/*
* ARM NEON optimised DSP functions
* Copyright (c) 2008 Mans Rullgard <mans@mansr.com>
*
* This file is part of FFmpeg.
*
* 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.
*
* 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 FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "libavutil/arm/asm.S"
function ff_clear_block_neon, export=1
vmov.i16 q0, #0
.rept 8
vst1.16 {q0}, [r0,:128]!
.endr
bx lr
endfunc
function ff_clear_blocks_neon, export=1
vmov.i16 q0, #0
.rept 8*6
vst1.16 {q0}, [r0,:128]!
.endr
bx lr
endfunc
function ff_put_pixels_clamped_neon, export=1
vld1.16 {d16-d19}, [r0,:128]!
vqmovun.s16 d0, q8
vld1.16 {d20-d23}, [r0,:128]!
vqmovun.s16 d1, q9
vld1.16 {d24-d27}, [r0,:128]!
vqmovun.s16 d2, q10
vld1.16 {d28-d31}, [r0,:128]!
vqmovun.s16 d3, q11
vst1.8 {d0}, [r1,:64], r2
vqmovun.s16 d4, q12
vst1.8 {d1}, [r1,:64], r2
vqmovun.s16 d5, q13
vst1.8 {d2}, [r1,:64], r2
vqmovun.s16 d6, q14
vst1.8 {d3}, [r1,:64], r2
vqmovun.s16 d7, q15
vst1.8 {d4}, [r1,:64], r2
vst1.8 {d5}, [r1,:64], r2
vst1.8 {d6}, [r1,:64], r2
vst1.8 {d7}, [r1,:64], r2
bx lr
endfunc
function ff_put_signed_pixels_clamped_neon, export=1
vmov.u8 d31, #128
vld1.16 {d16-d17}, [r0,:128]!
vqmovn.s16 d0, q8
vld1.16 {d18-d19}, [r0,:128]!
vqmovn.s16 d1, q9
vld1.16 {d16-d17}, [r0,:128]!
vqmovn.s16 d2, q8
vld1.16 {d18-d19}, [r0,:128]!
vadd.u8 d0, d0, d31
vld1.16 {d20-d21}, [r0,:128]!
vadd.u8 d1, d1, d31
vld1.16 {d22-d23}, [r0,:128]!
vadd.u8 d2, d2, d31
vst1.8 {d0}, [r1,:64], r2
vqmovn.s16 d3, q9
vst1.8 {d1}, [r1,:64], r2
vqmovn.s16 d4, q10
vst1.8 {d2}, [r1,:64], r2
vqmovn.s16 d5, q11
vld1.16 {d24-d25}, [r0,:128]!
vadd.u8 d3, d3, d31
vld1.16 {d26-d27}, [r0,:128]!
vadd.u8 d4, d4, d31
vadd.u8 d5, d5, d31
vst1.8 {d3}, [r1,:64], r2
vqmovn.s16 d6, q12
vst1.8 {d4}, [r1,:64], r2
vqmovn.s16 d7, q13
vst1.8 {d5}, [r1,:64], r2
vadd.u8 d6, d6, d31
vadd.u8 d7, d7, d31
vst1.8 {d6}, [r1,:64], r2
vst1.8 {d7}, [r1,:64], r2
bx lr
endfunc
function ff_add_pixels_clamped_neon, export=1
mov r3, r1
vld1.8 {d16}, [r1,:64], r2
vld1.16 {d0-d1}, [r0,:128]!
vaddw.u8 q0, q0, d16
vld1.8 {d17}, [r1,:64], r2
vld1.16 {d2-d3}, [r0,:128]!
vqmovun.s16 d0, q0
vld1.8 {d18}, [r1,:64], r2
vaddw.u8 q1, q1, d17
vld1.16 {d4-d5}, [r0,:128]!
vaddw.u8 q2, q2, d18
vst1.8 {d0}, [r3,:64], r2
vqmovun.s16 d2, q1
vld1.8 {d19}, [r1,:64], r2
vld1.16 {d6-d7}, [r0,:128]!
vaddw.u8 q3, q3, d19
vqmovun.s16 d4, q2
vst1.8 {d2}, [r3,:64], r2
vld1.8 {d16}, [r1,:64], r2
vqmovun.s16 d6, q3
vld1.16 {d0-d1}, [r0,:128]!
vaddw.u8 q0, q0, d16
vst1.8 {d4}, [r3,:64], r2
vld1.8 {d17}, [r1,:64], r2
vld1.16 {d2-d3}, [r0,:128]!
vaddw.u8 q1, q1, d17
vst1.8 {d6}, [r3,:64], r2
vqmovun.s16 d0, q0
vld1.8 {d18}, [r1,:64], r2
vld1.16 {d4-d5}, [r0,:128]!
vaddw.u8 q2, q2, d18
vst1.8 {d0}, [r3,:64], r2
vqmovun.s16 d2, q1
vld1.8 {d19}, [r1,:64], r2
vqmovun.s16 d4, q2
vld1.16 {d6-d7}, [r0,:128]!
vaddw.u8 q3, q3, d19
vst1.8 {d2}, [r3,:64], r2
vqmovun.s16 d6, q3
vst1.8 {d4}, [r3,:64], r2
vst1.8 {d6}, [r3,:64], r2
bx lr
endfunc
function ff_vector_clipf_neon, export=1
VFP vdup.32 q1, d0[1]
VFP vdup.32 q0, d0[0]
NOVFP vdup.32 q0, r2
NOVFP vdup.32 q1, r3
NOVFP ldr r2, [sp]
vld1.f32 {q2},[r1,:128]!
vmin.f32 q10, q2, q1
vld1.f32 {q3},[r1,:128]!
vmin.f32 q11, q3, q1
1: vmax.f32 q8, q10, q0
vmax.f32 q9, q11, q0
subs r2, r2, #8
beq 2f
vld1.f32 {q2},[r1,:128]!
vmin.f32 q10, q2, q1
vld1.f32 {q3},[r1,:128]!
vmin.f32 q11, q3, q1
vst1.f32 {q8},[r0,:128]!
vst1.f32 {q9},[r0,:128]!
b 1b
2: vst1.f32 {q8},[r0,:128]!
vst1.f32 {q9},[r0,:128]!
bx lr
endfunc
function ff_apply_window_int16_neon, export=1
push {r4,lr}
add r4, r1, r3, lsl #1
add lr, r0, r3, lsl #1
sub r4, r4, #16
sub lr, lr, #16
mov r12, #-16
1:
vld1.16 {q0}, [r1,:128]!
vld1.16 {q2}, [r2,:128]!
vld1.16 {q1}, [r4,:128], r12
vrev64.16 q3, q2
vqrdmulh.s16 q0, q0, q2
vqrdmulh.s16 d2, d2, d7
vqrdmulh.s16 d3, d3, d6
vst1.16 {q0}, [r0,:128]!
vst1.16 {q1}, [lr,:128], r12
subs r3, r3, #16
bgt 1b
pop {r4,pc}
endfunc
function ff_vector_clip_int32_neon, export=1
vdup.32 q0, r2
vdup.32 q1, r3
ldr r2, [sp]
1:
vld1.32 {q2-q3}, [r1,:128]!
vmin.s32 q2, q2, q1
vmin.s32 q3, q3, q1
vmax.s32 q2, q2, q0
vmax.s32 q3, q3, q0
vst1.32 {q2-q3}, [r0,:128]!
subs r2, r2, #8
bgt 1b
bx lr
endfunc