From 20f2bf5530472e4e8ef5be3f7d83a0455a1cac42 Mon Sep 17 00:00:00 2001 From: Zhao Zhili Date: Fri, 6 Sep 2024 17:58:18 +0800 Subject: [PATCH] aarch64/vvc: Add put_qpel_h_* and put_qpel_uni_h_* Just share hevc implementation. checkasm --test=vvc_mc --benchmark: put_luma_h_8_4x4_c: 0.2 ( 1.00x) put_luma_h_8_4x4_neon: 0.2 ( 1.00x) put_luma_h_8_8x8_c: 1.0 ( 1.00x) put_luma_h_8_8x8_neon: 0.2 ( 4.33x) put_luma_h_8_16x16_c: 3.2 ( 1.00x) put_luma_h_8_16x16_neon: 1.2 ( 2.63x) put_luma_h_8_32x32_c: 13.7 ( 1.00x) put_luma_h_8_32x32_neon: 4.0 ( 3.45x) put_luma_h_8_64x64_c: 48.2 ( 1.00x) put_luma_h_8_64x64_neon: 15.7 ( 3.07x) put_luma_h_8_128x128_c: 203.5 ( 1.00x) put_luma_h_8_128x128_neon: 62.0 ( 3.28x) put_uni_h_luma_8_4x4_c: 0.2 ( 1.00x) put_uni_h_luma_8_4x4_neon: 0.2 ( 1.00x) put_uni_h_luma_8_8x8_c: 1.5 ( 1.00x) put_uni_h_luma_8_8x8_neon: 0.2 ( 6.56x) put_uni_h_luma_8_16x16_c: 5.7 ( 1.00x) put_uni_h_luma_8_16x16_neon: 1.2 ( 4.67x) put_uni_h_luma_8_32x32_c: 24.0 ( 1.00x) put_uni_h_luma_8_32x32_neon: 4.7 ( 5.07x) put_uni_h_luma_8_64x64_c: 90.0 ( 1.00x) put_uni_h_luma_8_64x64_neon: 17.0 ( 5.30x) put_uni_h_luma_8_128x128_c: 357.7 ( 1.00x) put_uni_h_luma_8_128x128_neon: 67.5 ( 5.30x) --- libavcodec/aarch64/h26x/dsp.h | 13 ++ libavcodec/aarch64/h26x/qpel_neon.S | 232 +++++++++++++++++++--------- libavcodec/aarch64/vvc/Makefile | 1 + libavcodec/aarch64/vvc/dsp_init.c | 14 ++ 4 files changed, 186 insertions(+), 74 deletions(-) diff --git a/libavcodec/aarch64/h26x/dsp.h b/libavcodec/aarch64/h26x/dsp.h index 902286872d..f72746ce03 100644 --- a/libavcodec/aarch64/h26x/dsp.h +++ b/libavcodec/aarch64/h26x/dsp.h @@ -235,4 +235,17 @@ NEON8_FNPROTO(qpel_bi_hv, (uint8_t *dst, ptrdiff_t dststride, const uint8_t *src, ptrdiff_t srcstride, const int16_t *src2, int height, intptr_t mx, intptr_t my, int width), _i8mm); +#undef NEON8_FNPROTO_PARTIAL_4 +#define NEON8_FNPROTO_PARTIAL_4(fn, args, ext) \ + void ff_vvc_put_##fn##_h4_8_neon##ext args; \ + void ff_vvc_put_##fn##_h8_8_neon##ext args; \ + void ff_vvc_put_##fn##_h16_8_neon##ext args; \ + void ff_vvc_put_##fn##_h32_8_neon##ext args; + +NEON8_FNPROTO_PARTIAL_4(qpel, (int16_t *dst, const uint8_t *_src, ptrdiff_t _srcstride, int height, + const int8_t *hf, const int8_t *vf, int width),) + +NEON8_FNPROTO_PARTIAL_4(qpel_uni, (uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, + ptrdiff_t _srcstride, int height, const int8_t *hf, const int8_t *vf, int width),) + #endif diff --git