xref: /external/chromium_org/third_party/libvpx/source/libvpx/vp8/common/mips/dspr2/loopfilter_filters_dspr2.c

/*
 *  Copyright (c) 2012 The WebM project authors. All Rights Reserved.
 *
 *  Use of this source code is governed by a BSD-style license
 *  that can be found in the LICENSE file in the root of the source
 *  tree. An additional intellectual property rights grant can be found
 *  in the file PATENTS.  All contributing project authors may
 *  be found in the AUTHORS file in the root of the source tree.
 */


#include <stdlib.h>
#include "vp8_rtcd.h"
#include "vp8/common/onyxc_int.h"

#if HAVE_DSPR2
typedef unsigned char uc;

/* prefetch data for load */
inline void prefetch_load_lf(unsigned char *src)
{
    __asm__ __volatile__ (
        "pref   0,  0(%[src])   \n\t"
        :
        : [src] "r" (src)
    );
}


/* prefetch data for store */
inline void prefetch_store_lf(unsigned char *dst)
{
    __asm__ __volatile__ (
        "pref   1,  0(%[dst])   \n\t"
        :
        : [dst] "r" (dst)
    );
}

/* processing 4 pixels at the same time
 * compute hev and mask in the same function
 */
static __inline void vp8_filter_mask_vec_mips
(
    uint32_t limit,
    uint32_t flimit,
    uint32_t p1,
    uint32_t p0,
    uint32_t p3,
    uint32_t p2,
    uint32_t q0,
    uint32_t q1,
    uint32_t q2,
    uint32_t q3,
    uint32_t thresh,
    uint32_t *hev,
    uint32_t *mask
)
{
    uint32_t c, r, r3, r_k;
    uint32_t s1, s2, s3;
    uint32_t ones = 0xFFFFFFFF;
    uint32_t hev1;

    __asm__ __volatile__ (
        /* mask |= (abs(p3 - p2) > limit) */
        "subu_s.qb      %[c],   %[p3],     %[p2]        \n\t"
        "subu_s.qb      %[r_k], %[p2],     %[p3]        \n\t"
        "or             %[r_k], %[r_k],    %[c]         \n\t"
        "cmpgu.lt.qb    %[c],   %[limit],  %[r_k]       \n\t"
        "or             %[r],   $0,        %[c]         \n\t"

        /* mask |= (abs(p2 - p1) > limit) */
        "subu_s.qb      %[c],   %[p2],     %[p1]        \n\t"
        "subu_s.qb      %[r_k], %[p1],     %[p2]        \n\t"
        "or             %[r_k], %[r_k],    %[c]         \n\t"
        "cmpgu.lt.qb    %[c],   %[limit],  %[r_k]       \n\t"
        "or             %[r],   %[r],      %[c]         \n\t"

        /* mask |= (abs(p1 - p0) > limit)
         * hev  |= (abs(p1 - p0) > thresh)
         */
        "subu_s.qb      %[c],   %[p1],     %[p0]        \n\t"
        "subu_s.qb      %[r_k], %[p0],     %[p1]        \n\t"
        "or             %[r_k], %[r_k],    %[c]         \n\t"
        "cmpgu.lt.qb    %[c],   %[thresh], %[r_k]       \n\t"
        "or             %[r3],  $0,        %[c]         \n\t"
        "cmpgu.lt.qb    %[c],   %[limit],  %[r_k]       \n\t"
        "or             %[r],   %[r],      %[c]         \n\t"

        /* mask |= (abs(q1 - q0) > limit)
         * hev  |= (abs(q1 - q0) > thresh)
         */
        "subu_s.qb      %[c],   %[q1],     %[q0]        \n\t"
        "subu_s.qb      %[r_k], %[q0],     %[q1]        \n\t"
        "or             %[r_k], %[r_k],    %[c]         \n\t"
        "cmpgu.lt.qb    %[c],   %[thresh], %[r_k]       \n\t"
        "or             %[r3],  %[r3],     %[c]         \n\t"
        "cmpgu.lt.qb    %[c],   %[limit],  %[r_k]       \n\t"
        "or             %[r],   %[r],      %[c]         \n\t"

