static void intra_predict_128dc_16x16_msa(uint8_t *dst, int32_t dst_stride) {
const v16u8 out = (v16u8)__msa_ldi_b(128);
ST_UB8(out, out, out, out, out, out, out, out, dst, dst_stride);
dst += (8 * dst_stride);
ST_UB8(out, out, out, out, out, out, out, out, dst, dst_stride);
}
static void intra_predict_vert_16x16_msa(uint8_t *src, uint8_t *dst,
int32_t dst_stride)
{
v16u8 out = LD_UB(src);
ST_UB8(out, out, out, out, out, out, out, out, dst, dst_stride);
dst += (8 * dst_stride);
ST_UB8(out, out, out, out, out, out, out, out, dst, dst_stride);
}
static void intra_predict_dc_tl_16x16_msa(const uint8_t *src, uint8_t *dst,
int32_t dst_stride) {
v16u8 data, out;
v8u16 sum_h;
v4u32 sum_w;
v2u64 sum_d;
data = LD_UB(src);
sum_h = __msa_hadd_u_h(data, data);
sum_w = __msa_hadd_u_w(sum_h, sum_h);
sum_d = __msa_hadd_u_d(sum_w, sum_w);
sum_w = (v4u32)__msa_pckev_w((v4i32)sum_d, (v4i32)sum_d);
sum_d = __msa_hadd_u_d(sum_w, sum_w);
sum_w = (v4u32)__msa_srari_w((v4i32)sum_d, 4);
out = (v16u8)__msa_splati_b((v16i8)sum_w, 0);
ST_UB8(out, out, out, out, out, out, out, out, dst, dst_stride);
dst += (8 * dst_stride);
ST_UB8(out, out, out, out, out, out, out, out, dst, dst_stride);
}
static void intra_predict_dc_16x16_msa(const uint8_t *src_top,
const uint8_t *src_left, uint8_t *dst,
int32_t dst_stride) {
v16u8 top, left, out;
v8u16 sum_h, sum_top, sum_left;
v4u32 sum_w;
v2u64 sum_d;
top = LD_UB(src_top);
left = LD_UB(src_left);
HADD_UB2_UH(top, left, sum_top, sum_left);
sum_h = sum_top + sum_left;
sum_w = __msa_hadd_u_w(sum_h, sum_h);
sum_d = __msa_hadd_u_d(sum_w, sum_w);
sum_w = (v4u32)__msa_pckev_w((v4i32)sum_d, (v4i32)sum_d);
sum_d = __msa_hadd_u_d(sum_w, sum_w);
sum_w = (v4u32)__msa_srari_w((v4i32)sum_d, 5);
out = (v16u8)__msa_splati_b((v16i8)sum_w, 0);
ST_UB8(out, out, out, out, out, out, out, out, dst, dst_stride);
dst += (8 * dst_stride);
ST_UB8(out, out, out, out, out, out, out, out, dst, dst_stride);
}
