diff options
| author | Mans Rullgard <mans@mansr.com> | 2012-08-11 04:18:53 +0100 |
|---|---|---|
| committer | Mans Rullgard <mans@mansr.com> | 2012-08-13 01:03:25 +0100 |
| commit | fddc5b9bea39968ed1f45c667869428865de7626 (patch) | |
| tree | 417219a2ad12b4bfe0ee0f9618ad42f4c8db0711 | |
| parent | 6c4975eaafd7f8f91e81ad8d6be744a434241fd3 (diff) | |
| download | ffmpeg-fddc5b9bea39968ed1f45c667869428865de7626.tar.gz | |
celp: optimise ff_celp_lp_synthesis_filter()
Adding instead of subtracting the products in the loop allows the
compiler to generate more efficient multiply-accumulate instructions
when 16-bit multiply-subtract is not available. ARM has only
multiply-accumulate for 16-bit operands. In general, if only one
variant exists, it is usually accumulate rather than subtract.
In the same spirit, using the dedicated saturation function enables
use of any special optimised versions of this.
Signed-off-by: Mans Rullgard <mans@mansr.com>
| -rw-r--r-- | libavcodec/celp_filters.c | 15 |
1 files changed, 7 insertions, 8 deletions
diff --git a/libavcodec/celp_filters.c b/libavcodec/celp_filters.c index 4e5bcda79a..d764d19219 100644 --- a/libavcodec/celp_filters.c +++ b/libavcodec/celp_filters.c @@ -63,17 +63,16 @@ int ff_celp_lp_synthesis_filter(int16_t *out, const int16_t *filter_coeffs, int i,n; for (n = 0; n < buffer_length; n++) { - int sum = rounder; + int sum = -rounder, sum1; for (i = 1; i <= filter_length; i++) - sum -= filter_coeffs[i-1] * out[n-i]; + sum += filter_coeffs[i-1] * out[n-i]; - sum = ((sum >> 12) + in[n]) >> shift; + sum1 = ((-sum >> 12) + in[n]) >> shift; + sum = av_clip_int16(sum1); + + if (stop_on_overflow && sum != sum1) + return 1; - if (sum + 0x8000 > 0xFFFFU) { - if (stop_on_overflow) - return 1; - sum = (sum >> 31) ^ 32767; - } out[n] = sum; } |