yading@10: /*
yading@10:  * Copyright (c) 2008 Siarhei Siamashka <ssvb@users.sourceforge.net>
yading@10:  *
yading@10:  * This file is part of FFmpeg.
yading@10:  *
yading@10:  * FFmpeg is free software; you can redistribute it and/or
yading@10:  * modify it under the terms of the GNU Lesser General Public
yading@10:  * License as published by the Free Software Foundation; either
yading@10:  * version 2.1 of the License, or (at your option) any later version.
yading@10:  *
yading@10:  * FFmpeg is distributed in the hope that it will be useful,
yading@10:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
yading@10:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
yading@10:  * Lesser General Public License for more details.
yading@10:  *
yading@10:  * You should have received a copy of the GNU Lesser General Public
yading@10:  * License along with FFmpeg; if not, write to the Free Software
yading@10:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
yading@10:  */
yading@10: 
yading@10: #include "config.h"
yading@10: #include "libavutil/arm/asm.S"
yading@10: 
yading@10: /**
yading@10:  * ARM VFP optimized float to int16 conversion.
yading@10:  * Assume that len is a positive number and is multiple of 8, destination
yading@10:  * buffer is at least 4 bytes aligned (8 bytes alignment is better for
yading@10:  * performance), little-endian byte sex.
yading@10:  */
yading@10: @ void ff_float_to_int16_vfp(int16_t *dst, const float *src, int len)
yading@10: function ff_float_to_int16_vfp, export=1
yading@10:         push            {r4-r8,lr}
yading@10:         vpush           {d8-d11}
yading@10:         vldmia          r1!, {s16-s23}
yading@10:         vcvt.s32.f32    s0,  s16
yading@10:         vcvt.s32.f32    s1,  s17
yading@10:         vcvt.s32.f32    s2,  s18
yading@10:         vcvt.s32.f32    s3,  s19
yading@10:         vcvt.s32.f32    s4,  s20
yading@10:         vcvt.s32.f32    s5,  s21
yading@10:         vcvt.s32.f32    s6,  s22
yading@10:         vcvt.s32.f32    s7,  s23
yading@10: 1:
yading@10:         subs            r2,  r2,  #8
yading@10:         vmov            r3,  r4,  s0, s1
yading@10:         vmov            r5,  r6,  s2, s3
yading@10:         vmov            r7,  r8,  s4, s5
yading@10:         vmov            ip,  lr,  s6, s7
yading@10:         it              gt
yading@10:         vldmiagt        r1!, {s16-s23}
yading@10:         ssat            r4,  #16, r4
yading@10:         ssat            r3,  #16, r3
yading@10:         ssat            r6,  #16, r6
yading@10:         ssat            r5,  #16, r5
yading@10:         pkhbt           r3,  r3,  r4, lsl #16
yading@10:         pkhbt           r4,  r5,  r6, lsl #16
yading@10:         itttt           gt
yading@10:         vcvtgt.s32.f32  s0,  s16
yading@10:         vcvtgt.s32.f32  s1,  s17
yading@10:         vcvtgt.s32.f32  s2,  s18
yading@10:         vcvtgt.s32.f32  s3,  s19
yading@10:         itttt           gt
yading@10:         vcvtgt.s32.f32  s4,  s20
yading@10:         vcvtgt.s32.f32  s5,  s21
yading@10:         vcvtgt.s32.f32  s6,  s22
yading@10:         vcvtgt.s32.f32  s7,  s23
yading@10:         ssat            r8,  #16, r8
yading@10:         ssat            r7,  #16, r7
yading@10:         ssat            lr,  #16, lr
yading@10:         ssat            ip,  #16, ip
yading@10:         pkhbt           r5,  r7,  r8, lsl #16
yading@10:         pkhbt           r6,  ip,  lr, lsl #16
yading@10:         stmia           r0!, {r3-r6}
yading@10:         bgt             1b
yading@10: 
yading@10:         vpop            {d8-d11}
yading@10:         pop             {r4-r8,pc}
yading@10: endfunc