arm: add ff_int32_to_float_fmul_array8_neon

Quite a bit faster than int32_to_float_fmul_array8_c calling
ff_int32_to_float_fmul_scalar_neon through FmtConvertContext.
Number of cycles per int32_to_float_fmul_array8 call while decoding
padded.dts on exynos5422:

               before  after   change
cortex-a7:     1270     951    -25%
cortex-a15:     434     285    -34%

checkasm --bench cycle counts:     cortex-a15   cortex-a7
int32_to_float_fmul_array8_c:      1730.4       4384.5
int32_to_float_fmul_array8_neon_c:  571.5       1694.3
int32_to_float_fmul_array8_neon:    374.0       1448.8

Interesting are the differences between
int32_to_float_fmul_array8_neon_c and int32_to_float_fmul_array8_neon.
The former is current behaviour of calling
ff_int32_to_float_fmul_scalar_neon repeatedly from the c function,
The raw numbers differ since checkasm uses different lengths than the
dca decoder.

1 files changed, 4 insertions, 0 deletions

diff --git a/libavcodec/arm/fmtconvert_init_arm.c b/libavcodec/arm/fmtconvert_init_arm.c
index 6a80bfb6b3..11396e898c 100644
--- a/libavcodec/arm/fmtconvert_init_arm.c
+++ b/libavcodec/arm/fmtconvert_init_arm.c
@@ -25,6 +25,9 @@
 #include "libavcodec/avcodec.h"
 #include "libavcodec/fmtconvert.h"
 
+void ff_int32_to_float_fmul_array8_neon(FmtConvertContext *c, float *dst,
+                                        const int32_t *src, const float *mul,
+                                        int len);
 void ff_int32_to_float_fmul_scalar_neon(float *dst, const int32_t *src,
                                         float mul, int len);
 
@@ -46,6 +49,7 @@ av_cold void ff_fmt_convert_init_arm(FmtConvertContext *c, AVCodecContext *avctx
     }
 
     if (have_neon(cpu_flags)) {
+        c->int32_to_float_fmul_array8 = ff_int32_to_float_fmul_array8_neon;
         c->int32_to_float_fmul_scalar = ff_int32_to_float_fmul_scalar_neon;
     }
 }