a/libavcodec/aarch64/h26x/qpel_neon.S b/libavcodec/aarch64/h26x/qpel_neon.S index 8ddaa32b70..a05009c9d6 100644 --- a/libavcodec/aarch64/h26x/qpel_neon.S +++ b/libavcodec/aarch64/h26x/qpel_neon.S @@ -21,7 +21,8 @@ */ #include "libavutil/aarch64/asm.S" -#define MAX_PB_SIZE 64 +#define HEVC_MAX_PB_SIZE 64 +#define VVC_MAX_PB_SIZE 128 const qpel_filters, align=4 .byte 0, 0, 0, 0, 0, 0, 0, 0 @@ -44,6 +45,11 @@ endconst sxtl v0.8h, v0.8b .endm +.macro vvc_load_filter m + ld1 {v0.8b}, [\m] + sxtl v0.8h, v0.8b +.endm + .macro load_qpel_filterb freg, xreg movrel \xreg, qpel_filters_abs add \xreg, \xreg, \freg, lsl #3 @@ -212,22 +218,40 @@ function ff_hevc_put_hevc_h4_8_neon, export=0 endfunc .endif -function ff_hevc_put_hevc_\type\()_h4_8_neon, export=1 - load_filter mx -.ifc \type, qpel_bi - mov x16, #(MAX_PB_SIZE << 2) // src2bstridel - add x15, x4, #(MAX_PB_SIZE << 1) // src2b -.endif +.ifnc \type, qpel_bi +function ff_vvc_put_\type\()_h4_8_neon, export=1 + vvc_load_filter mx sub src, src, #3 mov mx, x30 .ifc \type, qpel - mov dststride, #(MAX_PB_SIZE << 1) + mov dststride, #(VVC_MAX_PB_SIZE << 1) lsl x13, srcstride, #1 // srcstridel - mov x14, #(MAX_PB_SIZE << 2) + mov x14, #(VVC_MAX_PB_SIZE << 2) .else lsl x14, dststride, #1 // dststridel lsl x13, srcstride, #1 // srcstridel .endif + b 1f +endfunc +.endif // !qpel_bi + +function ff_hevc_put_hevc_\type\()_h4_8_neon, export=1 + load_filter mx +.ifc \type, qpel_bi + mov x16, #(HEVC_MAX_PB_SIZE << 2) // src2bstridel + add x15, x4, #(HEVC_MAX_PB_SIZE << 1) // src2b +.endif + sub src, src, #3 + mov mx, x30 +.ifc \type, qpel + mov dststride, #(HEVC_MAX_PB_SIZE << 1) + lsl x13, srcstride, #1 // srcstridel + mov x14, #(HEVC_MAX_PB_SIZE << 2) +.else + lsl x14, dststride, #1 // dststridel + lsl x13, srcstride, #1 // srcstridel +.endif +1: add x10, dst, dststride // dstb add x12, src, srcstride // srcb 0: ld1 {v16.8b, v17.8b}, [src], x13 @@ -283,15 +307,15 @@ endfunc function ff_hevc_put_hevc_\type\()_h6_8_neon, export=1 load_filter mx .ifc \type, qpel_bi - mov x16, #(MAX_PB_SIZE << 2) // src2bstridel - add x15, x4, #(MAX_PB_SIZE << 1) // src2b + mov x16, #(HEVC_MAX_PB_SIZE << 2) // src2bstridel + add x15, x4, #(HEVC_MAX_PB_SIZE << 1) // src2b .endif sub src, src, #3 mov mx, x30 .ifc \type, qpel - mov dststride, #(MAX_PB_SIZE << 1) + mov dststride, #(HEVC_MAX_PB_SIZE << 1) lsl x13, srcstride, #1 // srcstridel - mov x14, #((MAX_PB_SIZE << 2) - 8) + mov x14, #((HEVC_MAX_PB_SIZE << 