-
Hey folks!
I have a use-case where we already use `xxh3` for hashing various things, including the key for a previous cache's key. We're using `bigcache` now, but it's a bit awkward doing `strconv.…
-
Wasm supports simd. Could we also have a xxh128 variant?
-
great work!
will xxh64 work in ebpf?
otherwise, which version will work 100% in ebpf?
-
I nearly couldn't find this crate
-
So notably XXH3 assumes you got unaligned access, as it has become kinda standard on modern processors. However, it doesn't have to be as bad as it is for strict alignment targets.
With RISC-V gain…
-
The checksum performance in folly is not optimized on Arm64 with Neon, which induce that the performance is quite slow.
./folly/hash/detail/ChecksumDetail.h
Cachelib heavily rely on Folly to rea…
-
rocksdb/Makefile [is hardwired](https://github.com/facebook/rocksdb/blob/b965a5a80ee1a70fd5de33818e92b4afa0d9e08c/Makefile#L298) to use -march=armv8+a+crc+crypto. That won't take advantage of features…
-
As per the title: for performance reasons (single pass & overlapping writes) a version which performs xxh3 and memcpy would be nice to have. Ideally with the option for nt stores where available.
-
With pull request #713 , XXH3 is optimized by ARM SVE instructions. Since data is divided in blocks in XXH3, and vector instructions could handle data in parallel.
For XXH32 & XXH64, data is fetche…
-
There are 7-Zip codecs supporting Zstandard, Brotli, LZ4, LZ5, Lizard and Fast LZMA2.
https://github.com/mcmilk/7-Zip-zstd
https://www.tc4shell.com/en/7zip/modern7z/
Those algorithms are significan…