comparison dsputil.h @ 10644:5da7180afadf libavcodec

refactor and optimize scalarproduct 29-105% faster apply_filter, 6-90% faster ape decoding on core2 (Any x86 other than core2 probably gets much less, since this is mostly due to ssse3 cachesplit avoidance and I haven't written the full gamut of other cachesplit modes.) 9-123% faster ape decoding on G4.

10643:7f6911429cdc

    /* intrax8 functions */
    void (*x8_spatial_compensation[12])(uint8_t *src , uint8_t *dst, int linesize);
    void (*x8_setup_spatial_compensation)(uint8_t *src, uint8_t *dst, int linesize,
           int * range, int * sum,  int edges);

    /* ape functions */
    /**
     * Add contents of the second vector to the first one.
     * @param len length of vectors, should be multiple of 16
     */
    void (*add_int16)(int16_t *v1/*align 16*/, int16_t *v2, int len);
    /**
     * Add contents of the second vector to the first one.
     * @param len length of vectors, should be multiple of 16
     */
    void (*sub_int16)(int16_t *v1/*align 16*/, int16_t *v2, int len);
    /**
     * Calculate scalar product of two vectors.
     * @param len length of vectors, should be multiple of 16
     * @param shift number of bits to discard from product
     */
    int32_t (*scalarproduct_int16)(int16_t *v1, int16_t *v2/*align 16*/, int len, int shift);

    /* rv30 functions */
    qpel_mc_func put_rv30_tpel_pixels_tab[4][16];
    qpel_mc_func avg_rv30_tpel_pixels_tab[4][16];

10644:5da7180afadf

    /* intrax8 functions */
    void (*x8_spatial_compensation[12])(uint8_t *src , uint8_t *dst, int linesize);
    void (*x8_setup_spatial_compensation)(uint8_t *src, uint8_t *dst, int linesize,
           int * range, int * sum,  int edges);

    /**
     * Calculate scalar product of two vectors.
     * @param len length of vectors, should be multiple of 16
     * @param shift number of bits to discard from product
     */
    int32_t (*scalarproduct_int16)(int16_t *v1, int16_t *v2/*align 16*/, int len, int shift);
    /* ape functions */
    /**
     * Calculate scalar product of v1 and v2,
     * and v1[i] += v3[i] * mul
     * @param len length of vectors, should be multiple of 16
     */
    int32_t (*scalarproduct_and_madd_int16)(int16_t *v1/*align 16*/, int16_t *v2, int16_t *v3, int len, int mul);

    /* rv30 functions */
    qpel_mc_func put_rv30_tpel_pixels_tab[4][16];
    qpel_mc_func avg_rv30_tpel_pixels_tab[4][16];