diff options
Diffstat (limited to 'libavcodec/aarch64/vp9mc_neon.S')
| -rw-r--r-- | libavcodec/aarch64/vp9mc_neon.S | 90 |
1 files changed, 49 insertions, 41 deletions
diff --git a/libavcodec/aarch64/vp9mc_neon.S b/libavcodec/aarch64/vp9mc_neon.S index 720273b115..82a0f53133 100644 --- a/libavcodec/aarch64/vp9mc_neon.S +++ b/libavcodec/aarch64/vp9mc_neon.S @@ -1,20 +1,20 @@ /* * Copyright (c) 2016 Google Inc. * - * This file is part of Libav. + * This file is part of FFmpeg. * - * Libav is free software; you can redistribute it and/or + * FFmpeg is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either * version 2.1 of the License, or (at your option) any later version. * - * Libav is distributed in the hope that it will be useful, + * FFmpeg is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public - * License along with Libav; if not, write to the Free Software + * License along with FFmpeg; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ @@ -193,41 +193,52 @@ endfunc // for size >= 16), and multiply-accumulate into dst1 and dst3 (or // dst1-dst2 and dst3-dst4 for size >= 16) .macro extmla dst1, dst2, dst3, dst4, src1, src2, src3, src4, src5, src6, offset, size - ext v20.16b, \src1, \src2, #(2*\offset) - ext v22.16b, \src4, \src5, #(2*\offset) + ext v20.16b, \src1\().16b, \src2\().16b, #(2*\offset) + ext v22.16b, \src4\().16b, \src5\().16b, #(2*\offset) .if \size >= 16 - mla \dst1, v20.8h, v0.h[\offset] - ext v21.16b, \src2, \src3, #(2*\offset) - mla \dst3, v22.8h, v0.h[\offset] - ext v23.16b, \src5, \src6, #(2*\offset) - mla \dst2, v21.8h, v0.h[\offset] - mla \dst4, v23.8h, v0.h[\offset] + mla \dst1\().8h, v20.8h, v0.h[\offset] + ext v21.16b, \src2\().16b, \src3\().16b, #(2*\offset) + mla \dst3\().8h, v22.8h, v0.h[\offset] + ext v23.16b, \src5\().16b, \src6\().16b, #(2*\offset) + mla \dst2\().8h, v21.8h, v0.h[\offset] + mla \dst4\().8h, v23.8h, v0.h[\offset] +.elseif \size == 8 + mla \dst1\().8h, v20.8h, v0.h[\offset] + mla \dst3\().8h, v22.8h, v0.h[\offset] .else - mla \dst1, v20.8h, v0.h[\offset] - mla \dst3, v22.8h, v0.h[\offset] + mla \dst1\().4h, v20.4h, v0.h[\offset] + mla \dst3\().4h, v22.4h, v0.h[\offset] .endif .endm // The same as above, but don't accumulate straight into the // destination, but use a temp register and accumulate with saturation. .macro extmulqadd dst1, dst2, dst3, dst4, src1, src2, src3, src4, src5, src6, offset, size - ext v20.16b, \src1, \src2, #(2*\offset) - ext v22.16b, \src4, \src5, #(2*\offset) + ext v20.16b, \src1\().16b, \src2\().16b, #(2*\offset) + ext v22.16b, \src4\().16b, \src5\().16b, #(2*\offset) .if \size >= 16 mul v20.8h, v20.8h, v0.h[\offset] - ext v21.16b, \src2, \src3, #(2*\offset) + ext v21.16b, \src2\().16b, \src3\().16b, #(2*\offset) mul v22.8h, v22.8h, v0.h[\offset] - ext v23.16b, \src5, \src6, #(2*\offset) + ext v23.16b, \src5\().16b, \src6\().16b, #(2*\offset) mul v21.8h, v21.8h, v0.h[\offset] mul v23.8h, v23.8h, v0.h[\offset] -.else +.elseif \size == 8 mul v20.8h, v20.8h, v0.h[\offset] mul v22.8h, v22.8h, v0.h[\offset] +.else + mul v20.4h, v20.4h, v0.h[\offset] + mul v22.4h, v22.4h, v0.h[\offset] .endif - sqadd \dst1, \dst1, v20.8h - sqadd \dst3, \dst3, v22.8h +.if \size == 4 + sqadd \dst1\().4h, \dst1\().4h, v20.4h + sqadd \dst3\().4h, \dst3\().4h, v22.4h +.else + sqadd \dst1\().8h, \dst1\().8h, v20.8h + sqadd \dst3\().8h, \dst3\().8h, v22.8h .if \size >= 16 - sqadd \dst2, \dst2, v21.8h - sqadd \dst4, \dst4, v23.8h + sqadd \dst2\().8h, \dst2\().8h, v21.8h + sqadd \dst4\().8h, \dst4\().8h, v23.8h +.endif .endif .endm @@ -250,7 +261,7 @@ function \type\()_8tap_\size\()h_\idx1\idx2 .if \size >= 16 sub x1, x1, x5 .endif - // size >= 16 loads two qwords and increments r2, + // size >= 16 loads two qwords and increments x2, // for size 4/8 it's enough with one qword and no // postincrement .if \size >= 16 @@ -258,8 +269,7 @@ function \type\()_8tap_\size\()h_\idx1\idx2 sub x3, x3, #8 .endif // Load the filter vector - ld1 {v0.8b}, [x9] - sxtl v0.8h, v0.8b + ld1 {v0.8h}, [x9] 1: .if \size >= 16 mov x9, x5 @@ -292,13 +302,13 @@ function \type\()_8tap_\size\()h_\idx1\idx2 mul v2.8h, v5.8h, v0.h[0] mul v25.8h, v17.8h, v0.h[0] .endif - extmla v1.8h, v2.8h, v24.8h, v25.8h, v4.16b, v5.16b, v6.16b, v16.16b, v17.16b, v18.16b, 1, \size - extmla v1.8h, v2.8h, v24.8h, v25.8h, v4.16b, v5.16b, v6.16b, v16.16b, v17.16b, v18.16b, 2, \size - extmla v1.8h, v2.8h, v24.8h, v25.8h, v4.16b, v5.16b, v6.16b, v16.16b, v17.16b, v18.16b, \idx1, \size - extmla v1.8h, v2.8h, v24.8h, v25.8h, v4.16b, v5.16b, v6.16b, v16.16b, v17.16b, v18.16b, 5, \size - extmla v1.8h, v2.8h, v24.8h, v25.8h, v4.16b, v5.16b, v6.16b, v16.16b, v17.16b, v18.16b, 6, \size - extmla v1.8h, v2.8h, v24.8h, v25.8h, v4.16b, v5.16b, v6.16b, v16.16b, v17.16b, v18.16b, 7, \size - extmulqadd v1.8h, v2.8h, v24.8h, v25.8h, v4.16b, v5.16b, v6.16b, v16.16b, v17.16b, v18.16b, \idx2, \size + extmla v1, v2, v24, v25, v4, v5, v6, v16, v17, v18, 1, \size + extmla v1, v2, v24, v25, v4, v5, v6, v16, v17, v18, 2, \size + extmla v1, v2, v24, v25, v4, v5, v6, v16, v17, v18, \idx1, \size + extmla v1, v2, v24, v25, v4, v5, v6, v16, v17, v18, 5, \size + extmla v1, v2, v24, v25, v4, v5, v6, v16, v17, v18, 6, \size + extmla v1, v2, v24, v25, v4, v5, v6, v16, v17, v18, 7, \size + extmulqadd v1, v2, v24, v25, v4, v5, v6, v16, v17, v18, \idx2, \size // Round, shift and saturate sqrshrun v1.8b, v1.8h, #7 @@ -373,9 +383,9 @@ do_8tap_h_size 16 .macro do_8tap_h_func type, filter, offset, size function ff_vp9_\type\()_\filter\()\size\()_h_neon, export=1 - movrel x6, X(ff_vp9_subpel_filters), 120*\offset - 8 + movrel x6, X(ff_vp9_subpel_filters), 256*\offset cmp w5, #8 - add x9, x6, w5, uxtw #3 + add x9, x6, w5, uxtw #4 mov x5, #\size .if \size >= 16 bge \type\()_8tap_16h_34 @@ -505,8 +515,7 @@ do_8tap_h_filters 4 function \type\()_8tap_8v_\idx1\idx2 sub x2, x2, x3, lsl #1 sub x2, x2, x3 - ld1 {v0.8b}, [x6] - sxtl v0.8h, v0.8b + ld1 {v0.8h}, [x6] 1: .ifc \type,avg mov x7, x0 @@ -579,8 +588,7 @@ do_8tap_8v avg, 4, 3 function \type\()_8tap_4v_\idx1\idx2 sub x2, x2, x3, lsl #1 sub x2, x2, x3 - ld1 {v0.8b}, [x6] - sxtl v0.8h, v0.8b + ld1 {v0.8h}, [x6] .ifc \type,avg mov x7, x0 .endif @@ -649,9 +657,9 @@ do_8tap_4v avg, 4, 3 .macro do_8tap_v_func type, filter, offset, size function ff_vp9_\type\()_\filter\()\size\()_v_neon, export=1 uxtw x4, w4 - movrel x5, X(ff_vp9_subpel_filters), 120*\offset - 8 + movrel x5, X(ff_vp9_subpel_filters), 256*\offset cmp w6, #8 - add x6, x5, w6, uxtw #3 + add x6, x5, w6, uxtw #4 mov x5, #\size .if \size >= 8 b.ge \type\()_8tap_8v_34 |