@@ -1265,54 +1265,51 @@ static void hevc_sao_edge_filter_0degree_4width_msa(uint8_t *dst,
int16_t *sao_offset_val,
int32_t height) {
- int32_t h_cnt;
uint32_t dst_val0, dst_val1;
- v8i16 edge_idx = { 1, 2, 0, 3, 4, 0, 0, 0 };
+ v16u8 cmp_minus10, diff_minus10, diff_minus11, src_minus10, src_minus11;
+ v16i8 edge_idx = { 1, 2, 0, 3, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
+ v16i8 sao_offset = LD_SB(sao_offset_val);
+ v16i8 src_plus10, offset, src0, dst0;
v16u8 const1 = (v16u8) __msa_ldi_b(1);
- v16u8 cmp_minus10, diff_minus10, cmp_minus11, diff_minus11;
- v16u8 src_minus10, src_minus11;
v16i8 zero = { 0 };
- v16i8 src_zero0, src_zero1, src_plus10, src_plus11, dst0;
- v8i16 offset_mask0, offset_mask1;
- v8i16 sao_offset, src00, src01;
- sao_offset = LD_SH(sao_offset_val);
+ sao_offset = __msa_pckev_b(sao_offset, sao_offset);
src -= 1;
- for (h_cnt = (height >> 1); h_cnt--;) {
- LD_UB2(src, src_stride, src_minus10, src_minus11);
+ /* load in advance */
+ LD_UB2(src, src_stride, src_minus10, src_minus11);
+
+ for (height -= 2; height; height -= 2) {
src += (2 * src_stride);
- SLDI_B2_0_SB(src_minus10, src_minus11, src_zero0, src_zero1, 1);
- SLDI_B2_0_SB(src_minus10, src_minus11, src_plus10, src_plus11, 2);
- ILVR_B2_UB(src_plus10, src_minus10, src_plus11, src_minus11,
- src_minus10, src_minus11);
- ILVR_B2_SB(src_zero0, src_zero0, src_zero1, src_zero1, src_zero0,
- src_zero1);
+ src_minus10 = (v16u8) __msa_pckev_d((v2i64) src_minus11,
+ (v2i64) src_minus10);
- cmp_minus10 = ((v16u8) src_zero0 == src_minus10);
+ src0 = (v16i8) __msa_sldi_b(zero, (v16i8) src_minus10, 1);
+ src_plus10 = (v16i8) __msa_sldi_b(zero, (v16i8) src_minus10,
+ 2);
+
+ cmp_minus10 = ((v16u8) src0 == src_minus10);
diff_minus10 = __msa_nor_v(cmp_minus10, cmp_minus10);
- cmp_minus10 = (src_minus10 < (v16u8) src_zero0);
+ cmp_minus10 = (src_minus10 < (v16u8) src0);
diff_minus10 = __msa_bmnz_v(diff_minus10, const1, cmp_minus10);
- cmp_minus11 = ((v16u8) src_zero1 == src_minus11);
- diff_minus11 = __msa_nor_v(cmp_minus11, cmp_minus11);
- cmp_minus11 = (src_minus11 < (v16u8) src_zero1);
- diff_minus11 = __msa_bmnz_v(diff_minus11, const1, cmp_minus11);
+ cmp_minus10 = ((v16u8) src0 == (v16u8) src_plus10);
+ diff_minus11 = __msa_nor_v(cmp_minus10, cmp_minus10);
+ cmp_minus10 = ((v16u8) src_plus10 < (v16u8) src0);
+ diff_minus11 = __msa_bmnz_v(diff_minus11, const1, cmp_minus10);
- offset_mask0 = (v8i16) (__msa_hadd_u_h(diff_minus10, diff_minus10) + 2);
- offset_mask1 = (v8i16) (__msa_hadd_u_h(diff_minus11, diff_minus11) + 2);
+ offset = (v16i8) diff_minus10 + (v16i8) diff_minus11 + 2;
- VSHF_H2_SH(edge_idx, edge_idx, sao_offset, sao_offset, offset_mask0,
- offset_mask0, offset_mask0, offset_mask0);
- VSHF_H2_SH(edge_idx, edge_idx, sao_offset, sao_offset, offset_mask1,
- offset_mask1, offset_mask1, offset_mask1);
- ILVEV_B2_SH(src_zero0, zero, src_zero1, zero, src00, src01);
- ADD2(offset_mask0, src00, offset_mask1, src01, offset_mask0,
- offset_mask1);
- CLIP_SH2_0_255(offset_mask0, offset_mask1);
+ /* load in advance */
+ LD_UB2(src, src_stride, src_minus10, src_minus11);
+
+ VSHF_B2_SB(edge_idx, edge_idx, sao_offset, sao_offset, offset, offset,
+ offset, offset);
+
+ src0 = (v16i8) __msa_xori_b((v16u8) src0, 128);
+ dst0 = __msa_adds_s_b(src0, offset);
+ dst0 = (v16i8) __msa_xori_b((v16u8) dst0, 128);
- dst0 = __msa_pckev_b((v16i8) offset_mask1, (v16i8) offset_mask0);
dst_val0 = __msa_copy_u_w((v4i32) dst0, 0);
dst_val1 = __msa_copy_u_w((v4i32) dst0, 2);
SW(dst_val0, dst);
@@ -1320,6 +1317,37 @@ static void hevc_sao_edge_filter_0degree_4width_msa(uint8_t *dst,