void av1_fdct4x4_1_sse2(const int16_t *input, tran_low_t *output, int stride) {
__m128i in0, in1;
__m128i tmp;
const __m128i zero = _mm_setzero_si128();
in0 = _mm_loadl_epi64((const __m128i *)(input + 0 * stride));
in1 = _mm_loadl_epi64((const __m128i *)(input + 1 * stride));
in1 = _mm_unpacklo_epi64(
in1, _mm_loadl_epi64((const __m128i *)(input + 2 * stride)));
in0 = _mm_unpacklo_epi64(
in0, _mm_loadl_epi64((const __m128i *)(input + 3 * stride)));
tmp = _mm_add_epi16(in0, in1);
in0 = _mm_unpacklo_epi16(zero, tmp);
in1 = _mm_unpackhi_epi16(zero, tmp);
in0 = _mm_srai_epi32(in0, 16);
in1 = _mm_srai_epi32(in1, 16);
tmp = _mm_add_epi32(in0, in1);
in0 = _mm_unpacklo_epi32(tmp, zero);
in1 = _mm_unpackhi_epi32(tmp, zero);
tmp = _mm_add_epi32(in0, in1);
in0 = _mm_srli_si128(tmp, 8);
in1 = _mm_add_epi32(tmp, in0);
in0 = _mm_slli_epi32(in1, 1);
store_output(&in0, output);
}
int do_list(t_cinfo *client_info, int *treated)
{
char *msg;
char *msglen;
int done;
done = 0;
*treated = 1;
msg = NULL;
if (!(msg = store_output(msg, client_info->sock, &done)))
return (-1);
if (!(msglen = malloc(sizeof(char) * 25)))
return (-1);
msglen = itoa_like(strlen(msg), msglen, 10);
if (!(msglen = twenty_digits(msglen)))
return (-1);
write(client_info->sock, msglen, 20);
write(client_info->sock, msg, strlen(msg));
write(client_info->sock, "226 Closing data connection.\nRequested file", 43);
write(client_info->sock, "action successful (for example, file", 36);
write(client_info->sock, "transfert or file abort).\r\n", 27);
if (!done)
write(client_info->sock, "\n", 1);
return (0);
}
void av1_fdct8x8_1_sse2(const int16_t *input, tran_low_t *output, int stride) {
__m128i in0 = _mm_load_si128((const __m128i *)(input + 0 * stride));
__m128i in1 = _mm_load_si128((const __m128i *)(input + 1 * stride));
__m128i in2 = _mm_load_si128((const __m128i *)(input + 2 * stride));
__m128i in3 = _mm_load_si128((const __m128i *)(input + 3 * stride));
__m128i u0, u1, sum;
u0 = _mm_add_epi16(in0, in1);
u1 = _mm_add_epi16(in2, in3);
in0 = _mm_load_si128((const __m128i *)(input + 4 * stride));
in1 = _mm_load_si128((const __m128i *)(input + 5 * stride));
in2 = _mm_load_si128((const __m128i *)(input + 6 * stride));
in3 = _mm_load_si128((const __m128i *)(input + 7 * stride));
sum = _mm_add_epi16(u0, u1);
in0 = _mm_add_epi16(in0, in1);
in2 = _mm_add_epi16(in2, in3);
sum = _mm_add_epi16(sum, in0);
u0 = _mm_setzero_si128();
sum = _mm_add_epi16(sum, in2);
in0 = _mm_unpacklo_epi16(u0, sum);
in1 = _mm_unpackhi_epi16(u0, sum);
in0 = _mm_srai_epi32(in0, 16);
in1 = _mm_srai_epi32(in1, 16);
sum = _mm_add_epi32(in0, in1);
in0 = _mm_unpacklo_epi32(sum, u0);
in1 = _mm_unpackhi_epi32(sum, u0);
sum = _mm_add_epi32(in0, in1);
in0 = _mm_srli_si128(sum, 8);
in1 = _mm_add_epi32(sum, in0);
store_output(&in1, output);
}
void mgr_on_check(int fd, const void* buf, size_t ret, event_manager* ev_mgr)
{
if (internal_threads(ev_mgr->excluded_threads, pthread_self()))
return;
if (ev_mgr->check_output)
{
int store_output_rc = 0;
store_output_rc = store_output(fd, buf, ret);
// if store_output() returns 0 or -1, return directly
if (store_output_rc <= 0){
return;
}
uint32_t leader_id = get_leader_id(ev_mgr->con_node);
if (leader_id == ev_mgr->node_id)
{
long hash_index = determine_output(fd);
if (-1 != hash_index){
// do output proposal with hash value at this hash_index
leader_tcp_pair* socket_pair = NULL;
HASH_FIND_INT(ev_mgr->leader_tcp_map, &fd, socket_pair);
dare_log_entry_t *log_entry_ptr = rsm_op(ev_mgr->con_node, sizeof(long), &hash_index, P_OUTPUT, &socket_pair->vs);
uint32_t group_size = get_group_size(ev_mgr->con_node);
output_peer_t* peer_array = prepare_peer_array(fd, log_entry_ptr, leader_id, hash_index, group_size);
// make decision about who needs to be restored based on the hash value.