        /* mask |= (abs(q2 - q1) > limit) */
        "subu_s.qb      %[c],   %[q2],     %[q1]        \n\t"
        "subu_s.qb      %[r_k], %[q1],     %[q2]        \n\t"
        "or             %[r_k], %[r_k],    %[c]         \n\t"
        "cmpgu.lt.qb    %[c],   %[limit],  %[r_k]       \n\t"
        "or             %[r],   %[r],      %[c]         \n\t"
        "sll            %[r3],    %[r3],    24          \n\t"

        /* mask |= (abs(q3 - q2) > limit) */
        "subu_s.qb      %[c],   %[q3],     %[q2]        \n\t"
        "subu_s.qb      %[r_k], %[q2],     %[q3]        \n\t"
        "or             %[r_k], %[r_k],    %[c]         \n\t"
        "cmpgu.lt.qb    %[c],   %[limit],  %[r_k]       \n\t"
        "or             %[r],   %[r],      %[c]         \n\t"

        : [c] "=&r" (c), [r_k] "=&r" (r_k),
          [r] "=&r" (r), [r3] "=&r" (r3)
        : [limit] "r" (limit), [p3] "r" (p3), [p2] "r" (p2),
          [p1] "r" (p1), [p0] "r" (p0), [q1] "r" (q1), [q0] "r" (q0),
          [q2] "r" (q2), [q3] "r" (q3), [thresh] "r" (thresh)
    );

    __asm__ __volatile__ (
        /* abs(p0 - q0) */
        "subu_s.qb      %[c],   %[p0],     %[q0]        \n\t"
        "subu_s.qb      %[r_k], %[q0],     %[p0]        \n\t"
        "wrdsp          %[r3]                           \n\t"
        "or             %[s1],  %[r_k],    %[c]         \n\t"

        /* abs(p1 - q1) */
        "subu_s.qb      %[c],    %[p1],    %[q1]        \n\t"
        "addu_s.qb      %[s3],   %[s1],    %[s1]        \n\t"
        "pick.qb        %[hev1], %[ones],  $0           \n\t"
        "subu_s.qb      %[r_k],  %[q1],    %[p1]        \n\t"
        "or             %[s2],   %[r_k],   %[c]         \n\t"

        /* abs(p0 - q0) * 2 + abs(p1 - q1) / 2  > flimit * 2 + limit */
        "shrl.qb        %[s2],   %[s2],     1           \n\t"
        "addu_s.qb      %[s1],   %[s2],     %[s3]       \n\t"
        "cmpgu.lt.qb    %[c],    %[flimit], %[s1]       \n\t"
        "or             %[r],    %[r],      %[c]        \n\t"
        "sll            %[r],    %[r],      24          \n\t"

        "wrdsp          %[r]                            \n\t"
        "pick.qb        %[s2],  $0,         %[ones]     \n\t"

        : [c] "=&r" (c), [r_k] "=&r" (r_k), [s1] "=&r" (s1), [hev1] "=&r" (hev1),
          [s2] "=&r" (s2), [r] "+r" (r), [s3] "=&r" (s3)
        : [p0] "r" (p0), [q0] "r" (q0), [p1] "r" (p1), [r3] "r" (r3),
          [q1] "r" (q1), [ones] "r" (ones), [flimit] "r" (flimit)
    );

    *hev = hev1;
    *mask = s2;
}


/* inputs & outputs are quad-byte vectors */
static __inline void vp8_filter_mips
(
    uint32_t mask,
    uint32_t hev,
    uint32_t *ps1,
    uint32_t *ps0,
    uint32_t *qs0,
    uint32_t *qs1
)
{
    int32_t vp8_filter_l, vp8_filter_r;
    int32_t Filter1_l, Filter1_r, Filter2_l, Filter2_r;
    int32_t subr_r, subr_l;
    uint32_t t1, t2, HWM, t3;
    uint32_t hev_l, hev_r, mask_l, mask_r, invhev_l, invhev_r;

    int32_t vps1, vps0, vqs0, vqs1;
    int32_t vps1_l, vps1_r, vps0_l, vps0_r, vqs0_l, vqs0_r, vqs1_l, vqs1_r;
    uint32_t N128;

    N128 = 0x80808080;
    t1  = 0x03000300;
    t2  = 0x04000400;
    t3  = 0x01000100;
    HWM = 0xFF00FF00;

    vps0 = (*ps0) ^ N128;
    vps1 = (*ps1) ^ N128;
    vqs0 = (*qs0) ^ N128;
    vqs1 = (*qs1) ^ N128;

