19"
ordered-float = { version = "5.0.0", optional = true }
num-complex = { version = "0.4.6", optional = true }
blas = { version = "0.23.0", optional = true }
lapack = { version = "0.20.0", optional = true }
blas-src = { version = "0.10", features = ["openblas"], optional = true }
lapack-src = { version = "0.10", features = ["openblas"], optional = true }
openblas-src = { version = "0.10", features = ["system"], optional = true }
ahash = { version = "0.8.12", optional = true }
rand = "0.9.1"

# Parallel sorting
rayon = { version = "1.10", optional = true }

# SIMD-accelerated sorting
voracious_radix_sort = { version = "1.2", optional = true }

[build-dependencies]
cc = { version = "1", optional = true }

[features]
large_string = ["minarrow/large_string"]
datetime = ["minarrow/datetime"]
simd = ["minarrow/simd"]
# Replaces all hashmaps and hashsets used for count distinct operations
# and categorical dictionary interning with the faster ahash.
fast_hash = ["dep:ahash"]

# Enables linear algebra kernels. These call out to BLAS and LAPACK.
# Both `Intel` and `OpenBLAS` are supported. Prefer Intel if you have intel for 
# speed, except that it is not fully open source, and one should understand its
# license terms, with an FAQ available at
# https://www.intel.com/content/www/us/en/developer/articles/tool/onemkl-license-faq.html
#
# ⚠️ Warning: Multivariate modules are still under development, and are not fully tested.
linear_algebra = ["blas", "lapack", "blas-src", "lapack-src", "openblas-src"]

# Arbitrary element-wise functions. This pattern provides the basis,
# and it is extendable to any set.
universal_functions = []
# Enables FFT kernels
fourier_transforms = ["dep:num-complex"]
# Enables CDF and PDF probability distributions
probability_distributions = []
extended_numeric_types = ["minarrow/extended_numeric_types"]

# Parallel sorting using rayon
parallel_sort = ["dep:rayon"]

# SIMD-accelerated radix sort for maximum integer sorting performance
simd_sort = ["dep:voracious_radix_sort"]

default = [
    "large_string",
    "probability_distributions",
    "simd",
    "fourier_transforms",
    "universal_functions",
]