void ff_ac3_lshift_int16_neon(int16_t *src, unsigned len, unsigned shift);
void ff_ac3_rshift_int32_neon(int32_t *src, unsigned len, unsigned shift);
void ff_float_to_fixed24_neon(int32_t *dst, const float *src, unsigned int len);
+void ff_ac3_extract_exponents_neon(uint8_t *exp, int32_t *coef, int nb_coefs);
void ff_ac3_bit_alloc_calc_bap_armv6(int16_t *mask, int16_t *psd,
int start, int end,
c->ac3_lshift_int16 = ff_ac3_lshift_int16_neon;
c->ac3_rshift_int32 = ff_ac3_rshift_int32_neon;
c->float_to_fixed24 = ff_float_to_fixed24_neon;
+ c->extract_exponents = ff_ac3_extract_exponents_neon;
}
}
bgt 1b
bx lr
endfunc
+
+function ff_ac3_extract_exponents_neon, export=1
+ vmov.i32 q14, #24
+ vmov.i32 q15, #8
+1:
+ vld1.32 {q0}, [r1,:128]
+ vabs.s32 q1, q0
+ vclz.i32 q3, q1
+ vsub.i32 q3, q3, q15
+ vcge.s32 q2, q3, q14
+ vbit q3, q14, q2
+ vbic q0, q0, q2
+ vmovn.i32 d6, q3
+ vmovn.i16 d6, q3
+ vst1.32 {q0}, [r1,:128]!
+ vst1.32 {d6[0]}, [r0,:32]!
+ subs r2, r2, #4
+ bgt 1b
+ bx lr
+endfunc