    /* use halfword pairs instead quad-bytes because of accuracy */
    vps0_l = vps0 & HWM;
    vps0_r = vps0 << 8;
    vps0_r = vps0_r & HWM;

    vps1_l = vps1 & HWM;
    vps1_r = vps1 << 8;
    vps1_r = vps1_r & HWM;

    vqs0_l = vqs0 & HWM;
    vqs0_r = vqs0 << 8;
    vqs0_r = vqs0_r & HWM;

    vqs1_l = vqs1 & HWM;
    vqs1_r = vqs1 << 8;
    vqs1_r = vqs1_r & HWM;

    mask_l = mask & HWM;
    mask_r = mask << 8;
    mask_r = mask_r & HWM;

    hev_l = hev & HWM;
    hev_r = hev << 8;
    hev_r = hev_r & HWM;

    __asm__ __volatile__ (
        /* vp8_filter = vp8_signed_char_clamp(ps1 - qs1); */
        "subq_s.ph    %[vp8_filter_l], %[vps1_l],       %[vqs1_l]       \n\t"
        "subq_s.ph    %[vp8_filter_r], %[vps1_r],       %[vqs1_r]       \n\t"

        /* qs0 - ps0 */
        "subq_s.ph    %[subr_l],       %[vqs0_l],       %[vps0_l]       \n\t"
        "subq_s.ph    %[subr_r],       %[vqs0_r],       %[vps0_r]       \n\t"

        /* vp8_filter &= hev; */
        "and          %[vp8_filter_l], %[vp8_filter_l], %[hev_l]        \n\t"
        "and          %[vp8_filter_r], %[vp8_filter_r], %[hev_r]        \n\t"

        /* vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * (qs0 - ps0)); */
        "addq_s.ph    %[vp8_filter_l], %[vp8_filter_l], %[subr_l]       \n\t"
        "addq_s.ph    %[vp8_filter_r], %[vp8_filter_r], %[subr_r]       \n\t"
        "xor          %[invhev_l],     %[hev_l],        %[HWM]          \n\t"
        "addq_s.ph    %[vp8_filter_l], %[vp8_filter_l], %[subr_l]       \n\t"
        "addq_s.ph    %[vp8_filter_r], %[vp8_filter_r], %[subr_r]       \n\t"
        "xor          %[invhev_r],     %[hev_r],        %[HWM]          \n\t"
        "addq_s.ph    %[vp8_filter_l], %[vp8_filter_l], %[subr_l]       \n\t"
        "addq_s.ph    %[vp8_filter_r], %[vp8_filter_r], %[subr_r]       \n\t"

        /* vp8_filter &= mask; */
        "and          %[vp8_filter_l], %[vp8_filter_l], %[mask_l]       \n\t"
        "and          %[vp8_filter_r], %[vp8_filter_r], %[mask_r]       \n\t"

        : [vp8_filter_l] "=&r" (vp8_filter_l), [vp8_filter_r] "=&r" (vp8_filter_r),
          [subr_l] "=&r" (subr_l), [subr_r] "=&r" (subr_r),
          [invhev_l] "=&r" (invhev_l), [invhev_r] "=&r" (invhev_r)

        : [vps0_l] "r" (vps0_l), [vps0_r] "r" (vps0_r), [vps1_l] "r" (vps1_l),
          [vps1_r] "r" (vps1_r), [vqs0_l] "r" (vqs0_l), [vqs0_r] "r" (vqs0_r),
          [vqs1_l] "r" (vqs1_l), [vqs1_r] "r" (vqs1_r),
          [mask_l] "r" (mask_l), [mask_r] "r" (mask_r),
          [hev_l] "r" (hev_l), [hev_r] "r" (hev_r),
          [HWM] "r" (HWM)
    );

    /* save bottom 3 bits so that we round one side +4 and the other +3 */
    __asm__ __volatile__ (
        /* Filter2 = vp8_signed_char_clamp(vp8_filter + 3) >>= 3; */
        "addq_s.ph    %[Filter1_l],    %[vp8_filter_l], %[t2]           \n\t"
        "addq_s.ph    %[Filter1_r],    %[vp8_filter_r], %[t2]           \n\t"

