From 6ca821a3e7751b6b24e63ec1b531e3841d4d7bda Mon Sep 17 00:00:00 2001 From: Kaustubh Raste Date: Wed, 11 Oct 2017 14:20:28 +0530 Subject: [PATCH] avcodec/mips: Improve hevc uni horiz mc msa functions Update macros to remove adds. Signed-off-by: Kaustubh Raste Signed-off-by: Michael Niedermayer --- libavcodec/mips/hevc_mc_uni_msa.c | 612 ++++++++++++++++-------------- 1 file changed, 323 insertions(+), 289 deletions(-) diff --git a/libavcodec/mips/hevc_mc_uni_msa.c b/libavcodec/mips/hevc_mc_uni_msa.c index eead591ff4..3a6c5b0d48 100644 --- a/libavcodec/mips/hevc_mc_uni_msa.c +++ b/libavcodec/mips/hevc_mc_uni_msa.c @@ -22,6 +22,85 @@ #include "libavcodec/mips/hevcdsp_mips.h" #include "libavcodec/mips/hevc_macros_msa.h" +static const uint8_t ff_hevc_mask_arr[16 * 3] __attribute__((aligned(0x40))) = { + /* 8 width cases */ + 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, + /* 4 width cases */ + 0, 1, 1, 2, 2, 3, 3, 4, 16, 17, 17, 18, 18, 19, 19, 20, + /* 4 width cases */ + 8, 9, 9, 10, 10, 11, 11, 12, 24, 25, 25, 26, 26, 27, 27, 28 +}; + +#define HORIZ_8TAP_4WID_4VECS_FILT(src0, src1, src2, src3, \ + mask0, mask1, mask2, mask3, \ + filt0, filt1, filt2, filt3, \ + out0, out1) \ +{ \ + v16i8 vec0_m, vec1_m, vec2_m, vec3_m, vec4_m, vec5_m, vec6_m, vec7_m; \ + \ + VSHF_B2_SB(src0, src1, src2, src3, mask0, mask0, vec0_m, vec1_m); \ + DOTP_SB2_SH(vec0_m, vec1_m, filt0, filt0, out0, out1); \ + VSHF_B2_SB(src0, src1, src2, src3, mask1, mask1, vec2_m, vec3_m); \ + DPADD_SB2_SH(vec2_m, vec3_m, filt1, filt1, out0, out1); \ + VSHF_B2_SB(src0, src1, src2, src3, mask2, mask2, vec4_m, vec5_m); \ + DPADD_SB2_SH(vec4_m, vec5_m, filt2, filt2, out0, out1); \ + VSHF_B2_SB(src0, src1, src2, src3, mask3, mask3, vec6_m, vec7_m); \ + DPADD_SB2_SH(vec6_m, vec7_m, filt3, filt3, out0, out1); \ +} + +#define HORIZ_8TAP_8WID_4VECS_FILT(src0, src1, src2, src3, \ + mask0, mask1, mask2, mask3, \ + filt0, filt1, filt2, filt3, \ + out0, out1, out2, out3) \ +{ \ + v16i8 vec0_m, vec1_m, vec2_m, vec3_m, vec4_m, vec5_m, vec6_m, vec7_m; \ + \ + VSHF_B2_SB(src0, src0, src1, src1, mask0, mask0, vec0_m, vec1_m); \ + VSHF_B2_SB(src2, src2, src3, src3, mask0, mask0, vec2_m, vec3_m); \ + DOTP_SB4_SH(vec0_m, vec1_m, vec2_m, vec3_m, filt0, filt0, filt0, filt0, \ + out0, out1, out2, out3); \ + VSHF_B2_SB(src0, src0, src1, src1, mask2, mask2, vec0_m, vec1_m); \ + VSHF_B2_SB(src2, src2, src3, src3, mask2, mask2, vec2_m, vec3_m); \ + DPADD_SB4_SH(vec0_m, vec1_m, vec2_m, vec3_m, filt2, filt2, filt2, filt2, \ + out0, out1, out2, out3); \ + VSHF_B2_SB(src0, src0, src1, src1, mask1, mask1, vec4_m, vec5_m); \ + VSHF_B2_SB(src2, src2, src3, src3, mask1, mask1, vec6_m, vec7_m); \ + DPADD_SB4_SH(vec4_m, vec5_m, vec6_m, vec7_m, filt1, filt1, filt1, filt1, \ + out0, out1, out2, out3); \ + VSHF_B2_SB(src0, src0, src1, src1, mask3, mask3, vec4_m, vec5_m); \ + VSHF_B2_SB(src2, src2, src3, src3, mask3, mask3, vec6_m, vec7_m); \ + DPADD_SB4_SH(vec4_m, vec5_m, vec6_m, vec7_m, filt3, filt3, filt3, filt3, \ + out0, out1, out2, out3); \ +} + +#define HORIZ_4TAP_4WID_4VECS_FILT(src0, src1, src2, src3, \ + mask0, mask1, filt0, filt1, \ + out0, out1) \ +{ \ + v16i8 vec0_m, vec1_m, vec2_m, vec3_m; \ + \ + VSHF_B2_SB(src0, src1, src2, src3, mask0, mask0, vec0_m, vec1_m); \ + DOTP_SB2_SH(vec0_m, vec1_m, filt0, filt0, out0, out1); \ + VSHF_B2_SB(src0, src1, src2, src3, mask1, mask1, vec2_m, vec3_m); \ + DPADD_SB2_SH(vec2_m, vec3_m, filt1, filt1, out0, out1); \ +} + +#define HORIZ_4TAP_8WID_4VECS_FILT(src0, src1, src2, src3, \ + mask0, mask1, filt0, filt1, \ + out0, out1, out2, out3) \ +{ \ + v16i8 vec0_m, vec1_m, vec2_m, vec3_m; \ + \ + VSHF_B2_SB(src0, src0, src1, src1, mask0, mask0, vec0_m, vec1_m); \ + VSHF_B2_SB(src2, src2, src3, src3, mask0, mask0, vec2_m, vec3_m); \ + DOTP_SB4_SH(vec0_m, vec1_m, vec2_m, vec3_m, filt0, filt0, filt0, filt0, \ + out0, out1, out2, out3); \ + VSHF_B2_SB(src0, src0, src1, src1, mask1, mask1, vec0_m, vec1_m); \ + VSHF_B2_SB(src2, src2, src3, src3, mask1, mask1, vec2_m, vec3_m); \ + DPADD_SB4_SH(vec0_m, vec1_m, vec2_m, vec3_m, filt1, filt1, filt1, filt1, \ + out0, out1, out2, out3); \ +} + static void copy_width8_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, int32_t height) @@ -227,53 +306,6 @@ static const uint8_t mc_filt_mask_arr[16 * 3] = { tmp0; \ } ) -#define HORIZ_8TAP_4WID_4VECS_FILT(src0, src1, src2, src3, \ - mask0, mask1, mask2, mask3, \ - filt0, filt1, filt2, filt3, \ - out0, out1) \ -{ \ - v16i8 vec0_m, vec1_m, vec2_m, vec3_m, vec4_m, vec5_m, vec6_m, vec7_m; \ - v8i16 res0_m, res1_m, res2_m, res3_m; \ - \ - VSHF_B2_SB(src0, src1, src2, src3, mask0, mask0, vec0_m, vec1_m); \ - DOTP_SB2_SH(vec0_m, vec1_m, filt0, filt0, res0_m, res1_m); \ - VSHF_B2_SB(src0, src1, src2, src3, mask1, mask1, vec2_m, vec3_m); \ - DPADD_SB2_SH(vec2_m, vec3_m, filt1, filt1, res0_m, res1_m); \ - VSHF_B2_SB(src0, src1, src2, src3, mask2, mask2, vec4_m, vec5_m); \ - DOTP_SB2_SH(vec4_m, vec5_m, filt2, filt2, res2_m, res3_m); \ - VSHF_B2_SB(src0, src1, src2, src3, mask3, mask3, vec6_m, vec7_m); \ - DPADD_SB2_SH(vec6_m, vec7_m, filt3, filt3, res2_m, res3_m); \ - ADDS_SH2_SH(res0_m, res2_m, res1_m, res3_m, out0, out1); \ -} - -#define HORIZ_8TAP_8WID_4VECS_FILT(src0, src1, src2, src3, \ - mask0, mask1, mask2, mask3, \ - filt0, filt1, filt2, filt3, \ - out0, out1, out2, out3) \ -{ \ - v16i8 vec0_m, vec1_m, vec2_m, vec3_m, vec4_m, vec5_m, vec6_m, vec7_m; \ - v8i16 res0_m, res1_m, res2_m, res3_m, res4_m, res5_m, res6_m, res7_m; \ - \ - VSHF_B2_SB(src0, src0, src1, src1, mask0, mask0, vec0_m, vec1_m); \ - VSHF_B2_SB(src2, src2, src3, src3, mask0, mask0, vec2_m, vec3_m); \ - DOTP_SB4_SH(vec0_m, vec1_m, vec2_m, vec3_m, filt0, filt0, filt0, filt0, \ - res0_m, res1_m, res2_m, res3_m); \ - VSHF_B2_SB(src0, src0, src1, src1, mask2, mask2, vec0_m, vec1_m); \ - VSHF_B2_SB(src2, src2, src3, src3, mask2, mask2, vec2_m, vec3_m); \ - DOTP_SB4_SH(vec0_m, vec1_m, vec2_m, vec3_m, filt2, filt2, filt2, filt2, \ - res4_m, res5_m, res6_m, res7_m); \ - VSHF_B2_SB(src0, src0, src1, src1, mask1, mask1, vec4_m, vec5_m); \ - VSHF_B2_SB(src2, src2, src3, src3, mask1, mask1, vec6_m, vec7_m); \ - DPADD_SB4_SH(vec4_m, vec5_m, vec6_m, vec7_m, filt1, filt1, filt1, filt1, \ - res0_m, res1_m, res2_m, res3_m); \ - VSHF_B2_SB(src0, src0, src1, src1, mask3, mask3, vec4_m, vec5_m); \ - VSHF_B2_SB(src2, src2, src3, src3, mask3, mask3, vec6_m, vec7_m); \ - DPADD_SB4_SH(vec4_m, vec5_m, vec6_m, vec7_m, filt3, filt3, filt3, filt3, \ - res4_m, res5_m, res6_m, res7_m); \ - ADDS_SH4_SH(res0_m, res4_m, res1_m, res5_m, res2_m, res6_m, res3_m, \ - res7_m, out0, out1, out2, out3); \ -} - #define FILT_4TAP_DPADD_S_H(vec0, vec1, filt0, filt1) \ ( { \ v8i16 tmp0; \ @@ -284,34 +316,6 @@ static const uint8_t mc_filt_mask_arr[16 * 3] = { tmp0; \ } ) -#define HORIZ_4TAP_4WID_4VECS_FILT(src0, src1, src2, src3, \ - mask0, mask1, filt0, filt1, \ - out0, out1) \ -{ \ - v16i8 vec0_m, vec1_m, vec2_m, vec3_m; \ - \ - VSHF_B2_SB(src0, src1, src2, src3, mask0, mask0, vec0_m, vec1_m); \ - DOTP_SB2_SH(vec0_m, vec1_m, filt0, filt0, out0, out1); \ - VSHF_B2_SB(src0, src1, src2, src3, mask1, mask1, vec2_m, vec3_m); \ - DPADD_SB2_SH(vec2_m, vec3_m, filt1, filt1, out0, out1); \ -} - -#define HORIZ_4TAP_8WID_4VECS_FILT(src0, src1, src2, src3, \ - mask0, mask1, filt0, filt1, \ - out0, out1, out2, out3) \ -{ \ - v16i8 vec0_m, vec1_m, vec2_m, vec3_m; \ - \ - VSHF_B2_SB(src0, src0, src1, src1, mask0, mask0, vec0_m, vec1_m); \ - VSHF_B2_SB(src2, src2, src3, src3, mask0, mask0, vec2_m, vec3_m); \ - DOTP_SB4_SH(vec0_m, vec1_m, vec2_m, vec3_m, filt0, filt0, filt0, filt0, \ - out0, out1, out2, out3); \ - VSHF_B2_SB(src0, src0, src1, src1, mask1, mask1, vec0_m, vec1_m); \ - VSHF_B2_SB(src2, src2, src3, src3, mask1, mask1, vec2_m, vec3_m); \ - DPADD_SB4_SH(vec0_m, vec1_m, vec2_m, vec3_m, filt1, filt1, filt1, filt1, \ - out0, out1, out2, out3); \ -} - static void common_hz_8t_4x4_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, const int8_t *filter) @@ -320,7 +324,7 @@ static void common_hz_8t_4x4_msa(uint8_t *src, int32_t src_stride, v16i8 