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
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
|
# Getting libaec
The source code of libaec is hosted at DKRZ GitLab.
## Source code and binary releases
The latest releases of libaec can be downloaded at the following
locations:
https://gitlab.dkrz.de/k202009/libaec/-/releases
or
https://github.com/MathisRosenhauer/libaec/releases
## Developer snapshot
```shell
git clone https://gitlab.dkrz.de/k202009/libaec.git
```
# Installation
## General considerations
Libaec achieves the best performance on 64 bit systems. The library
will work correctly on 32 bit systems but encoding and decoding
performance will be much lower.
## Installation from source code release with configure
The most common installation procedure on Unix-like systems looks as
follows:
Unpack the tar archive and change into the unpacked directory.
```shell
mkdir build
cd build
../configure
make check install
```
## Installation from source code release with CMake
As an alternative, you can use CMake to install libaec.
Unpack the tar archive and change into the unpacked directory.
```shell
mkdir build
cd build
cmake ..
make install
```
You can set options for compiling using the CMake GUI by replacing the cmake
command with
```shell
cmake-gui ..
```
or by setting the options manually, e.g.
```shell
cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=~/local ..
```
CMake can also generate project files for Microsoft Visual Studio when
used in Windows.
## Installation from cloned repository
The configure script is not included in the repository. You can
generate it with autotools and gnulib:
```shell
cd libaec
gnulib-tool --import lib-symbol-visibility
autoreconf -iv
mkdir build
cd build
../configure
make check install
```
# Optimization
Libaec in general and encoding performance in particular can benefit
from vectorization and other compiler optimizations. You can try to
enable higher than default optimizations and check the benefits with
the bench target.
## Intel compiler
Assuming your CPU supports AVX2, the following options will increase
encoding speed.
```shell
../configure CC=icc
make CFLAGS="-O3 -xCORE-AVX2" bench
```
On a 3.4 GHz E3-1240 v3 we see more than 400 MiB/s for encoding
typical data.
## gcc
The default -O2 will already enable vectorization but -O3 yields even
better performance.
|