        /* Filter1 = vp8_signed_char_clamp(vp8_filter + 4) >>= 3; */
        "addq_s.ph    %[Filter2_l],    %[vp8_filter_l], %[t1]           \n\t"
        "addq_s.ph    %[Filter2_r],    %[vp8_filter_r], %[t1]           \n\t"
        "shra.ph      %[Filter1_r],    %[Filter1_r],    3               \n\t"
        "shra.ph      %[Filter1_l],    %[Filter1_l],    3               \n\t"

        "shra.ph      %[Filter2_l],    %[Filter2_l],    3               \n\t"
        "shra.ph      %[Filter2_r],    %[Filter2_r],    3               \n\t"

        "and          %[Filter1_l],    %[Filter1_l],    %[HWM]          \n\t"
        "and          %[Filter1_r],    %[Filter1_r],    %[HWM]          \n\t"

        /* vps0 = vp8_signed_char_clamp(ps0 + Filter2); */
        "addq_s.ph    %[vps0_l],       %[vps0_l],       %[Filter2_l]    \n\t"
        "addq_s.ph    %[vps0_r],       %[vps0_r],       %[Filter2_r]    \n\t"

        /* vqs0 = vp8_signed_char_clamp(qs0 - Filter1); */
        "subq_s.ph    %[vqs0_l],       %[vqs0_l],       %[Filter1_l]    \n\t"
        "subq_s.ph    %[vqs0_r],       %[vqs0_r],       %[Filter1_r]    \n\t"

        : [Filter1_l] "=&r" (Filter1_l), [Filter1_r] "=&r" (Filter1_r),
          [Filter2_l] "=&r" (Filter2_l), [Filter2_r] "=&r" (Filter2_r),
          [vps0_l] "+r" (vps0_l), [vps0_r] "+r" (vps0_r),
          [vqs0_l] "+r" (vqs0_l), [vqs0_r] "+r" (vqs0_r)

        : [t1] "r" (t1), [t2] "r" (t2),
          [vp8_filter_l] "r" (vp8_filter_l), [vp8_filter_r] "r" (vp8_filter_r),
          [HWM] "r" (HWM)
    );

    __asm__ __volatile__ (
        /* (vp8_filter += 1) >>= 1 */
        "addqh.ph    %[Filter1_l],    %[Filter1_l],     %[t3]           \n\t"
        "addqh.ph    %[Filter1_r],    %[Filter1_r],     %[t3]           \n\t"

        /* vp8_filter &= ~hev; */
        "and          %[Filter1_l],    %[Filter1_l],    %[invhev_l]     \n\t"
        "and          %[Filter1_r],    %[Filter1_r],    %[invhev_r]     \n\t"

        /* vps1 = vp8_signed_char_clamp(ps1 + vp8_filter); */
        "addq_s.ph    %[vps1_l],       %[vps1_l],       %[Filter1_l]    \n\t"
        "addq_s.ph    %[vps1_r],       %[vps1_r],       %[Filter1_r]    \n\t"

        /* vqs1 = vp8_signed_char_clamp(qs1 - vp8_filter); */
        "subq_s.ph    %[vqs1_l],       %[vqs1_l],       %[Filter1_l]    \n\t"
        "subq_s.ph    %[vqs1_r],       %[vqs1_r],       %[Filter1_r]    \n\t"

        : [Filter1_l] "+r" (Filter1_l), [Filter1_r] "+r" (Filter1_r),
          [vps1_l] "+r" (vps1_l), [vps1_r] "+r" (vps1_r),
          [vqs1_l] "+r" (vqs1_l), [vqs1_r] "+r" (vqs1_r)

        : [t3] "r" (t3), [invhev_l] "r" (invhev_l), [invhev_r] "r" (invhev_r)
    );

    /* Create quad-bytes from halfword pairs */
    vqs0_l = vqs0_l & HWM;
    vqs1_l = vqs1_l & HWM;
    vps0_l = vps0_l & HWM;
    vps1_l = vps1_l & HWM;

    __asm__ __volatile__ (
        "shrl.ph      %[vqs0_r],       %[vqs0_r],       8               \n\t"
        "shrl.ph      %[vps0_r],       %[vps0_r],       8               \n\t"
        "shrl.ph      %[vqs1_r],       %[vqs1_r],       8               \n\t"
        "shrl.ph      %[vps1_r],       %[vps1_r],       8               \n\t"

        : [vps1_r] "+r" (vps1_r), [vqs1_r] "+r" (vqs1_r),
          [vps0_r] "+r" (vps0_r), [vqs0_r] "+r" (vqs0_r)
        :
    );

    vqs0 = vqs0_l | vqs0_r;
    vqs1 = vqs1_l | vqs1_r;
    vps0 = vps0_l | vps0_r;
    vps1 = vps1_l | vps1_r;

