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
|
gf2x-1.3.0:
- gf2x.h now contains GF2X_VERSION_* macros ; previous versions of
offered no way to distinguish between versions. For developers, an
example of how this can be done in config scripts is in
check-gf2x-version.c
- the shared library major version number has been increased to 3.0.0
(as we should have done for 1.2, in fact)
- gf2x now compiles fine with icc
- the fft dynamic library is now a separate file, and a build with or
without --enable-fft-interface does not affect the generated
libgf2x.so in any way.
- the shared libraries now expose only the official functions, not the
internal ones. In particular, even though libgf2x in itself uses
the ternary fft code, that code is not exposed in the external abi
of libgf2x. If you intend to use it, you must use libgf2x-fft.so
(compiled with --enable-fft-interface).
- the fft interface api has been significantly modified and improved.
A new interface for computing middle products has been added.
The C++ wrapper code has been improved (no longer an ugly macro).
- packaging bugs that affected usability of the fft interface have
been fixed.
- two long-standing bugs in the ternary FFT for ridiculously small
transform lengths have been fixed.
- configuration file for pkg-config added: assuming gf2x installed in
<prefix> known to pkg-config, "pkg-config --cflags gf2x" etc will
work. (Otherwise, one would need PKG_CONFIG_PATH=<prefix>/lib/pkg_config).
gf2x-1.2:
- changed the type representing the polynomial sizes from "unsigned int"
to "unsigned long". On a 64-bit machine, the maximal degree was
2^38-1, now it is 2^70-1.
- gf2x_mul is now thread-safe. gf2x_mul_r is here in case a user
wants to provide the storage pool and avoid having gf2x_mul do
allocation for that. Note that currently, for FFT sizes, gf2x_mul
does allocation anyway.
- based (strongly) on the first above change and (mildly) on the
second one, bump the shared library version number to 1.2.0
- added support for armv7l
- made license more clear: only one file is gpl-licensed
(toom-gpl.c), all the rest is lgpl-happy.
gf2x-1.1: (bugfix release)
- We're considering using this file for the first time.
- Redone tuning.
- Fixed a couple of build failures, and autoconf warnings.
- Added a faster mul4
- Added some support for cross compilation (might also have been
broken in the meantime, we don't have regression testing for this).
The interface is unchanged, so the shared library version has not
changed and still is 1.0.0
|
