Changelog:
==========
* Improved CRC-32 performance on recent x86 CPUs by adding
VPCLMULQDQ-accelerated implementations using 256-bit and 512-bit vectors.
* Improved Adler-32 performance on recent x86 CPUs by adding
VNNI-accelerated implementations using 256-bit and 512-bit vectors.
* Improved CRC-32 and Adler-32 performance on short inputs.
* Optimized the portable implementation of Adler-32.
* Added some basic optimizations for RISC-V.
* Dropped support for gcc versions older than v4.9 (released in 2014)
and clang versions older than v3.9 (released in 2016).
* Dropped support for CRC-32 acceleration on 32-bit ARM using the ARMv8 pmull or
crc32 instructions.
Signed-off-by: Wang Mingyu <wangmy@fujitsu.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
* Added new functions 'libdeflate_alloc_compressor_ex()' and
'libdeflate_alloc_decompressor_ex()'. These functions allow specifying a
custom memory allocator on a per-compressor basis.
* libdeflate now always generates Huffman codes with at least 2 codewords. This
fixes a compatibility issue where Windows Explorer's ZIP unpacker could not
decompress DEFLATE streams created by libdeflate. libdeflate's behavior was
allowed by the DEFLATE RFC, but not all software was okay with it. In rare
cases, compression ratios can be slightly reduced by this change.
* Disabled the use of some compiler intrinsics on MSVC versions where they don't
work correctly.
* libdeflate can now compress up to the exact size of the output buffer.
* Slightly improved compression performance at levels 1-9.
* Improved the compression ratio of very short inputs.
Signed-off-by: Wang Mingyu <wangmy@fujitsu.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Changelog:
==========
* Fixed a bug where the build type didn't default to "Release" when using
CMake 3.10 or earlier.
* Fixed a bug where some optimized code wasn't used when building with
Clang 15 or later (x86), or with Clang 16 or later (aarch64).
* Fixed build errors with some architecture and compiler combos:
* aarch64 with Clang 16
* armv6kz or armv7e-m with gcc
* armhf with gcc (on Debian only)
Signed-off-by: Wang Mingyu <wangmy@fujitsu.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Martin Jansa
Wang Mingyu
Markus Volk