mdct15: add inverse transform postrotation SIMD

2.5ms frames:
Before   (c):  2638 decicycles in postrotate, 2097040 runs,    112 skips
After (sse3):  1467 decicycles in postrotate, 2097083 runs,     69 skips
After (avx2):  1244 decicycles in postrotate, 2097085 runs,     67 skips

5ms frames:
Before   (c):  4987 decicycles in postrotate, 1048371 runs,    205 skips
After (sse3):  2644 decicycles in postrotate, 1048509 runs,     67 skips
After (avx2):  2031 decicycles in postrotate, 1048523 runs,     53 skips

10ms frames:
Before   (c):  9153 decicycles in postrotate,  523575 runs,    713 skips
After (sse3):  5110 decicycles in postrotate,  523726 runs,    562 skips
After (avx2):  3738 decicycles in postrotate,  524223 runs,     65 skips

20ms frames:
Before   (c): 17857 decicycles in postrotate,  261866 runs,    278 skips
After (sse3): 10041 decicycles in postrotate,  261746 runs,    398 skips
After (avx2):  7050 decicycles in postrotate,  262116 runs,     28 skips

Improves total decoding performance for real world content by 9% with avx2.

Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>

1 files changed, 3 insertions, 0 deletions

diff --git a/libavcodec/mdct15.h b/libavcodec/mdct15.h
index 1c2149d436..42e60f3e10 100644
--- a/libavcodec/mdct15.h
+++ b/libavcodec/mdct15.h
@@ -42,6 +42,9 @@ typedef struct MDCT15Context {
     /* 15-point FFT */
     void (*fft15)(FFTComplex *out, FFTComplex *in, FFTComplex *exptab, ptrdiff_t stride);
 
+    /* PFA postrotate and exptab */
+    void (*postreindex)(FFTComplex *out, FFTComplex *in, FFTComplex *exp, int *lut, ptrdiff_t len8);
+
     /* Calculate a full 2N -> N MDCT */
     void (*mdct)(struct MDCT15Context *s, float *dst, const float *src, ptrdiff_t stride);