src0, src1, src2, src3, filt0, filt1, filt2, filt3; v8i16 filt, out0, out1; - mask0 = LD_UB(&mc_filt_mask_arr[16]); + mask0 = LD_UB(&ff_hevc_mask_arr[16]); src -= 3; /* rearranging filter */ @@ -350,7 +354,7 @@ static void common_hz_8t_4x8_msa(uint8_t *src, int32_t src_stride, v16u8 mask0, mask1, mask2, mask3, out; v8i16 filt, out0, out1, out2, out3; - mask0 = LD_UB(&mc_filt_mask_arr[16]); + mask0 = LD_UB(&ff_hevc_mask_arr[16]); src -= 3; /* rearranging filter */ @@ -387,7 +391,7 @@ static void common_hz_8t_4x16_msa(uint8_t *src, int32_t src_stride, v16i8 src0, src1, src2, src3, filt0, filt1, filt2, filt3; v8i16 filt, out0, out1, out2, out3; - mask0 = LD_UB(&mc_filt_mask_arr[16]); + mask0 = LD_UB(&ff_hevc_mask_arr[16]); src -= 3; /* rearranging filter */ @@ -450,47 +454,17 @@ static void common_hz_8t_4w_msa(uint8_t *src, int32_t src_stride, } } -static void common_hz_8t_8x4_msa(uint8_t *src, int32_t src_stride, - uint8_t *dst, int32_t dst_stride, - const int8_t *filter) -{ - v16i8 src0, src1, src2, src3, filt0, filt1, filt2, filt3; - v16u8 mask0, mask1, mask2, mask3, tmp0, tmp1; - v8i16 filt, out0, out1, out2, out3; - - mask0 = LD_UB(&mc_filt_mask_arr[0]); - src -= 3; - - /* rearranging filter */ - filt = LD_SH(filter); - SPLATI_H4_SB(filt, 0, 1, 2, 3, filt0, filt1, filt2, filt3); - - mask1 = mask0 + 2; - mask2 = mask0 + 4; - mask3 = mask0 + 6; - - LD_SB4(src, src_stride, src0, src1, src2, src3); - XORI_B4_128_SB(src0, src1, src2, src3); - HORIZ_8TAP_8WID_4VECS_FILT(src0, src1, src2, src3, mask0, mask1, mask2, - mask3, filt0, filt1, filt2, filt3, out0, out1, - out2, out3); - SRARI_H4_SH(out0, out1, out2, out3, 6); - SAT_SH4_SH(out0, out1, out2, out3, 7); - tmp0 = PCKEV_XORI128_UB(out0, out1); - tmp1 = PCKEV_XORI128_UB(out2, out3); - ST8x4_UB(tmp0, tmp1, dst, dst_stride); -} - -static void common_hz_8t_8x8mult_msa(uint8_t *src, int32_t src_stride, - uint8_t *dst, int32_t dst_stride, - const int8_t *filter, int32_t height) +static void common_hz_8t_8w_msa(uint8_t *src, int32_t src_stride, + uint8_t *dst, int32_t dst_stride, + const int8_t *filter, int32_t height) { uint32_t loop_cnt; v16i8 src0, src1, src2, src3, filt0, filt1, filt2, filt3; v16u8 mask0, mask1, mask2, mask3, tmp0, tmp1; + v16i8 vec0_m, vec1_m, vec2_m, vec3_m, vec4_m, vec5_m, vec6_m, vec7_m; v8i16 filt, out0, out1, out2, out3; - mask0 = LD_UB(&mc_filt_mask_arr[0]); + mask0 = LD_UB(&ff_hevc_mask_arr[0]); src -= 3; /* rearranging filter */ @@ -505,9 +479,24 @@ static void common_hz_8t_8x8mult_msa(uint8_t *src, int32_t src_stride, LD_SB4(src, src_stride, src0, src1, src2, src3); XORI_B4_128_SB(src0, src1, src2, src3); src += (4 * src_stride); - HORIZ_8TAP_8WID_4VECS_FILT(src0, src1, src2, src3, mask0, mask1, mask2, - mask3, filt0, filt1, filt2, filt3, out0, - out1, out2, out3); + + VSHF_B2_SB(src0, src0, src1, src1, mask0, mask0, vec0_m, vec1_m); + VSHF_B2_SB(src2, src2, src3, src3, mask0, mask0, vec2_m, vec3_m); + DOTP_SB4_SH(vec0_m, vec1_m, vec2_m, vec3_m, filt0, filt0, filt0, filt0, + out0, out1, out2, out3); + VSHF_B2_SB(src0, src0, src1, src1, mask2, mask2, vec0_m, vec1_m); + VSHF_B2_SB(src2, src2, src3, src3, mask2, mask2, vec2_m, vec3_m); + DPADD_SB4_SH(vec0_m, vec1_m, vec2_m, vec3_m, filt2, filt2, filt2, filt2, + out0, out1, out2, out3); + VSHF_B2_SB(src0, src0, src1, src1, mask1, mask1, vec4_m, vec5_m); + VSHF_B2_SB(src2, src2, src3, src3, mask1, mask1, vec6_m, vec7_m); + DPADD_SB4_SH(vec4_m, vec5_m, vec6_m, vec7_m, filt1, filt1, filt1, filt1, + out0, out1, out2, out3); + VSHF_B2_SB(src0, src0, src1, src1, mask3, mask3, vec4_m, vec5_m); + VSHF_B2_SB(src2, src2, src3, src3, mask3, mask3, vec6_m, vec7_m); + DPADD_SB4_SH(vec4_m, vec5_m, vec6_m, vec7_m, filt3, filt3, filt3, filt3, + out0, out1, out2, out3); + SRARI_H4_SH(out0, out1, out2, out3, 6); SAT_SH4_SH(out0, out1, out2, out3, 7); tmp0 = PCKEV_XORI128_UB(out0, out1); @@ -517,36 +506,22 @@ static void common_hz_8t_8x8mult_msa(uint8_t *src, int32_t src_stride, } } -static void common_hz_8t_8w_msa(uint8_t *src, int32_t src_stride, - uint8_t *dst, int32_t dst_stride, - const int8_t *filter, int32_t height) -{ - if (4 == height) { - common_hz_8t_8x4_msa(src, src_stride, dst, dst_stride, filter); - } else { - common_hz_8t_8x8mult_msa(src, src_stride, dst, dst_stride, filter, - height); - } -} - static void common_hz_8t_12w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, const int8_t *filter, int32_t height) { - uint8_t *src1_ptr, *dst1; uint32_t loop_cnt; - v16i8 src0, src1, src2, src3, filt0, filt1, filt2, filt3; - v8i16 filt, out0, out1, out2, out3; - v16u8 mask0, mask1, mask2, mask3, mask4, mask5, mask6, mask00, tmp0, tmp1; - - mask00 = LD_UB(&mc_filt_mask_arr[0]); - mask0 = LD_UB(&mc_filt_mask_arr[16]); + v16u8 mask0, mask1, mask2, mask3, mask4, mask5, mask6, mask00; + v16u8 tmp0, tmp1, tmp2; + v16i8 src0, src1, src2, src3, src4, src5, src6, src7; + v16i8 vec0, vec1, vec2, vec3, vec4, vec5, vec6, vec7; + v16i8 filt0, filt1, filt2, filt3; + v8i16 filt, out0, out1, out2, out3, out4, out5; - src1_ptr = src - 3; - dst1 = dst; + mask00 = LD_UB(&ff_hevc_mask_arr[0]); + mask0 = LD_UB(&ff_hevc_mask_arr[16]); - dst = dst1 + 8; - src = src1_ptr + 8; + src = src - 3; /* rearranging filter */ filt = LD_SH(filter); @@ -559,32 +534,53 @@ static void common_hz_8t_12w_msa(uint8_t *src, int32_t src_stride, mask5 = mask0 + 4; mask6 = mask0 + 6; - for (loop_cnt = (height >> 2); loop_cnt--;) { + for (loop_cnt = 4; loop_cnt--;) { /* 8 width */ - LD_SB4(src1_ptr, src_stride, src0, src1, src2, src3); + LD_SB4(src, src_stride, src0, src1, src2, src3); + /* 4 width */ + LD_SB4(src + 8, src_stride, src4, src5, src6, src7); + XORI_B4_128_SB(src0, src1, src2, src3); - src1_ptr += (4 * src_stride); - HORIZ_8TAP_8WID_4VECS_FILT(src0, src1, src2, src3, mask00, mask1, mask2, - mask3, filt0, filt1, filt2, filt3, out0, - out1, out2, out3); + XORI_B4_128_SB(src4, src5, src6, src7); + src += (4 * src_stride); + + VSHF_B2_SB(src0, src0, src1, src1, mask00, mask00, vec0, vec1); + VSHF_B2_SB(src2, src2, src3, src3, mask00, mask00, vec2, vec3); + DOTP_SB4_SH(vec0, vec1, vec2, vec3, filt0, filt0, filt0, filt0, out0, + out1, out2, out3); + VSHF_B2_SB(src0, src0, src1, src1, mask2, mask2, vec0, vec1); + VSHF_B2_SB(src2, src2, src3, src3, mask2, mask2, vec2, vec3); + DPADD_SB4_SH(vec0, vec1, vec2, vec3, filt2, filt2, filt2, filt2, out0, + out1, out2, out3); + VSHF_B2_SB(src0, src0, src1, src1, mask1, mask1, vec4, vec5); + VSHF_B2_SB(src2, src2, src3, src3, mask1, mask1, vec6, vec7); + DPADD_SB4_SH(vec4, vec5, vec6, vec7, filt1, filt1, filt1, filt1, out0, + out1, out2, out3); + VSHF_B2_SB(src0, src0, src1, src1, mask3, mask3, vec4, vec5); + VSHF_B2_SB(src2, src2, src3, src3, mask3, mask3, vec6, vec7); + DPADD_SB4_SH(vec4, vec5, vec6, vec7, filt3, filt3, filt3, filt3, out0, + out1, out2, out3); + + /* 4 width */ + VSHF_B2_SB(src4, src5, src6, src7, mask0, mask0, vec0, vec1); + DOTP_SB2_SH(vec0, vec1, filt0, filt0, out4, out5); + VSHF_B2_SB(src4, src5, src6, src7, mask4, mask4, vec2, vec3); + DPADD_SB2_SH(vec2, vec3, filt1, filt1, out4, out5); + VSHF_B2_SB(src4, src5, src6, src7, mask5, mask5, vec4, vec5); + DPADD_SB2_SH(vec4, vec5, filt2, filt2, out4, out5); + VSHF_B2_SB(src4, src5, src6, src7, mask6, mask6, vec6, vec7); + DPADD_SB2_SH(vec6, vec7, filt3, filt3, out4, out5); + SRARI_H4_SH(out0, out1, out2, out3, 6); + SRARI_H2_SH(out4, out5, 6); SAT_SH4_SH(out0, out1, out2, out3, 7); + SAT_SH2_SH(out4, out5, 7); tmp0 = PCKEV_XORI128_UB(out0, out1); tmp1 = PCKEV_XORI128_UB(out2, out3); - ST8x4_UB(tmp0, tmp1, dst1, dst_stride); - dst1 += (4 * dst_stride); + tmp2 = PCKEV_XORI128_UB(out4, out5); - /* 4 width */ - LD_SB4(src, src_stride, src0, src1, src2, src3); - XORI_B4_128_SB(src0, src1, src2, src3); - src += (4 * src_stride); - HORIZ_8TAP_4WID_4VECS_FILT(src0, src1, src2, src3, mask0, mask4, mask5, - mask6, filt0, filt1, filt2, filt3, out0, - out1); - SRARI_H2_SH(out0, out1, 6); - SAT_SH2_SH(out0, out1, 7); - tmp0 = PCKEV_XORI128_UB(out0, out1); - ST4x4_UB(tmp0, tmp0, 0, 1, 2, 3, dst, dst_stride); + ST8x4_UB(tmp0, tmp1, dst, dst_stride); + ST4x4_UB(tmp2, tmp2, 0, 1, 2, 3, dst + 