    *ps0 = vps0 ^ N128;
    *ps1 = vps1 ^ N128;
    *qs0 = vqs0 ^ N128;
    *qs1 = vqs1 ^ N128;
}

void vp8_loop_filter_horizontal_edge_mips
(
    unsigned char *s,
    int p,
    unsigned int flimit,
    unsigned int limit,
    unsigned int thresh,
    int count
)
{
    uint32_t mask;
    uint32_t hev;
    uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
    unsigned char *sm1, *s0, *s1, *s2, *s3, *s4, *s5, *s6;

    mask = 0;
    hev = 0;
    p1 = 0;
    p2 = 0;
    p3 = 0;
    p4 = 0;

    /* prefetch data for store */
    prefetch_store_lf(s);

    /* loop filter designed to work using chars so that we can make maximum use
     * of 8 bit simd instructions.
     */

    sm1 = s - (p << 2);
    s0 = s - p - p - p;
    s1 = s - p - p ;
    s2 = s - p;
    s3 = s;
    s4 = s + p;
    s5 = s + p + p;
    s6 = s + p + p + p;

    /* load quad-byte vectors
     * memory is 4 byte aligned
     */
    p1 = *((uint32_t *)(s1));
    p2 = *((uint32_t *)(s2));
    p3 = *((uint32_t *)(s3));
    p4 = *((uint32_t *)(s4));

    /* if (p1 - p4 == 0) and (p2 - p3 == 0)
     * mask will be zero and filtering is not needed
     */
    if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
    {

        pm1 = *((uint32_t *)(sm1));
        p0  = *((uint32_t *)(s0));
        p5  = *((uint32_t *)(s5));
        p6  = *((uint32_t *)(s6));

        vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
                                 thresh, &hev, &mask);

        /* if mask == 0 do filtering is not needed */
        if (mask)
        {
            /* filtering */
            vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);

            /* unpack processed 4x4 neighborhood */
            *((uint32_t *)s1) = p1;
            *((uint32_t *)s2) = p2;
            *((uint32_t *)s3) = p3;
            *((uint32_t *)s4) = p4;
        }
    }

    sm1 += 4;
    s0  += 4;
    s1  += 4;
    s2  += 4;
    s3  += 4;
    s4  += 4;
    s5  += 4;
    s6  += 4;

    /* load quad-byte vectors
     * memory is 4 byte aligned
     */
    p1 = *((uint32_t *)(s1));
    p2 = *((uint32_t *)(s2));
    p3 = *((uint32_t *)(s3));
    p4 = *((uint32_t *)(s4));

    /* if (p1 - p4 == 0) and (p2 - p3 == 0)
     * mask will be zero and filtering is not needed
     */
    if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
    {

        pm1 = *((uint32_t *)(sm1));
        p0  = *((uint32_t *)(s0));
        p5  = *((uint32_t *)(s5));
        p6  = *((uint32_t *)(s6));

        vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
                                 thresh, &hev, &mask);

        /* if mask == 0 do filtering is not needed */
        if (mask)
        {
            /* filtering */
            vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);

            /* unpack processed 4x4 neighborhood */
            *((uint32_t *)s1) = p1;
            *((uint32_t *)s2) = p2;
            *((uint32_t *)s3) = p3;
            *((uint32_t *)s4) = p4;
        }
    }

    sm1 += 4;
    s0  += 4;
    s1  += 4;
    s2  += 4;
    s3  += 4;
    s4  += 4;
    s5  += 4;
    s6  += 4;

    /* load quad-byte vectors
     * memory is 4 byte aligned
     */
    p1 = *((uint32_t *)(s1));
    p2 = *((uint32_t *)(s2));
    p3 = *((uint32_t *)(s3));
    p4 = *((uint32_t *)(s4));

    /* if (p1 - p4 == 0) and (p2 - p3 == 0)
     * mask will be zero and filtering is not needed
     */
    if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
    {

        pm1 = *((uint32_t *)(sm1));
        p0  = *((uint32_t *)(s0));
        p5  = *((uint32_t *)(s5));
        p6  = *((uint32_t *)(s6));

        vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
                                 thresh, &hev, &mask);