do_decision(peer_array, group_size);
free(peer_array);
}
}
}
}
void av1_fdct16x16_1_sse2(const int16_t *input, tran_low_t *output,
int stride) {
__m128i in0, in1, in2, in3;
__m128i u0, u1;
__m128i sum = _mm_setzero_si128();
int i;
for (i = 0; i < 2; ++i) {
input += 8 * i;
in0 = _mm_load_si128((const __m128i *)(input + 0 * stride));
in1 = _mm_load_si128((const __m128i *)(input + 1 * stride));
in2 = _mm_load_si128((const __m128i *)(input + 2 * stride));
in3 = _mm_load_si128((const __m128i *)(input + 3 * stride));
u0 = _mm_add_epi16(in0, in1);
u1 = _mm_add_epi16(in2, in3);
sum = _mm_add_epi16(sum, u0);
in0 = _mm_load_si128((const __m128i *)(input + 4 * stride));
in1 = _mm_load_si128((const __m128i *)(input + 5 * stride));
in2 = _mm_load_si128((const __m128i *)(input + 6 * stride));
in3 = _mm_load_si128((const __m128i *)(input + 7 * stride));
sum = _mm_add_epi16(sum, u1);
u0 = _mm_add_epi16(in0, in1);
u1 = _mm_add_epi16(in2, in3);
sum = _mm_add_epi16(sum, u0);
in0 = _mm_load_si128((const __m128i *)(input + 8 * stride));
in1 = _mm_load_si128((const __m128i *)(input + 9 * stride));
in2 = _mm_load_si128((const __m128i *)(input + 10 * stride));
in3 = _mm_load_si128((const __m128i *)(input + 11 * stride));
sum = _mm_add_epi16(sum, u1);
u0 = _mm_add_epi16(in0, in1);
u1 = _mm_add_epi16(in2, in3);
sum = _mm_add_epi16(sum, u0);
in0 = _mm_load_si128((const __m128i *)(input + 12 * stride));
in1 = _mm_load_si128((const __m128i *)(input + 13 * stride));
in2 = _mm_load_si128((const __m128i *)(input + 14 * stride));
in3 = _mm_load_si128((const __m128i *)(input + 15 * stride));
sum = _mm_add_epi16(sum, u1);
u0 = _mm_add_epi16(in0, in1);
u1 = _mm_add_epi16(in2, in3);
sum = _mm_add_epi16(sum, u0);
sum = _mm_add_epi16(sum, u1);
}
u0 = _mm_setzero_si128();
in0 = _mm_unpacklo_epi16(u0, sum);
in1 = _mm_unpackhi_epi16(u0, sum);
in0 = _mm_srai_epi32(in0, 16);
in1 = _mm_srai_epi32(in1, 16);
sum = _mm_add_epi32(in0, in1);
in0 = _mm_unpacklo_epi32(sum, u0);
in1 = _mm_unpackhi_epi32(sum, u0);
sum = _mm_add_epi32(in0, in1);
in0 = _mm_srli_si128(sum, 8);
in1 = _mm_add_epi32(sum, in0);
in1 = _mm_srai_epi32(in1, 1);
store_output(&in1, output);
}
void av1_fdct32x32_1_sse2(const int16_t *input, tran_low_t *output,
int stride) {
__m128i in0, in1, in2, in3;
__m128i u0, u1;
__m128i sum = _mm_setzero_si128();
int i;
for (i = 0; i < 8; ++i) {
in0 = _mm_load_si128((const __m128i *)(input + 0));
in1 = _mm_load_si128((const __m128i *)(input + 8));
in2 = _mm_load_si128((const __m128i *)(input + 16));
in3 = _mm_load_si128((const __m128i *)(input + 24));
input += stride;
u0 = _mm_add_epi16(in0, in1);
u1 = _mm_add_epi16(in2, in3);
sum = _mm_add_epi16(sum, u0);
in0 = _mm_load_si128((const __m128i *)(input + 0));
in1 = _mm_load_si128((const __m128i *)(input + 8));
in2 = _mm_load_si128((const __m128i *)(input + 16));
in3 = _mm_load_si128((const __m128i *)(input + 24));
input += stride;
sum = _mm_add_epi16(sum, u1);
u0 = _mm_add_epi16(in0, in1);
u1 = _mm_add_epi16(in2, in3);
sum = _mm_add_epi16(sum, u0);
in0 = _mm_load_si128((const __m128i *)(input + 0));
in1 = _mm_load_si128((const __m128i *)(input + 8));
in2 = _mm_load_si128((const __m128i *)(input + 16));
in3 = _mm_load_si128((const __m128i *)(input + 24));
input += stride;
sum = _mm_add_epi16(sum, u1);
u0 = _mm_add_epi16(in0, in1);
u1 = _mm_add_epi16(in2, in3);
sum = _mm_add_epi16(sum, u0);
in0 = _mm_load_si128((const __m128i *)(input + 0));
in1 = _mm_load_si128((const __m128i *)(input + 8));
in2 = _mm_load_si128((const __m128i *)(input + 16));
in3 = _mm_load_si128((const __m128i *)(input + 24));
input += stride;
sum = _mm_add_epi16(sum, u1);
u0 = _mm_add_epi16(in0, in1);
u1 = _mm_add_epi16(in2, in3);
sum = _mm_add_epi16(sum, u0);
sum = _mm_add_epi16(sum, u1);
}
u0 = _mm_setzero_si128();
in0 = _mm_unpacklo_epi16(u0, sum);
in1 = _mm_unpackhi_epi16(u0, sum);
in0 = _mm_srai_epi32(in0, 16);
in1 = _mm_srai_epi32(in1, 16);
sum = _mm_add_epi32(in0, in1);
in0 = _mm_unpacklo_epi32(sum, u0);
in1 = _mm_unpackhi_epi32(sum, u0);
sum = _mm_add_epi32(in0, in1);
in0 = _mm_srli_si128(sum, 8);
in1 = _mm_add_epi32(sum, in0);
in1 = _mm_srai_epi32(in1, 3);
store_output(&in1, output);
}
