comparison dsputil.h @ 5737:efa3c1f9259a libavcodec

sse2 version of compute_autocorr(). 4x faster than c (somehow, even though doubles only allow 2x simd). overal flac encoding: 15-50% faster on core2, 4-11% on k8, 3-13% on p4.

5736:810067f2c33d

    void (*h261_loop_filter)(uint8_t *src, int stride);

    /* assume len is a multiple of 4, and arrays are 16-byte aligned */
    void (*vorbis_inverse_coupling)(float *mag, float *ang, int blocksize);
    /* assume len is a multiple of 8, and arrays are 16-byte aligned */
    void (*vector_fmul)(float *dst, const float *src, int len);
    void (*vector_fmul_reverse)(float *dst, const float *src0, const float *src1, int len);
    /* assume len is a multiple of 8, and src arrays are 16-byte aligned */
    void (*vector_fmul_add_add)(float *dst, const float *src0, const float *src1, const float *src2, int src3, int len, int step);

5737:efa3c1f9259a

    void (*h261_loop_filter)(uint8_t *src, int stride);

    /* assume len is a multiple of 4, and arrays are 16-byte aligned */
    void (*vorbis_inverse_coupling)(float *mag, float *ang, int blocksize);
    /* no alignment needed */
    void (*flac_compute_autocorr)(const int32_t *data, int len, int lag, double *autoc);
    /* assume len is a multiple of 8, and arrays are 16-byte aligned */
    void (*vector_fmul)(float *dst, const float *src, int len);
    void (*vector_fmul_reverse)(float *dst, const float *src0, const float *src1, int len);
    /* assume len is a multiple of 8, and src arrays are 16-byte aligned */
    void (*vector_fmul_add_add)(float *dst, const float *src0, const float *src1, const float *src2, int src3, int len, int step);