lookup8.c

#include <immintrin.h>
#include <stdio.h>
#include <intrin.h>

/*
lookup8:
	subq	$12, %rsp
	.seh_stackalloc	12
	.seh_endprologue
	movl	%edx, (%rsp)
	vmovdqu	(%rcx), %ymm0
	vbroadcastss	(%rsp), %ymm1
	vpcmpeqd	%ymm0, %ymm1, %ymm0
	vptest	%ymm0, %ymm0
	je	.L3
	vpmovmskb	%ymm0, %eax
	bsrl	%eax, %eax
	shrl	$2, %eax
.L4:
	vzeroupper
	addq	$12, %rsp
	ret
.L3:
	movl	$-1, %eax
	jmp	.L4
*/

int table[8] = {
    50,
    13,
    9,
    12,

    15,
    89,
    93,
    92
};

int
lookup8(int *table, int val)
{
    __m256i v256 = _mm256_set1_epi32(val);

    __m256i cmp = _mm256_cmpeq_epi32(v256, _mm256_loadu_si256((__m256i*)table));

    if (_mm256_testz_si256(cmp,cmp)) {
        return -1;
    }

    int bits = _mm256_movemask_epi8(cmp);
    //printf("%x\n", bits);
#ifdef __GNUC__
    unsigned int pos = _bit_scan_reverse(bits);
#else
    unsigned int pos;
    _BitScanReverse(&pos, bits);
#endif

    return pos >> 2;
}

int main(int argc, char **argv)
{
    if (argc < 2) {
        puts("usage : lookup8 val");
        return 1;
    }

    int val = atoi(argv[1]);

    printf("%d\n", lookup8(table, val));
}