2) - 8) .else lsl x14, dststride, #1 // dststridel lsl x13, srcstride, #1 // srcstridel @@ -333,22 +357,40 @@ function ff_hevc_put_hevc_\type\()_h6_8_neon, export=1 ret mx endfunc -function ff_hevc_put_hevc_\type\()_h8_8_neon, export=1 - load_filter mx -.ifc \type, qpel_bi - mov x16, #(MAX_PB_SIZE << 2) // src2bstridel - add x15, x4, #(MAX_PB_SIZE << 1) // src2b -.endif +.ifnc \type, qpel_bi +function ff_vvc_put_\type\()_h8_8_neon, export=1 + vvc_load_filter mx sub src, src, #3 mov mx, x30 .ifc \type, qpel - mov dststride, #(MAX_PB_SIZE << 1) + mov dststride, #(VVC_MAX_PB_SIZE << 1) lsl x13, srcstride, #1 // srcstridel - mov x14, #(MAX_PB_SIZE << 2) + mov x14, #(VVC_MAX_PB_SIZE << 2) .else lsl x14, dststride, #1 // dststridel lsl x13, srcstride, #1 // srcstridel .endif + b 1f +endfunc +.endif // !qpel_bi + +function ff_hevc_put_hevc_\type\()_h8_8_neon, export=1 + load_filter mx +.ifc \type, qpel_bi + mov x16, #(HEVC_MAX_PB_SIZE << 2) // src2bstridel + add x15, x4, #(HEVC_MAX_PB_SIZE << 1) // src2b +.endif + sub src, src, #3 + mov mx, x30 +.ifc \type, qpel + mov dststride, #(HEVC_MAX_PB_SIZE << 1) + lsl x13, srcstride, #1 // srcstridel + mov x14, #(HEVC_MAX_PB_SIZE << 2) +.else + lsl x14, dststride, #1 // dststridel + lsl x13, srcstride, #1 // srcstridel +.endif +1: add x10, dst, dststride // dstb add x12, src, srcstride // srcb 0: ld1 {v16.8b, v17.8b}, [src], x13 @@ -415,16 +457,16 @@ function ff_hevc_put_hevc_\type\()_h12_8_neon, export=1 sxtw height, heightw .ifc \type, qpel_bi ldrh w8, [sp] // width - mov x16, #(MAX_PB_SIZE << 2) // src2bstridel - lsl x17, height, #7 // src2b reset (height * (MAX_PB_SIZE << 1)) - add x15, x4, #(MAX_PB_SIZE << 1) // src2b + mov x16, #(HEVC_MAX_PB_SIZE << 2) // src2bstridel + lsl x17, height, #7 // src2b reset (height * (HEVC_MAX_PB_SIZE << 1)) + add x15, x4, #(HEVC_MAX_PB_SIZE << 1) // src2b .endif sub src, src, #3 mov mx, x30 .ifc \type, qpel - mov dststride, #(MAX_PB_SIZE << 1) + mov dststride, #(HEVC_MAX_PB_SIZE << 1) lsl x13, srcstride, #1 // srcstridel - mov x14, #((MAX_PB_SIZE << 2) - 16) + mov x14, #((HEVC_MAX_PB_SIZE << 2) - 16) .else lsl x14, dststride, #1 // dststridel lsl x13, srcstride, #1 // srcstridel @@ -497,25 +539,45 @@ function ff_hevc_put_hevc_\type\()_h12_8_neon, export=1 ret mx endfunc +.ifnc \type, qpel_bi +function ff_vvc_put_\type\()_h16_8_neon, export=1 + vvc_load_filter mx + sxtw height, heightw + mov mx, x30 + sub src, src, #3 + mov mx, x30 +.ifc \type, qpel + mov dststride, #(VVC_MAX_PB_SIZE << 1) + lsl x13, srcstride, #1 // srcstridel + mov x14, #(VVC_MAX_PB_SIZE << 2) +.else + lsl x14, dststride, #1 // dststridel + lsl x13, srcstride, #1 // srcstridel +.endif + b 0f +endfunc +.endif // !qpel_bi + function ff_hevc_put_hevc_\type\()_h16_8_neon, export=1 load_filter mx sxtw height, heightw mov mx, x30 .ifc \type, qpel_bi ldrh w8, [sp] // width - mov x16, #(MAX_PB_SIZE << 2) // src2bstridel - add x15, x4, #(MAX_PB_SIZE << 1) // src2b + mov x16, #(HEVC_MAX_PB_SIZE << 2) // src2bstridel + add x15, x4, #(HEVC_MAX_PB_SIZE << 1) // src2b .endif sub src, src, #3 mov mx, x30 .ifc \type, qpel - mov dststride, #(MAX_PB_SIZE << 1) + mov dststride, #(HEVC_MAX_PB_SIZE << 1) lsl x13, srcstride, #1 // srcstridel - mov x14, #(MAX_PB_SIZE << 2) + mov x14, #(HEVC_MAX_PB_SIZE << 2) .else lsl x14, dststride, #1 // dststridel lsl x13, srcstride, #1 // srcstridel .endif +0: add x10, dst, dststride // dstb add x12, src, srcstride // srcb @@ -555,29 +617,51 @@ function ff_hevc_put_hevc_\type\()_h16_8_neon, export=1 ret mx endfunc -function ff_hevc_put_hevc_\type\()_h32_8_neon, export=1 - load_filter mx +.ifnc \type, qpel_bi +function ff_vvc_put_\type\()_h32_8_neon, export=1 + vvc_load_filter mx sxtw height, heightw mov mx, x30 -.ifc \type, qpel_bi - ldrh w8, [sp] // width - mov x16, #(MAX_PB_SIZE << 2) // src2bstridel - lsl x17, x5, #7 // src2b reset - add x15, x4, #(MAX_PB_SIZE << 1) // src2b - sub x16, x16, width, uxtw #1 -.endif sub src, src, #3 mov mx, x30 .ifc \type, qpel - mov dststride, #(MAX_PB_SIZE << 1) + mov dststride, #(VVC_MAX_PB_SIZE << 1) lsl x13, srcstride, #1 // srcstridel - mov x14, #(MAX_PB_SIZE << 2) + mov x14, #(VVC_MAX_PB_SIZE << 2) sub x14, x14, width, uxtw #1 .else lsl x14, dststride, #1 // dststridel lsl x13, srcstride, #1 // srcstridel sub x14, x14, width, uxtw .endif + b 1f +endfunc +.endif // !qpel_bi + +function ff_hevc_put_hevc_\type\()_h32_8_neon, export=1 + load_filter mx + sxtw height, heightw + mov mx, x30 +.ifc \type, qpel_bi + ldrh w8, [sp] // width + mov x16, #(HEVC_MAX_PB_SIZE << 2) // src2bstridel + lsl x17, x5, #7 // src2b reset + add x15, x4, #(HEVC_MAX_PB_SIZE << 1) // src2b + sub x16, x16, width, uxtw #1 +.endif + sub src, src, #3 + mov mx, x30 +.ifc \type, qpel + mov dststride, #(HEVC_MAX_PB_SIZE << 1) + lsl x13, srcstride, #1 // srcstridel + mov x14, #(HEVC_MAX_PB_SIZE << 2) + sub x14, x14, width, uxtw #1 +.else + lsl x14, dststride, #1 // dststridel + lsl x13, srcstride, #1 // srcstridel + sub x14, x14, width, uxtw +.endif +1: sub x13, x13, width, uxtw sub x13, x13, #8 add x10, dst, dststride // dstb @@ -651,7 +735,7 @@ put_hevc qpel_bi function ff_hevc_put_hevc_qpel_v4_8_neon, export=1 load_qpel_filterb x5, x4 sub x1, x1, x2, lsl #1 - mov x9, #(MAX_PB_SIZE * 2) + mov x9, #(HEVC_MAX_PB_SIZE * 2) sub x1, x1, x2 ldr s16, [x1] ldr s17, [x1, x2] @@ -680,7 +764,7 @@ endfunc function ff_hevc_put_hevc_qpel_v6_8_neon, export=1 load_qpel_filterb x5, x4 sub x1, x1, x2, lsl #1 - mov x9, #(MAX_PB_SIZE * 2 - 8) + mov x9, #(HEVC_MAX_PB_SIZE * 2 - 8) sub x1, x1, x2 ldr d16, [x1] ldr d17, [x1, x2] @@ -709,7 +793,7 @@ endfunc function ff_hevc_put_hevc_qpel_v8_8_neon, export=1 load_qpel_filterb x5, x4 sub x1, x1, x2, lsl #1 - mov x9, #(MAX_PB_SIZE * 2) + mov x9, #(HEVC_MAX_PB_SIZE * 2) sub x1, x1, x2 ldr d16, [x1] ldr d17, [x1, x2] @@ -737,7 +821,7 @@ endfunc function ff_hevc_put_hevc_qpel_v12_8_neon, export=1 load_qpel_filterb x5, x4 sub x1, x1, x2, lsl #1 - mov x9, #(MAX_PB_SIZE * 2 - 16) + mov x9, #(HEVC_MAX_PB_SIZE * 2 - 16) sub x1, x1, x2 ldr q16, [x1] ldr q17, [x1, x2] @@ -768,7 +852,7 @@ endfunc function ff_hevc_put_hevc_qpel_v16_8_neon, export=1 load_qpel_filterb x5, x4 sub x1, x1, x2, lsl #1 - mov x9, #(MAX_PB_SIZE * 2) + mov x9, #(HEVC_MAX_PB_SIZE * 2) sub x1, x1, x2 ldr q16, [x1] ldr q17, [x1, x2] @@ -802,7 +886,7 @@ function ff_hevc_put_hevc_qpel_v24_8_neon, export=1 load_qpel_filterb x5, x4 sub x1, x1, x2, lsl #1 sub x1, x1, x2 - mov x9, #(MAX_PB_SIZE * 2) + mov x9, #(HEVC_MAX_PB_SIZE * 2) ld1 {v16.16b, v17.16b}, [x1], x2 ld1 {v18.16b, v19.16b}, [x1], x2 ld1 {v20.16b, v21.16b}, [x1], x2 @@ -833,7 +917,7 @@ function ff_hevc_put_hevc_qpel_v32_8_neon, export=1 st1 {v8.8b-v11.8b}, [sp] load_qpel_filterb x5, x4 sub x1, x1, x2, lsl #1 - mov x9, #(MAX_PB_SIZE * 2) + mov x9, #(HEVC_MAX_PB_SIZE * 2) sub x1, x1, x2 ld1 {v16.16b, v17.16b}, [x1], x2 ld1 {v18.16b, v19.16b}, [x1], x2 @@ -883,7 +967,7 @@ function ff_hevc_put_hevc_qpel_v64_8_neon, export=1 load_qpel_filterb x5, x4 sub x1, x1, x2, lsl #1 sub x1, x1, x2 - mov x9, #(MAX_PB_SIZE * 2) + mov x9, #(HEVC_MAX_PB_SIZE * 2) 0: mov x8, x1 // src ld1 {v16.16b, v17.16b}, [x8], x2 mov w11, w3 // height @@ -921,7 +1005,7 @@ function ff_hevc_put_hevc_qpel_bi_v4_8_neon, export=1 load_qpel_filterb x7, x6 sub x2, x2, x3, lsl #1 sub x2, x2, x3 - mov x12, #(MAX_PB_SIZE * 2) + mov x12, #(HEVC_MAX_PB_SIZE * 