8, dst_stride); dst += (4 * dst_stride); } } @@ -594,11 +590,12 @@ static void common_hz_8t_16w_msa(uint8_t *src, int32_t src_stride, const int8_t *filter, int32_t height) { uint32_t loop_cnt; - v16i8 src0, src1, src2, src3, filt0, filt1, filt2, filt3; v16u8 mask0, mask1, mask2, mask3, out; + v16i8 src0, src1, src2, src3, src4, src5, src6, src7; + v16i8 filt0, filt1, filt2, filt3; v8i16 filt, out0, out1, out2, out3; - mask0 = LD_UB(&mc_filt_mask_arr[0]); + mask0 = LD_UB(&ff_hevc_mask_arr[0]); src -= 3; /* rearranging filter */ @@ -609,11 +606,17 @@ static void common_hz_8t_16w_msa(uint8_t *src, int32_t src_stride, mask2 = mask0 + 4; mask3 = mask0 + 6; - for (loop_cnt = (height >> 1); loop_cnt--;) { + for (loop_cnt = (height >> 2); loop_cnt--;) { LD_SB2(src, src_stride, src0, src2); LD_SB2(src + 8, src_stride, src1, src3); - XORI_B4_128_SB(src0, src1, src2, src3); src += (2 * src_stride); + + LD_SB2(src, src_stride, src4, src6); + LD_SB2(src + 8, src_stride, src5, src7); + src += (2 * src_stride); + + XORI_B4_128_SB(src0, src1, src2, src3); + XORI_B4_128_SB(src4, src5, src6, src7); HORIZ_8TAP_8WID_4VECS_FILT(src0, src1, src2, src3, mask0, mask1, mask2, mask3, filt0, filt1, filt2, filt3, out0, out1, out2, out3); @@ -625,6 +628,18 @@ static void common_hz_8t_16w_msa(uint8_t *src, int32_t src_stride, out = PCKEV_XORI128_UB(out2, out3); ST_UB(out, dst); dst += dst_stride; + + HORIZ_8TAP_8WID_4VECS_FILT(src4, src5, src6, src7, mask0, mask1, mask2, + mask3, filt0, filt1, filt2, filt3, out0, + out1, out2, out3); + SRARI_H4_SH(out0, out1, out2, out3, 6); + SAT_SH4_SH(out0, out1, out2, out3, 7); + out = PCKEV_XORI128_UB(out0, out1); + ST_UB(out, dst); + dst += dst_stride; + out = PCKEV_XORI128_UB(out2, out3); + ST_UB(out, dst); + dst += dst_stride; } } @@ -637,10 +652,9 @@ static void common_hz_8t_24w_msa(uint8_t *src, int32_t src_stride, v16u8 mask0, mask1, mask2, mask3, mask4, mask5, mask6, mask7, out; v16i8 vec0, vec1, vec2, vec3, vec4, vec5, vec6, vec7, vec8, vec9, vec10; v16i8 vec11; - v8i16 out0, out1, out2, out3, out4, out5, out6, out7, out8, out9, out10; - v8i16 out11, filt; + v8i16 out0, out1, out2, out3, out8, out9, filt; - mask0 = LD_UB(&mc_filt_mask_arr[0]); + mask0 = LD_UB(&ff_hevc_mask_arr[0]); src -= 3; /* rearranging filter */ @@ -655,7 +669,7 @@ static void common_hz_8t_24w_msa(uint8_t *src, int32_t src_stride, mask6 = mask0 + 12; mask7 = mask0 + 14; - for (loop_cnt = (height >> 1); loop_cnt--;) { + for (loop_cnt = 16; loop_cnt--;) { LD_SB2(src, src_stride, src0, src2); LD_SB2(src + 16, src_stride, src1, src3); XORI_B4_128_SB(src0, src1, src2, src3); @@ -669,9 +683,9 @@ static void common_hz_8t_24w_msa(uint8_t *src, int32_t src_stride, VSHF_B2_SB(src0, src0, src1, src1, mask2, mask2, vec0, vec8); VSHF_B2_SB(src2, src2, src3, src3, mask2, mask2, vec2, vec9); VSHF_B2_SB(src0, src1, src2, src3, mask6, mask6, vec1, vec3); - DOTP_SB4_SH(vec0, vec8, vec2, vec9, filt2, filt2, filt2, filt2, out4, - out10, out6, out11); - DOTP_SB2_SH(vec1, vec3, filt2, filt2, out5, out7); + DPADD_SB4_SH(vec0, vec8, vec2, vec9, filt2, filt2, filt2, filt2, + out0, out8, out2, out9); + DPADD_SB2_SH(vec1, vec3, filt2, filt2, out1, out3); VSHF_B2_SB(src0, src0, src1, src1, mask1, mask1, vec4, vec10); VSHF_B2_SB(src2, src2, src3, src3, mask1, mask1, vec6, vec11); VSHF_B2_SB(src0, src1, src2, src3, mask5, mask5, vec5, vec7); @@ -682,11 +696,8 @@ static void common_hz_8t_24w_msa(uint8_t *src, int32_t src_stride, VSHF_B2_SB(src2, src2, src3, src3, mask3, mask3, vec6, vec11); VSHF_B2_SB(src0, src1, src2, src3, mask7, mask7, vec5, vec7); DPADD_SB4_SH(vec4, vec10, vec6, vec11, filt3, filt3, filt3, filt3, - out4, out10, out6, out11); - DPADD_SB2_SH(vec5, vec7, filt3, filt3, out5, out7); - ADDS_SH4_SH(out0, out4, out8, out10, out2, out6, out9, out11, out0, - out8, out2, out9); - ADDS_SH2_SH(out1, out5, out3, out7, out1, out3); + out0, out8, out2, out9); + DPADD_SB2_SH(vec5, vec7, filt3, filt3, out1, out3); SRARI_H4_SH(out0, out8, out2, out9, 6); SRARI_H2_SH(out1, out3, 6); SAT_SH4_SH(out0, out8, out2, out9, 