2 * Copyright (c) 2008 Siarhei Siamashka <ssvb@users.sourceforge.net>
4 * This file is part of FFmpeg
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
25 * Assume that len is a positive number and is multiple of 8
27 @ void ff_vector_fmul_vfp(float *dst, const float *src0, const float *src1, int len)
28 function ff_vector_fmul_vfp, export=1
31 orr r12, r12, #(3 << 16) /* set vector size to 4 */
43 vldmiage r1!, {s16-s19}
44 vldmiage r2!, {s24-s27}
45 vldmiage r1!, {s20-s23}
46 vldmiage r2!, {s28-s31}
48 vmulge.f32 s24, s16, s24
52 vmulge.f32 s28, s20, s28
55 vldmiagt r2!, {s8-s11}
57 vldmiagt r2!, {s12-s15}
60 vstmiage r0!, {s24-s27}
61 vstmiage r0!, {s28-s31}
64 bic r12, r12, #(7 << 16) /* set vector size back to 1 */
71 * ARM VFP optimized implementation of 'vector_fmul_reverse_c' function.
72 * Assume that len is a positive number and is multiple of 8
74 @ void ff_vector_fmul_reverse_vfp(float *dst, const float *src0,
75 @ const float *src1, int len)
76 function ff_vector_fmul_reverse_vfp, export=1
78 add r2, r2, r3, lsl #2
90 vldmdbge r2!, {s16-s19}
93 vldmiage r1!, {s24-s27}
96 vldmdbge r2!, {s20-s23}
99 vldmiage r1!, {s28-s31}
100 vmul.f32 s15, s4, s15
102 vmulge.f32 s24, s19, s24
104 vldmdbgt r2!, {s0-s3}
106 vmulge.f32 s25, s18, s25
109 vmulge.f32 s26, s17, s26
111 vldmiagt r1!, {s8-s11}
113 vmulge.f32 s27, s16, s27
114 vmulge.f32 s28, s23, s28
116 vldmdbgt r2!, {s4-s7}
118 vmulge.f32 s29, s22, s29
119 vstmia r0!, {s14-s15}
121 vmulge.f32 s30, s21, s30
122 vmulge.f32 s31, s20, s31
123 vmulge.f32 s8, s3, s8
125 vldmiagt r1!, {s12-s15}
127 vmulge.f32 s9, s2, s9
128 vmulge.f32 s10, s1, s10
129 vstmiage r0!, {s24-s27}
130 vmulge.f32 s11, s0, s11
132 vstmiage r0!, {s28-s31}