1 /*-------------------------------------------------------------------------
4 * Compute CRC-32C checksum using Intel SSE 4.2 instructions.
6 * Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group
7 * Portions Copyright (c) 1994, Regents of the University of California
11 * src/port/pg_crc32c_sse42.c
13 *-------------------------------------------------------------------------
17 #include <nmmintrin.h>
19 #include "port/pg_crc32c.h"
22 pg_comp_crc32c_sse42(pg_crc32c crc, const void *data, size_t len)
24 const unsigned char *p = data;
25 const unsigned char *pend = p + len;
28 * Process eight bytes of data at a time.
30 * NB: We do unaligned accesses here. The Intel architecture allows that,
31 * and performance testing didn't show any performance gain from aligning
37 crc = (uint32) _mm_crc32_u64(crc, *((const uint64 *) p));
41 /* Process remaining full four bytes if any */
44 crc = _mm_crc32_u32(crc, *((const unsigned int *) p));
50 * Process four bytes at a time. (The eight byte instruction is not
51 * available on the 32-bit x86 architecture).
55 crc = _mm_crc32_u32(crc, *((const unsigned int *) p));
58 #endif /* __x86_64__ */
60 /* Process any remaining bytes one at a time. */
63 crc = _mm_crc32_u8(crc, *p);