7); @@ -707,11 +718,12 @@ static void common_hz_8t_32w_msa(uint8_t *src, int32_t src_stride, const int8_t *filter, int32_t height) { uint32_t loop_cnt; - v16i8 src0, src1, src2, src3, filt0, filt1, filt2, filt3; v16u8 mask0, mask1, mask2, mask3, out; + v16i8 src0, src1, src2, src3, src4, src5, src6, src7; + v16i8 filt0, filt1, filt2, filt3; v8i16 filt, out0, out1, out2, out3; - mask0 = LD_UB(&mc_filt_mask_arr[0]); + mask0 = LD_UB(&ff_hevc_mask_arr[0]); src -= 3; /* rearranging filter */ @@ -724,31 +736,32 @@ static void common_hz_8t_32w_msa(uint8_t *src, int32_t src_stride, for (loop_cnt = (height >> 1); loop_cnt--;) { src0 = LD_SB(src); + src1 = LD_SB(src + 8); src2 = LD_SB(src + 16); src3 = LD_SB(src + 24); - src1 = __msa_sldi_b(src2, src0, 8); src += src_stride; XORI_B4_128_SB(src0, src1, src2, src3); + + src4 = LD_SB(src); + src5 = LD_SB(src + 8); + src6 = LD_SB(src + 16); + src7 = LD_SB(src + 24); + src += src_stride; + XORI_B4_128_SB(src4, src5, src6, src7); + HORIZ_8TAP_8WID_4VECS_FILT(src0, src1, src2, src3, mask0, mask1, mask2, mask3, filt0, filt1, filt2, filt3, out0, out1, out2, out3); SRARI_H4_SH(out0, out1, out2, out3, 6); SAT_SH4_SH(out0, out1, out2, out3, 7); - src0 = LD_SB(src); - src2 = LD_SB(src + 16); - src3 = LD_SB(src + 24); - src1 = __msa_sldi_b(src2, src0, 8); - src += src_stride; - out = PCKEV_XORI128_UB(out0, out1); ST_UB(out, dst); out = PCKEV_XORI128_UB(out2, out3); ST_UB(out, dst + 16); dst += dst_stride; - XORI_B4_128_SB(src0, src1, src2, src3); - HORIZ_8TAP_8WID_4VECS_FILT(src0, src1, src2, src3, mask0, mask1, mask2, + HORIZ_8TAP_8WID_4VECS_FILT(src4, src5, src6, src7, mask0, mask1, mask2, mask3, filt0, filt1, filt2, filt3, out0, out1, out2, out3); SRARI_H4_SH(out0, out1, out2, out3, 6); @@ -767,10 +780,11 @@ static void common_hz_8t_48w_msa(uint8_t *src, int32_t src_stride, { uint32_t loop_cnt; v16i8 src0, src1, src2, src3, filt0, filt1, filt2, filt3, vec0, vec1, vec2; + v16i8 src4; v16u8 mask0, mask1, mask2, mask3, mask4, mask5, mask6, mask7, out; - v8i16 filt, out0, out1, out2, out3, out4, out5, out6; + v8i16 filt, out0, out1, out2, out3; - mask0 = LD_UB(&mc_filt_mask_arr[0]); + mask0 = LD_UB(&ff_hevc_mask_arr[0]); src -= 3; /* rearranging filter */ @@ -785,11 +799,17 @@ static void common_hz_8t_48w_msa(uint8_t *src, int32_t src_stride, mask6 = mask0 + 12; mask7 = mask0 + 14; - for (loop_cnt = height; loop_cnt--;) { - LD_SB3(src, 16, src0, src2, src3); - src1 = __msa_sldi_b(src2, src0, 8); + for (loop_cnt = 64; loop_cnt--;) { + src0 = LD_SB(src); + src1 = LD_SB(src + 8); + src2 = LD_SB(src + 16); + src3 = LD_SB(src + 32); + src4 = LD_SB(src + 40); + src += src_stride; XORI_B4_128_SB(src0, src1, src2, src3); + src4 = (v16i8) __msa_xori_b((v16u8) src4, 128); + VSHF_B3_SB(src0, src0, src1, src1, src2, src2, mask0, mask0, mask0, vec0, vec1, vec2); DOTP_SB3_SH(vec0, vec1, vec2, filt0, filt0, filt0, out0, out1, out2); @@ -799,45 +819,42 @@ static void common_hz_8t_48w_msa(uint8_t *src, int32_t src_stride, out2 = __msa_dpadd_s_h(out2, vec2, filt1); VSHF_B3_SB(src0, src0, src1, src1, src2, src2, mask2, mask2, mask2, vec0, vec1, vec2); - DOTP_SB3_SH(vec0, vec1, vec2, filt2, filt2, filt2, out3, out4, out5); + DPADD_SB2_SH(vec0, vec1, filt2, filt2, out0, out1); + out2 = __msa_dpadd_s_h(out2, vec2, filt2); + VSHF_B3_SB(src0, src0, src1, src1, src2, src2, mask3, mask3, mask3, vec0, vec1, vec2); - DPADD_SB2_SH(vec0, vec1, filt3, filt3, out3, out4); - out5 = __msa_dpadd_s_h(out5, vec2, filt3); - ADDS_SH2_SH(out0, out3, out1, out4, out0, out1); - out2 = __msa_adds_s_h(out2, out5); + DPADD_SB2_SH(vec0, vec1, filt3, filt3, out0, out1); + out2 = __msa_dpadd_s_h(out2, vec2, filt3); + SRARI_H2_SH(out0, out1, 6); - out6 = __msa_srari_h(out2, 6); - SAT_SH3_SH(out0, out1, out6, 7); + out3 = __msa_srari_h(out2, 6); + SAT_SH3_SH(out0, out1, out3, 7); out = PCKEV_XORI128_UB(out0, out1); ST_UB(out, dst); - src1 = LD_SB(src + 40); - src += src_stride; - src1 = (v16i8) __msa_xori_b((v16u8) src1, 