blob: fedd47a13246c877d1770a56187c8362405d8fe0 (
plain)
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
|
.. title:: clang-tidy - portability-simd-intrinsics
portability-simd-intrinsics
===========================
Finds SIMD intrinsics calls and suggests ``std::experimental::simd`` (`P0214`_)
alternatives.
If the option ``Suggest`` is set to non-zero, for
.. code-block:: c++
_mm_add_epi32(a, b); // x86
vec_add(a, b); // Power
the check suggests an alternative: ``operator+`` on ``std::experimental::simd``
objects.
Otherwise, it just complains the intrinsics are non-portable (and there are
`P0214`_ alternatives).
Many architectures provide SIMD operations (e.g. x86 SSE/AVX, Power AltiVec/VSX,
ARM NEON). It is common that SIMD code implementing the same algorithm, is
written in multiple target-dispatching pieces to optimize for different
architectures or micro-architectures.
The C++ standard proposal `P0214`_ and its extensions cover many common SIMD
operations. By migrating from target-dependent intrinsics to `P0214`_
operations, the SIMD code can be simplified and pieces for different targets can
be unified.
Refer to `P0214`_ for introduction and motivation for the data-parallel standard
library.
Options
-------
.. option:: Suggest
If this option is set to non-zero (default is `0`), the check will suggest
`P0214`_ alternatives, otherwise it only points out the intrinsic function is
non-portable.
.. option:: Std
The namespace used to suggest `P0214`_ alternatives. If not specified, `std::`
for `-std=c++2a` and `std::experimental::` for `-std=c++11`.
.. _P0214: https://wg21.link/p0214
|