2) ld1 {v16.s}[0], [x2], x3 ld1 {v17.s}[0], [x2], x3 ld1 {v18.s}[0], [x2], x3 @@ -951,7 +1035,7 @@ function ff_hevc_put_hevc_qpel_bi_v6_8_neon, export=1 ld1 {v16.8b}, [x2], x3 sub x1, x1, #4 ld1 {v17.8b}, [x2], x3 - mov x12, #(MAX_PB_SIZE * 2) + mov x12, #(HEVC_MAX_PB_SIZE * 2) ld1 {v18.8b}, [x2], x3 ld1 {v19.8b}, [x2], x3 ld1 {v20.8b}, [x2], x3 @@ -977,7 +1061,7 @@ function ff_hevc_put_hevc_qpel_bi_v8_8_neon, export=1 load_qpel_filterb x7, x6 sub x2, x2, x3, lsl #1 sub x2, x2, x3 - mov x12, #(MAX_PB_SIZE * 2) + mov x12, #(HEVC_MAX_PB_SIZE * 2) ld1 {v16.8b}, [x2], x3 ld1 {v17.8b}, [x2], x3 ld1 {v18.8b}, [x2], x3 @@ -1006,7 +1090,7 @@ function ff_hevc_put_hevc_qpel_bi_v12_8_neon, export=1 sub x2, x2, x3 sub x1, x1, #8 ld1 {v16.16b}, [x2], x3 - mov x12, #(MAX_PB_SIZE * 2) + mov x12, #(HEVC_MAX_PB_SIZE * 2) ld1 {v17.16b}, [x2], x3 ld1 {v18.16b}, [x2], x3 ld1 {v19.16b}, [x2], x3 @@ -1037,7 +1121,7 @@ function ff_hevc_put_hevc_qpel_bi_v16_8_neon, export=1 load_qpel_filterb x7, x6 sub x2, x2, x3, lsl #1 sub x2, x2, x3 - mov x12, #(MAX_PB_SIZE * 2) + mov x12, #(HEVC_MAX_PB_SIZE * 2) ld1 {v16.16b}, [x2], x3 ld1 {v17.16b}, [x2], x3 ld1 {v18.16b}, [x2], x3 @@ -1092,7 +1176,7 @@ function ff_hevc_put_hevc_qpel_bi_v32_8_neon, export=1 sub x2, x2, x3 load_qpel_filterb x7, x6 ldr w6, [sp, #64] - mov x12, #(MAX_PB_SIZE * 2) + mov x12, #(HEVC_MAX_PB_SIZE * 2) 0: mov x8, x2 // src ld1 {v16.16b, v17.16b}, [x8], x3 mov w11, w5 // height @@ -2147,7 +2231,7 @@ function ff_hevc_put_hevc_qpel_uni_w_v64_8_neon, export=1 endfunc function hevc_put_hevc_qpel_uni_hv4_8_end_neon - mov x9, #(MAX_PB_SIZE * 2) + mov x9, #(HEVC_MAX_PB_SIZE * 2) load_qpel_filterh x6, x5 ldr d16, [sp] ldr d17, [sp, x9] @@ -2174,7 +2258,7 @@ function hevc_put_hevc_qpel_uni_hv4_8_end_neon endfunc function hevc_put_hevc_qpel_uni_hv6_8_end_neon - mov x9, #(MAX_PB_SIZE * 2) + mov x9, #(HEVC_MAX_PB_SIZE * 2) load_qpel_filterh x6, x5 sub x1, x1, #4 ldr q16, [sp] @@ -2204,7 +2288,7 @@ function hevc_put_hevc_qpel_uni_hv6_8_end_neon endfunc function hevc_put_hevc_qpel_uni_hv8_8_end_neon - mov x9, #(MAX_PB_SIZE * 2) + mov x9, #(HEVC_MAX_PB_SIZE * 2) load_qpel_filterh x6, x5 ldr q16, [sp] ldr q17, [sp, x9] @@ -2232,7 +2316,7 @@ function hevc_put_hevc_qpel_uni_hv8_8_end_neon endfunc function hevc_put_hevc_qpel_uni_hv12_8_end_neon - mov x9, #(MAX_PB_SIZE * 2) + mov x9, #(HEVC_MAX_PB_SIZE * 2) load_qpel_filterh x6, x5 sub x1, x1, #8 ld1 {v16.8h, v17.8h}, [sp], x9 @@ -2260,7 +2344,7 @@ function hevc_put_hevc_qpel_uni_hv12_8_end_neon endfunc function hevc_put_hevc_qpel_uni_hv16_8_end_neon - mov x9, #(MAX_PB_SIZE * 2) + mov x9, #(HEVC_MAX_PB_SIZE * 2) load_qpel_filterh x6, x5 sub w12, w9, w7, lsl #1 0: mov x8, sp // src @@ -3355,7 +3439,7 @@ endfunc function ff_hevc_put_hevc_qpel_h4_8_neon_i8mm, export=1 QPEL_H_HEADER - mov x10, #MAX_PB_SIZE * 2 + mov x10, #HEVC_MAX_PB_SIZE * 2 1: ld1 {v0.16b}, [x1], x2 ext v1.16b, v0.16b, v0.16b, #1 @@ -3378,7 +3462,7 @@ endfunc function ff_hevc_put_hevc_qpel_h6_8_neon_i8mm, export=1 QPEL_H_HEADER - mov x10, #MAX_PB_SIZE * 2 + mov x10, #HEVC_MAX_PB_SIZE * 2 add x15, x0, #8 1: ld1 {v0.16b}, [x1], x2 @@ -3411,7 +3495,7 @@ endfunc function ff_hevc_put_hevc_qpel_h8_8_neon_i8mm, export=1 QPEL_H_HEADER - mov x10, #MAX_PB_SIZE * 2 + mov x10, #HEVC_MAX_PB_SIZE * 2 1: ld1 {v0.16b}, [x1], x2 ext v1.16b, v0.16b, v0.16b, #1 @@ -3457,7 +3541,7 @@ endfunc function ff_hevc_put_hevc_qpel_h12_8_neon_i8mm, export=1 QPEL_H_HEADER - mov x10, #MAX_PB_SIZE * 2 + mov x10, #HEVC_MAX_PB_SIZE * 2 add x15, x0, #16 1: ld1 {v16.16b, v17.16b}, [x1], x2 @@ -3495,7 +3579,7 @@ endfunc function ff_hevc_put_hevc_qpel_h16_8_neon_i8mm, export=1 QPEL_H_HEADER - mov x10, #MAX_PB_SIZE * 2 + mov x10, #HEVC_MAX_PB_SIZE * 2 1: ld1 {v16.16b, v17.16b}, [x1], x2 ext v1.16b, v16.16b, v17.16b, #1 @@ -3533,7 +3617,7 @@ endfunc function ff_hevc_put_hevc_qpel_h24_8_neon_i8mm, export=1 QPEL_H_HEADER - mov x10, #MAX_PB_SIZE * 2 + mov x10, #HEVC_MAX_PB_SIZE * 2 add x15, x0, #32 1: ld1 {v16.16b, v17.16b}, [x1], x2 @@ -3585,7 +3669,7 @@ endfunc function ff_hevc_put_hevc_qpel_h32_8_neon_i8mm, export=1 QPEL_H_HEADER - mov x10, #MAX_PB_SIZE * 2 + mov x10, #HEVC_MAX_PB_SIZE * 2 add x15, x0, #32 1: ld1 {v16.16b, v17.16b, v18.16b}, [x1], x2 @@ -3642,7 +3726,7 @@ endfunc function ff_hevc_put_hevc_qpel_h48_8_neon_i8mm, export=1 QPEL_H_HEADER - mov x10, #MAX_PB_SIZE * 2 - 64 + mov x10, #HEVC_MAX_PB_SIZE * 2 - 64 1: ld1 {v16.16b, v17.16b, v18.16b, v19.16b}, [x1], x2 ext v1.16b, v16.16b, v17.16b, #1 @@ -4173,7 +4257,7 @@ DISABLE_I8MM stp x24, x25, [sp, #48] stp x26, x27, [sp, #64] mov x19, sp - mov x11, #(MAX_PB_SIZE*(MAX_PB_SIZE+8)*2) + mov x11, #(HEVC_MAX_PB_SIZE*(HEVC_MAX_PB_SIZE+8)*2) sub sp, sp, x11 mov x20, x0 mov x21, x1 @@ -4204,7 +4288,7 @@ DISABLE_I8MM add x9, x9, x23, lsl #3 ld1 {v0.8b}, [x9] sxtl v0.8h, v0.8b - mov x10, #(MAX_PB_SIZE * 2) + mov x10, #(HEVC_MAX_PB_SIZE * 2) dup v28.4s, w24 dup v29.4s, w25 