Build Packages / Unit tests (push) Skipped
Build Packages / build:windows:nocuda (push) Successful in 15m31s
Build Packages / build:viewer-tgz:cpu (push) Successful in 5m46s
Build Packages / build:viewer-tgz:cuda (push) Successful in 6m9s
Build Packages / build:rpm (rocky8_nocuda) (push) Successful in 9m25s
Build Packages / build:rpm (rocky9_nocuda) (push) Successful in 10m21s
Build Packages / build:rpm (ubuntu2204_nocuda) (push) Successful in 9m41s
Build Packages / build:rpm (ubuntu2404_nocuda) (push) Successful in 9m18s
Build Packages / build:rpm (rocky8_sls9) (push) Successful in 10m26s
Build Packages / build:rpm (rocky9_sls9) (push) Successful in 11m33s
Build Packages / build:rpm (rocky8) (push) Successful in 10m32s
Build Packages / build:rpm (rocky9) (push) Successful in 12m23s
Build Packages / build:rpm (ubuntu2204) (push) Successful in 10m50s
Build Packages / build:rpm (ubuntu2404) (push) Successful in 10m12s
Build Packages / DIALS test (push) Successful in 12m6s
Build Packages / XDS test (durin plugin) (push) Successful in 8m15s
Build Packages / XDS test (JFJoch plugin) (push) Successful in 7m12s
Build Packages / XDS test (neggia plugin) (push) Successful in 5m35s
Build Packages / Generate python client (push) Successful in 27s
Build Packages / Build documentation (push) Successful in 54s
Build Packages / Create release (push) Skipped
Build Packages / build:windows:cuda (push) Successful in 12m37s
This is an UNSTABLE release. It includes many experimental features, as well as many AI generated fixes. We recommend using rc.152 for production use. * jfjoch_process: Major rotation (rot3d) data processing overhaul - robust profile-fit integration, Cauchy-loss scaling with optional absorption surface, de-novo indexing and space-group/centering determination fixes, and merging statistics + ISa in the mmCIF output. * jfjoch_process: Add EXPERIMENTAL ice-ring detection (--detect-ice-rings) that excludes ice reflections from scaling. * Compression: Add BSHUF_ZSTD_RLE_HUFF, make compression size-aware (drop frames that don't fit rather than aborting), and add the jfjoch_recompress tool. * jfjoch_viewer: Report "Multiple lattices detected" and grey out "Analyze dataset" on a live connection. * jfjoch_broker: Write smargon chi/phi goniometer positions to NXmx; read sensor thickness/material from HDF5 metadata. * CI: Build Windows (CUDA and non-CUDA) installers.Reviewed-on: #66 Co-authored-by: Filip Leonarski <filip.leonarski@psi.ch>
60 lines
2.5 KiB
C++
60 lines
2.5 KiB
C++
// SPDX-FileCopyrightText: 2024 Filip Leonarski, Paul Scherrer Institute <filip.leonarski@psi.ch>
|
|
// SPDX-License-Identifier: GPL-3.0-only
|
|
|
|
#pragma once
|
|
|
|
#include <bitshuffle/bitshuffle.h>
|
|
|
|
#include <vector>
|
|
#include <cstdint>
|
|
#include <cstddef>
|
|
#include "CompressionAlgorithmEnum.h"
|
|
#include "MaxCompressedSize.h"
|
|
|
|
#include "JFJochZstdCompressor.h"
|
|
#include "JFJochZstdHuffCompressor.h"
|
|
|
|
class JFJochBitShuffleCompressor {
|
|
JFJochZstdCompressor zstd_compressor;
|
|
JFJochZstdHuffCompressor huff_compressor;
|
|
CompressionAlgorithm algorithm;
|
|
std::vector<char> tmp_space;
|
|
std::vector<char> scratch;
|
|
|
|
size_t CompressBlock(char *dest, const char * source, size_t nelements, size_t elem_size);
|
|
public:
|
|
// The bitshuffle block size is chosen as a byte target rather than a fixed element count, so
|
|
// the per-block working set - and thus the cache behaviour - stays constant across pixel bit
|
|
// depths. The target is per-algorithm: LZ4 favours a small, cache-resident block (throughput),
|
|
// ZSTD/RLE a large block (ratio). The element block size scales inversely with elem_size,
|
|
// which is always a power of two here, so the result is a power of two >= 8 (a valid bitshuffle
|
|
// block, multiple of 8). E.g. ZSTD 128 kB -> 65536 elem at 16-bit, 32768 at 32-bit.
|
|
constexpr static size_t DefaultBlockSizeBytes(CompressionAlgorithm algorithm) {
|
|
return algorithm == CompressionAlgorithm::BSHUF_LZ4 ? 16384 : 131072;
|
|
}
|
|
constexpr static size_t BlockSize(CompressionAlgorithm algorithm, size_t elem_size) {
|
|
return DefaultBlockSizeBytes(algorithm) / elem_size;
|
|
}
|
|
|
|
explicit JFJochBitShuffleCompressor(CompressionAlgorithm algorithm);
|
|
|
|
template<class T>
|
|
size_t Compress(void *dest, size_t dest_size, const std::vector<T> &src) {
|
|
return Compress(dest, dest_size, src.data(), src.size(), sizeof(T));
|
|
};
|
|
|
|
template<class T>
|
|
std::vector<uint8_t> Compress(const std::vector<T> &src) {
|
|
return Compress(src.data(), src.size(), sizeof(T));
|
|
}
|
|
std::vector<uint8_t> Compress(const void* source, size_t nelements, size_t elem_size);
|
|
// Throws CompressionBufferTooSmallException if the compressed output would not fit dest_size.
|
|
size_t Compress(void *dest, size_t dest_size, const void* source, size_t nelements, size_t elem_size);
|
|
};
|
|
|
|
template <class T> std::vector<T> bitshuffle(const std::vector<T> &input, size_t block_size) {
|
|
std::vector<T> ret(input.size());
|
|
bshuf_bitshuffle(input.data(), ret.data(), input.size(), sizeof(T), block_size);
|
|
return ret;
|
|
}
|