mirror of
https://github.com/librempeg/librempeg
synced 2024-11-23 03:28:27 +00:00
ARM: libavresample: NEON optimised stereo fltp to s16 conversion
Signed-off-by: Mans Rullgard <mans@mansr.com>
This commit is contained in:
parent
643b5b794f
commit
d26de339af
@ -26,6 +26,8 @@
|
||||
#include "libavresample/audio_convert.h"
|
||||
|
||||
void ff_conv_flt_to_s16_neon(int16_t *dst, const float *src, int len);
|
||||
void ff_conv_fltp_to_s16_2ch_neon(int16_t *dst, float *const *src,
|
||||
int len, int channels);
|
||||
|
||||
av_cold void ff_audio_convert_init_arm(AudioConvert *ac)
|
||||
{
|
||||
@ -35,5 +37,8 @@ av_cold void ff_audio_convert_init_arm(AudioConvert *ac)
|
||||
ff_audio_convert_set_func(ac, AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_FLT,
|
||||
0, 16, 8, "NEON",
|
||||
ff_conv_flt_to_s16_neon);
|
||||
ff_audio_convert_set_func(ac, AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_FLTP,
|
||||
2, 16, 8, "NEON",
|
||||
ff_conv_fltp_to_s16_2ch_neon);
|
||||
}
|
||||
}
|
||||
|
@ -64,3 +64,67 @@ function ff_conv_flt_to_s16_neon, export=1
|
||||
vst1.16 {q2}, [r0,:128]!
|
||||
bx lr
|
||||
endfunc
|
||||
|
||||
function ff_conv_fltp_to_s16_2ch_neon, export=1
|
||||
ldm r1, {r1, r3}
|
||||
subs r2, r2, #8
|
||||
vld1.32 {q0}, [r1,:128]!
|
||||
vcvt.s32.f32 q8, q0, #31
|
||||
vld1.32 {q1}, [r1,:128]!
|
||||
vcvt.s32.f32 q9, q1, #31
|
||||
vld1.32 {q10}, [r3,:128]!
|
||||
vcvt.s32.f32 q10, q10, #31
|
||||
vld1.32 {q11}, [r3,:128]!
|
||||
vcvt.s32.f32 q11, q11, #31
|
||||
beq 3f
|
||||
bics r12, r2, #15
|
||||
beq 2f
|
||||
1: subs r12, r12, #16
|
||||
vld1.32 {q0}, [r1,:128]!
|
||||
vcvt.s32.f32 q0, q0, #31
|
||||
vsri.32 q10, q8, #16
|
||||
vld1.32 {q1}, [r1,:128]!
|
||||
vcvt.s32.f32 q1, q1, #31
|
||||
vld1.32 {q12}, [r3,:128]!
|
||||
vcvt.s32.f32 q12, q12, #31
|
||||
vld1.32 {q13}, [r3,:128]!
|
||||
vsri.32 q11, q9, #16
|
||||
vst1.16 {q10}, [r0,:128]!
|
||||
vcvt.s32.f32 q13, q13, #31
|
||||
vst1.16 {q11}, [r0,:128]!
|
||||
vsri.32 q12, q0, #16
|
||||
vld1.32 {q8}, [r1,:128]!
|
||||
vsri.32 q13, q1, #16
|
||||
vst1.16 {q12}, [r0,:128]!
|
||||
vcvt.s32.f32 q8, q8, #31
|
||||
vld1.32 {q9}, [r1,:128]!
|
||||
vcvt.s32.f32 q9, q9, #31
|
||||
vld1.32 {q10}, [r3,:128]!
|
||||
vcvt.s32.f32 q10, q10, #31
|
||||
vld1.32 {q11}, [r3,:128]!
|
||||
vcvt.s32.f32 q11, q11, #31
|
||||
vst1.16 {q13}, [r0,:128]!
|
||||
bne 1b
|
||||
ands r2, r2, #15
|
||||
beq 3f
|
||||
2: vsri.32 q10, q8, #16
|
||||
vld1.32 {q0}, [r1,:128]!
|
||||
vcvt.s32.f32 q0, q0, #31
|
||||
vld1.32 {q1}, [r1,:128]!
|
||||
vcvt.s32.f32 q1, q1, #31
|
||||
vld1.32 {q12}, [r3,:128]!
|
||||
vcvt.s32.f32 q12, q12, #31
|
||||
vsri.32 q11, q9, #16
|
||||
vld1.32 {q13}, [r3,:128]!
|
||||
vcvt.s32.f32 q13, q13, #31
|
||||
vst1.16 {q10}, [r0,:128]!
|
||||
vsri.32 q12, q0, #16
|
||||
vst1.16 {q11}, [r0,:128]!
|
||||
vsri.32 q13, q1, #16
|
||||
vst1.16 {q12-q13},[r0,:128]!
|
||||
bx lr
|
||||
3: vsri.32 q10, q8, #16
|
||||
vsri.32 q11, q9, #16
|
||||
vst1.16 {q10-q11},[r0,:128]!
|
||||
bx lr
|
||||
endfunc
|
||||
|
Loading…
Reference in New Issue
Block a user