        /* if mask == 0 do filtering is not needed */
        if (mask)
        {
            /* filtering */
            vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);

            /* unpack processed 4x4 neighborhood */
            *((uint32_t *)s1) = p1;
            *((uint32_t *)s2) = p2;
            *((uint32_t *)s3) = p3;
            *((uint32_t *)s4) = p4;
        }
    }

    sm1 += 4;
    s0  += 4;
    s1  += 4;
    s2  += 4;
    s3  += 4;
    s4  += 4;
    s5  += 4;
    s6  += 4;

    /* load quad-byte vectors
     * memory is 4 byte aligned
     */
    p1 = *((uint32_t *)(s1));
    p2 = *((uint32_t *)(s2));
    p3 = *((uint32_t *)(s3));
    p4 = *((uint32_t *)(s4));

    /* if (p1 - p4 == 0) and (p2 - p3 == 0)
     * mask will be zero and filtering is not needed
     */
    if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
    {

        pm1 = *((uint32_t *)(sm1));
        p0  = *((uint32_t *)(s0));
        p5  = *((uint32_t *)(s5));
        p6  = *((uint32_t *)(s6));

        vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
                                 thresh, &hev, &mask);

        /* if mask == 0 do filtering is not needed */
        if (mask)
        {
            /* filtering */
            vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);

            /* unpack processed 4x4 neighborhood */
            *((uint32_t *)s1) = p1;
            *((uint32_t *)s2) = p2;
            *((uint32_t *)s3) = p3;
            *((uint32_t *)s4) = p4;
        }
    }
}

void vp8_loop_filter_uvhorizontal_edge_mips
(
    unsigned char *s,
    int p,
    unsigned int flimit,
    unsigned int limit,
    unsigned int thresh,
    int count
)
{
    uint32_t mask;
    uint32_t hev;
    uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
    unsigned char *sm1, *s0, *s1, *s2, *s3, *s4, *s5, *s6;

    mask = 0;
    hev = 0;
    p1 = 0;
    p2 = 0;
    p3 = 0;
    p4 = 0;

    /* loop filter designed to work using chars so that we can make maximum use
     * of 8 bit simd instructions.
     */

    sm1 = s - (p << 2);
    s0  = s - p - p - p;
    s1  = s - p - p ;
    s2  = s - p;
    s3  = s;
    s4  = s + p;
    s5  = s + p + p;
    s6  = s + p + p + p;

    /* load quad-byte vectors
     * memory is 4 byte aligned
     */
    p1 = *((uint32_t *)(s1));
    p2 = *((uint32_t *)(s2));
    p3 = *((uint32_t *)(s3));
    p4 = *((uint32_t *)(s4));

    /* if (p1 - p4 == 0) and (p2 - p3 == 0)
     * mask will be zero and filtering is not needed
     */
    if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
    {

        pm1 = *((uint32_t *)(sm1));
        p0  = *((uint32_t *)(s0));
        p5  = *((uint32_t *)(s5));
        p6  = *((uint32_t *)(s6));

        vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
                                 thresh, &hev, &mask);

        /* if mask == 0 do filtering is not needed */
        if (mask)
        {
            /* filtering */
            vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);

            /* unpack processed 4x4 neighborhood */
            *((uint32_t *)s1) = p1;
            *((uint32_t *)s2) = p2;
            *((uint32_t *)s3) = p3;
            *((uint32_t *)s4) = p4;
        }
    }

    sm1 += 4;
    s0  += 4;
    s1  += 4;
    s2  += 4;
    s3  += 4;
    s4  += 4;
    s5  += 4;
    s6  += 4;

    /* load quad-byte vectors
     * memory is 4 byte aligned
     */
    p1 = *((uint32_t *)(s1));
    p2 = *((uint32_t *)(s2));
    p3 = *((uint32_t *)(s3));
    p4 = *((uint32_t *)(s4));

    /* if (p1 - p4 == 0) and (p2 - p3 == 0)
     * mask will be zero and filtering is not needed
     */
    if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
    {

        pm1 = *((uint32_t *)(sm1));
        p0  = *((uint32_t *)(s0));
        p5  = *((uint32_t *)(s5));
        p6  = *((uint32_t *)(s6));

        vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
                                 thresh, &hev, &mask);

        /* if mask == 0 do filtering is not needed */
        if (mask)
        {
            /* filtering */
            vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);

