SIMD-accelerated carryless-multiplication CRC-64/NVME checksum computation
(similar to crc32fast and forked from crc64fast which calculates CRC-64/XZ [a.k.a CRC-64/GO-ECMA]).
CRC-64/NVME comes from the NVM Express® NVM Command Set Specification (Revision 1.0d, December 2023) and has also been implemented in the Linux kernel (where it's called CRC-64/Rocksoft) and AWS S3's recommended checksum option as CRC64-NVME. (Note that the Check value in the spec uses incorrect endianness (Section 5.2.1.3.4, Figure 120, page 83).
SIMD-accelerated carryless-multiplication is based on the Intel Fast CRC Computation for Generic Polynomials Using PCLMULQDQ Instruction paper.
Changes from crc64fast
Primarily changes the CRC-64/XZ (aka CRC-64/GO-ECMA) polynomial from crc64fast (which uses the ECMA-182 polynomial [0x42F0E1EBA9EA3693]) to use the NVME polynomial (0xAD93D23594C93659), plus re-calculates the input parameters (tables, keys, mu, and reciprocal polynomial) for fast operations.
use crc64fast_nvme::Digest;
let mut c = Digest::new();
c.write(b"hello ");
c.write(b"world!");
let checksum = c.sum64();
assert_eq!(checksum, 0xd9160d1fa8e418e3);A simple CLI implementation can be found in crc_64_nvme_checksum.rs, which will calculate the CRC-64/NVME checksum for a file on disk.
Tooling to re-calculate input parameters for other CRC-64 implementations/polynomials is supplied in src\bin.
crc64fast-nvme provides two fast implementations, and the most performance one will
be chosen based on CPU feature at runtime.
- a fast, platform-agnostic table-based implementation, processing 16 bytes at a time.
- a SIMD-carryless-multiplication based implementation on modern processors:
- using PCLMULQDQ + SSE 4.1 on x86/x86_64
- using PMULL + NEON on AArch64 (64-bit ARM)
| Algorithm | Throughput (x86_64) | Throughput (aarch64) |
|---|---|---|
| crc 3.0.1 | 0.5 GiB/s | 0.3 GiB/s |
| crc64fast-nvme (table) | 2.3 GiB/s | 1.8 GiB/s |
| crc64fast-nvme (simd) | 28.2 GiB/s | 20.0 GiB/s |
- crc32-fast - Original
crc32implementation in Rust. - crc64-fast - Original
CRC-64/XZimplementation in Rust (from which this project was forked). - Fast CRC Computation for Generic Polynomials Using PCLMULQDQ Instruction - Intel's paper.
- NVM Express® NVM Command Set Specification - The NVMe spec, including
CRC-64-NVME(with incorrect endian Check value). - CRC-64/NVME - The
CRC-64/NVMEquick definition. - Linux implementation - Linux implementation of
CRC-64/NVME. - C++ artifacts implementation - Inspiration C++ for the Rust code in calculate_pclmulqdq_artifacts.rs.
- Intel isa-l GH issue #88 - Additional insight into generating artifacts.
- StackOverflow PCLMULQDQ CRC32 answer - Insightful answer to implementation details for CRC32.
- StackOverflow PCLMULQDQ CRC32 question - Insightful question & answer to CRC32 implementation details.
- AWS S3 announcement about CRC64-NVME support
- AWS S3 docs on checking object integrity using CRC64-NVME
crc64fast-nvme is dual-licensed under
- Apache 2.0 license (LICENSE-Apache or http://www.apache.org/licenses/LICENSE-2.0)
- MIT license (LICENSE-MIT or https://opensource.org/licenses/MIT)