128); - - VSHF_B3_SB(src2, src3, src3, src3, src1, src1, mask4, mask0, mask0, + VSHF_B3_SB(src2, src3, src3, src3, src4, src4, mask4, mask0, mask0, vec0, vec1, vec2); DOTP_SB3_SH(vec0, vec1, vec2, filt0, filt0, filt0, out0, out1, out2); - VSHF_B3_SB(src2, src3, src3, src3, src1, src1, mask5, mask1, mask1, + VSHF_B3_SB(src2, src3, src3, src3, src4, src4, mask5, mask1, mask1, vec0, vec1, vec2); DPADD_SB2_SH(vec0, vec1, filt1, filt1, out0, out1); out2 = __msa_dpadd_s_h(out2, vec2, filt1); - VSHF_B3_SB(src2, src3, src3, src3, src1, src1, mask6, mask2, mask2, + VSHF_B3_SB(src2, src3, src3, src3, src4, src4, mask6, mask2, mask2, vec0, vec1, vec2); - DOTP_SB3_SH(vec0, vec1, vec2, filt2, filt2, filt2, out3, out4, out5); - VSHF_B3_SB(src2, src3, src3, src3, src1, src1, mask7, mask3, mask3, + DPADD_SB2_SH(vec0, vec1, filt2, filt2, out0, out1); + out2 = __msa_dpadd_s_h(out2, vec2, filt2); + VSHF_B3_SB(src2, src3, src3, src3, src4, src4, mask7, mask3, mask3, vec0, vec1, vec2); - DPADD_SB2_SH(vec0, vec1, filt3, filt3, out3, out4); - out5 = __msa_dpadd_s_h(out5, vec2, filt3); - ADDS_SH2_SH(out0, out3, out1, out4, out3, out4); - out5 = __msa_adds_s_h(out2, out5); - SRARI_H2_SH(out3, out4, 6); - out5 = __msa_srari_h(out5, 6); - SAT_SH3_SH(out3, out4, out5, 7); - out = PCKEV_XORI128_UB(out6, out3); + DPADD_SB2_SH(vec0, vec1, filt3, filt3, out0, out1); + out2 = __msa_dpadd_s_h(out2, vec2, filt3); + + SRARI_H2_SH(out0, out1, 6); + out2 = __msa_srari_h(out2, 6); + SAT_SH3_SH(out0, out1, out2, 7); + out = PCKEV_XORI128_UB(out3, out0); ST_UB(out, dst + 16); - out = PCKEV_XORI128_UB(out4, out5); + out = PCKEV_XORI128_UB(out1, out2); ST_UB(out, dst + 32); dst += dst_stride; } @@ -848,11 +865,13 @@ static void common_hz_8t_64w_msa(uint8_t *src, int32_t src_stride, const int8_t *filter, int32_t height) { int32_t loop_cnt; - v16i8 src0, src1, src2, src3, filt0, filt1, filt2, filt3; v16u8 mask0, mask1, mask2, mask3, out; - v8i16 filt, out0, out1, out2, out3; + v16i8 src0, src1, src2, src3, src4, src5, src6, src7; + v16i8 vec0, vec1, vec2, vec3, vec4, vec5, vec6, vec7; + v16i8 filt0, filt1, filt2, filt3; + v8i16 res0, res1, res2, res3, filt; - mask0 = LD_UB(&mc_filt_mask_arr[0]); + mask0 = LD_UB(&ff_hevc_mask_arr[0]); src -= 3; /* rearranging filter */ @@ -864,37 +883,57 @@ static void common_hz_8t_64w_msa(uint8_t *src, int32_t src_stride, mask3 = mask0 + 6; for (loop_cnt = height; loop_cnt--;) { - src0 = LD_SB(src); - src2 = LD_SB(src + 16); - src3 = LD_SB(src + 24); - src1 = __msa_sldi_b(src2, src0, 8); + LD_SB8(src, 8, src0, src1, src2, src3, src4, src5, src6, src7); + src += src_stride; - XORI_B4_128_SB(src0, src1, src2, src3); - HORIZ_8TAP_8WID_4VECS_FILT(src0, src1, src2, src3, mask0, mask1, - mask2, mask3, filt0, filt1, filt2, filt3, - out0, out1, out2, out3); - SRARI_H4_SH(out0, out1, out2, out3, 6); - SAT_SH4_SH(out0, out1, out2, out3, 7); - out = PCKEV_XORI128_UB(out0, out1); + XORI_B8_128_SB(src0, src1, src2, src3, src4, src5, src6, src7); + + VSHF_B2_SB(src0, src0, src1, src1, mask0, mask0, vec0, vec1); + VSHF_B2_SB(src2, src2, src3, src3, mask0, mask0, vec2, vec3); + DOTP_SB4_SH(vec0, vec1, vec2, vec3, filt0, filt0, filt0, filt0, res0, + res1, res2, res3); + VSHF_B2_SB(src0, src0, src1, src1, mask2, mask2, vec0, vec1); + VSHF_B2_SB(src2, src2, src3, src3, mask2, mask2, vec2, vec3); + DPADD_SB4_SH(vec0, vec1, vec2, vec3, filt2, filt2, filt2, filt2, res0, + res1, res2, res3); + VSHF_B2_SB(src0, src0, src1, src1, mask1, mask1, vec4, vec5); + VSHF_B2_SB(src2, src2, src3, src3, mask1, mask1, vec6, vec7); + DPADD_SB4_SH(vec4, vec5, vec6, vec7, filt1, filt1, filt1, filt1, res0, + res1, res2, res3); + VSHF_B2_SB(src0, src0, src1, src1, mask3, mask3, vec4, vec5); + VSHF_B2_SB(src2, src2, src3, src3, mask3, mask3, vec6, vec7); + DPADD_SB4_SH(vec4, vec5, vec6, vec7, filt3, filt3, filt3, filt3, res0, + res1, res2, res3); + + SRARI_H4_SH(res0, res1, res2, res3, 6); + SAT_SH4_SH(res0, res1, res2, res3, 7); + out = PCKEV_XORI128_UB(res0, res1); ST_UB(out, dst); - out = PCKEV_XORI128_UB(out2, out3); + out = PCKEV_XORI128_UB(res2, res3); ST_UB(out, dst + 16); - src0 = LD_SB(src + 32); - src2 = LD_SB(src + 48); - src3 = LD_SB(src + 56); - src1 = __msa_sldi_b(src2, src0, 8); - src += src_stride; - - XORI_B4_128_SB(src0, src1, src2, src3); - HORIZ_8TAP_8WID_4VECS_FILT(src0, src1, src2, src3, mask0, mask1, - mask2, mask3, filt0, filt1, filt2, filt3, - out0, out1, out2, out3); - SRARI_H4_SH(out0, out1, out2, out3, 6); - SAT_SH4_SH(out0, out1, out2, out3, 7); - out = PCKEV_XORI128_UB(out0, out1); + VSHF_B2_SB(src4, src4, src5, src5, mask0, mask0, vec0, vec1); + VSHF_B2_SB(src6, src6, src7, src7, mask0, mask0, vec2, vec3); + DOTP_SB4_SH(vec0, vec1, vec2, vec3, filt0, filt0, filt0, filt0, res0, + res1, res2, res3); + VSHF_B2_SB(src4, src4, src5, src5, mask2, mask2, vec0, vec1); + VSHF_B2_SB(src6, src6, src7, src7, mask2, mask2, vec2, vec3); + DPADD_SB4_SH(vec0, vec1, vec2, vec3, filt2, filt2, filt2, filt2, res0, + res1, res2, res3); + VSHF_B2_SB(src4, src4, src5, src5, mask1, mask1, vec4, vec5); + VSHF_B2_SB(src6, src6, src7, src7, mask1, mask1, vec6, vec7); + DPADD_SB4_SH(vec4, vec5, vec6, vec7, filt1, filt1, filt1, filt1, res0, + res1, res2, res3); + VSHF_B2_SB(src4, src4, src5, src5, mask3, mask3, vec4, vec5); + VSHF_B2_SB(src6, src6, src7, src7, mask3, mask3, vec6, vec7); + DPADD_SB4_SH(vec4, vec5, vec6, vec7, filt3, filt3, filt3, filt3, res0, + res1, res2, res3); + + SRARI_H4_SH(res0, res1, res2, res3, 6); + SAT_SH4_SH(res0, res1, res2, res3, 7); + out = PCKEV_XORI128_UB(res0, res1); ST_UB(out, dst + 32); - out = PCKEV_XORI128_UB(out2, out3); + out = PCKEV_XORI128_UB(res2, res3); ST_UB(out, dst + 48); dst += dst_stride; } @@ -3721,15 +3760,13 @@ UNI_MC_COPY(64); #define UNI_MC(PEL, DIR, WIDTH, TAP, DIR1, FILT_DIR) \ void ff_hevc_put_hevc_uni_##PEL##_##DIR##WIDTH##_8_msa(uint8_t *dst, \ - ptrdiff_t \ - dst_stride, \ - uint8_t *src, \ - ptrdiff_t \ - src_stride, \ - int height, \ - intptr_t mx, \ - intptr_t my, \ - int width) \ + ptrdiff_t dst_stride, \ + uint8_t *src, \ + ptrdiff_t src_stride, \ + int height, \ + intptr_t mx, \ + intptr_t my, \ + int width) \ { \ const int8_t *filter = ff_hevc_##PEL##_filters[FILT_DIR - 1]; \ \ @@ -3773,41 +3810,38 @@ UNI_MC(epel, v, 32, 4, vt, my); #undef UNI_MC -#define UNI_MC_HV(PEL, DIR, WIDTH, TAP, DIR1) \ -void ff_hevc_put_hevc_uni_##PEL##_##DIR##WIDTH##_8_msa(uint8_t *dst, \ - ptrdiff_t \ - dst_stride, \ - uint8_t *src, \ - ptrdiff_t \ - src_stride, \ - int height, \ - intptr_t mx, \ - intptr_t my, \ - int width) \ -{ \ - const int8_t *filter_x = ff_hevc_##PEL##_filters[mx - 1]; \ - const int8_t *filter_y = ff_hevc_##PEL##_filters[my - 1]; \ - \ - hevc_##DIR1##_uni_##TAP##t_##WIDTH##w_msa(src, src_stride, dst, \ - dst_stride, filter_x, \ - filter_y, height); \ -} - -UNI_MC_HV(qpel, hv, 4, 8, hv); -UNI_MC_HV(qpel, hv, 8, 8, hv); -UNI_MC_HV(qpel, hv, 12, 8, hv); -UNI_MC_HV(qpel, hv, 16, 8, hv); -UNI_MC_HV(qpel, hv, 24, 8, hv); -UNI_MC_HV(qpel, hv, 32, 8, hv); -UNI_MC_HV(qpel, hv, 48, 8, hv); -UNI_MC_HV(qpel, hv, 64, 8, hv); - -UNI_MC_HV(epel, hv, 4, 4, hv); -UNI_MC_HV(epel, hv, 6, 4, hv); -UNI_MC_HV(epel, hv, 8, 4, hv); -UNI_MC_HV(epel, hv, 12, 4, hv); -UNI_MC_HV(epel, hv, 16, 4, hv); -UNI_MC_HV(epel, hv, 24, 4, hv); -UNI_MC_HV(epel, hv, 32, 4, hv); +#define UNI_MC_HV(PEL, WIDTH, TAP) \ +void ff_hevc_put_hevc_uni_##PEL##_hv##WIDTH##_8_msa(uint8_t *dst, \ + ptrdiff_t dst_stride, \ + uint8_t *src, \ + ptrdiff_t src_stride, \ + int height, \ + intptr_t mx, \ + intptr_t my, \ + int width) \ +{ \ + const int8_t *filter_x = ff_hevc_##PEL##_filters[mx - 1]; \ + const int8_t *filter_y = ff_hevc_##PEL##_filters[my - 1]; \ + \ + hevc_hv_uni_##TAP##t_##WIDTH##w_msa(src, src_stride, dst, dst_stride, \ + filter_x, filter_y, height); \ +} + +UNI_MC_HV(qpel, 4, 8); +UNI_MC_HV(qpel, 8, 8); +UNI_MC_HV(qpel, 12, 8); +UNI_MC_HV(qpel, 16, 8); +UNI_MC_HV(qpel, 24, 8); +UNI_MC_HV(qpel, 32, 8); +UNI_MC_HV(qpel, 48, 8); +UNI_MC_HV(qpel, 64, 8); + +UNI_MC_HV(epel, 4, 4); +UNI_MC_HV(epel, 6, 4); +UNI_MC_HV(epel, 8, 4); +UNI_MC_HV(epel, 12, 4); +UNI_MC_HV(epel, 16, 4); +UNI_MC_HV(epel, 24, 4); +UNI_MC_HV(epel, 32, 4); #undef UNI_MC_HV