dup v30.4s, w26 @@ -4591,7 +4675,7 @@ endfunc qpel_uni_w_hv neon function hevc_put_hevc_qpel_bi_hv4_8_end_neon - mov x9, #(MAX_PB_SIZE * 2) + mov x9, #(HEVC_MAX_PB_SIZE * 2) load_qpel_filterh x7, x6 ld1 {v16.4h}, [sp], x9 ld1 {v17.4h}, [sp], x9 @@ -4617,7 +4701,7 @@ function hevc_put_hevc_qpel_bi_hv4_8_end_neon endfunc function hevc_put_hevc_qpel_bi_hv6_8_end_neon - mov x9, #(MAX_PB_SIZE * 2) + mov x9, #(HEVC_MAX_PB_SIZE * 2) load_qpel_filterh x7, x6 sub x1, x1, #4 ld1 {v16.8h}, [sp], x9 @@ -4648,7 +4732,7 @@ function hevc_put_hevc_qpel_bi_hv6_8_end_neon endfunc function hevc_put_hevc_qpel_bi_hv8_8_end_neon - mov x9, #(MAX_PB_SIZE * 2) + mov x9, #(HEVC_MAX_PB_SIZE * 2) load_qpel_filterh x7, x6 ld1 {v16.8h}, [sp], x9 ld1 {v17.8h}, [sp], x9 @@ -4678,7 +4762,7 @@ endfunc function hevc_put_hevc_qpel_bi_hv16_8_end_neon load_qpel_filterh x7, x8 - mov x9, #(MAX_PB_SIZE * 2) + mov x9, #(HEVC_MAX_PB_SIZE * 2) mov x10, x6 0: mov x8, sp // src ld1 {v16.8h, v17.8h}, [x8], x9 diff --git a/libavcodec/aarch64/vvc/Makefile b/libavcodec/aarch64/vvc/Makefile index 54c49fea92..a5ad24dfc5 100644 --- a/libavcodec/aarch64/vvc/Makefile +++ b/libavcodec/aarch64/vvc/Makefile @@ -3,4 +3,5 @@ clean:: OBJS-$(CONFIG_VVC_DECODER) += aarch64/vvc/dsp_init.o NEON-OBJS-$(CONFIG_VVC_DECODER) += aarch64/vvc/alf.o \ + aarch64/h26x/qpel_neon.o \ aarch64/h26x/sao_neon.o diff --git a/libavcodec/aarch64/vvc/dsp_init.c b/libavcodec/aarch64/vvc/dsp_init.c index 0aac140a8f..ea6245d9a3 100644 --- a/libavcodec/aarch64/vvc/dsp_init.c +++ b/libavcodec/aarch64/vvc/dsp_init.c @@ -46,6 +46,20 @@ void ff_vvc_dsp_init_aarch64(VVCDSPContext *const c, const int bd) return; if (bd == 8) { + c->inter.put[0][1][0][1] = ff_vvc_put_qpel_h4_8_neon; + c->inter.put[0][2][0][1] = ff_vvc_put_qpel_h8_8_neon; + c->inter.put[0][3][0][1] = ff_vvc_put_qpel_h16_8_neon; + c->inter.put[0][4][0][1] = + c->inter.put[0][5][0][1] = + c->inter.put[0][6][0][1] = ff_vvc_put_qpel_h32_8_neon; + + c->inter.put_uni[0][1][0][1] = ff_vvc_put_qpel_uni_h4_8_neon; + c->inter.put_uni[0][2][0][1] = ff_vvc_put_qpel_uni_h8_8_neon; + c->inter.put_uni[0][3][0][1] = ff_vvc_put_qpel_uni_h16_8_neon; + c->inter.put_uni[0][4][0][1] = + c->inter.put_uni[0][5][0][1] = + c->inter.put_uni[0][6][0][1] = ff_vvc_put_qpel_uni_h32_8_neon; + for (int i = 0; i < FF_ARRAY_ELEMS(c->sao.band_filter); i++) c->sao.band_filter[i] = ff_h26x_sao_band_filter_8x8_8_neon; c->sao.edge_filter[0] = ff_vvc_sao_edge_filter_8x8_8_neon;