/* https://github.com/cirosantilli/linux-kernel-module-cheat#x86-simd
 *
 * Add a bunch of integers in one go.
 */

#include "common.h"

ENTRY
.data
    input0:       .long 0xF1F1F1F1, 0xF2F2F2F2, 0xF3F3F3F3, 0xF4F4F4F4
    input1:       .long 0x12121212, 0x13131313, 0x14141414, 0x15151515
    paddb_expect: .long 0x03030303, 0x05050505, 0x07070707, 0x09090909
    paddw_expect: .long 0x04030403, 0x06050605, 0x08070807, 0x0A090A09
    paddd_expect: .long 0x04040403, 0x06060605, 0x08080807, 0x0A0A0A09
    paddq_expect: .long 0x04040403, 0x06060606, 0x08080807, 0x0A0A0A0A
.bss
    result: .skip 16
.text
    movups input1, %xmm1
#define TEST(size) \
    movups input0, %xmm0; \
    padd ## size %xmm1, %xmm0; \
    movups %xmm0, result; \
    ASSERT_MEMCMP(result, padd ## size ## _expect, $0x10)

    TEST(b)
    TEST(w)
    TEST(d)
    TEST(q)
#undef TEST
EXIT