            /* unpack processed 4x4 neighborhood */
            *((uint32_t *)s1) = p1;
            *((uint32_t *)s2) = p2;
            *((uint32_t *)s3) = p3;
            *((uint32_t *)s4) = p4;
        }
    }
}

void vp8_loop_filter_vertical_edge_mips
(
    unsigned char *s,
    int p,
    const unsigned int flimit,
    const unsigned int limit,
    const unsigned int thresh,
    int count
)
{
    int i;
    uint32_t mask, hev;
    uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
    unsigned char *s1, *s2, *s3, *s4;
    uint32_t prim1, prim2, sec3, sec4, prim3, prim4;

    hev = 0;
    mask = 0;
    i = 0;
    pm1 = 0;
    p0 = 0;
    p1 = 0;
    p2 = 0;
    p3 = 0;
    p4 = 0;
    p5 = 0;
    p6 = 0;

    /* loop filter designed to work using chars so that we can make maximum use
     * of 8 bit simd instructions.
     */

    /* apply filter on 4 pixesl at the same time */
    do
    {

        /* prefetch data for store */
        prefetch_store_lf(s + p);

        s1 = s;
        s2 = s + p;
        s3 = s2 + p;
        s4 = s3 + p;
        s  = s4 + p;

        /* load quad-byte vectors
         * memory is 4 byte aligned
         */
        p2  = *((uint32_t *)(s1 - 4));
        p6  = *((uint32_t *)(s1));
        p1  = *((uint32_t *)(s2 - 4));
        p5  = *((uint32_t *)(s2));
        p0  = *((uint32_t *)(s3 - 4));
        p4  = *((uint32_t *)(s3));
        pm1 = *((uint32_t *)(s4 - 4));
        p3  = *((uint32_t *)(s4));

        /* transpose pm1, p0, p1, p2 */
        __asm__ __volatile__ (
            "precrq.qb.ph   %[prim1],   %[p2],      %[p1]       \n\t"
            "precr.qb.ph    %[prim2],   %[p2],      %[p1]       \n\t"
            "precrq.qb.ph   %[prim3],   %[p0],      %[pm1]      \n\t"
            "precr.qb.ph    %[prim4],   %[p0],      %[pm1]      \n\t"

            "precrq.qb.ph   %[p1],      %[prim1],   %[prim2]    \n\t"
            "precr.qb.ph    %[pm1],     %[prim1],   %[prim2]    \n\t"
            "precrq.qb.ph   %[sec3],    %[prim3],   %[prim4]    \n\t"
            "precr.qb.ph    %[sec4],    %[prim3],   %[prim4]    \n\t"

            "precrq.ph.w    %[p2],      %[p1],      %[sec3]     \n\t"
            "precrq.ph.w    %[p0],      %[pm1],     %[sec4]     \n\t"
            "append         %[p1],      %[sec3],    16          \n\t"
            "append         %[pm1],     %[sec4],    16          \n\t"

            : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
              [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
              [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
              [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
            :
        );

        /* transpose p3, p4, p5, p6 */
        __asm__ __volatile__ (
            "precrq.qb.ph   %[prim1],   %[p6],      %[p5]       \n\t"
            "precr.qb.ph    %[prim2],   %[p6],      %[p5]       \n\t"
            "precrq.qb.ph   %[prim3],   %[p4],      %[p3]       \n\t"
            "precr.qb.ph    %[prim4],   %[p4],      %[p3]       \n\t"

            "precrq.qb.ph   %[p5],      %[prim1],   %[prim2]    \n\t"
            "precr.qb.ph    %[p3],      %[prim1],   %[prim2]    \n\t"
            "precrq.qb.ph   %[sec3],    %[prim3],   %[prim4]    \n\t"
            "precr.qb.ph    %[sec4],    %[prim3],   %[prim4]    \n\t"

            "precrq.ph.w    %[p6],      %[p5],      %[sec3]     \n\t"
            "precrq.ph.w    %[p4],      %[p3],      %[sec4]     \n\t"
            "append         %[p5],      %[sec3],    16          \n\t"
            "append         %[p3],      %[sec4],    16          \n\t"

            : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
              [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
              [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
              [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
            :
        );

        /* if (p1 - p4 == 0) and (p2 - p3 == 0)
         * mask will be zero and filtering is not needed
         */
        if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
        {

            vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
                                     thresh, &hev, &mask);

            /* if mask == 0 do filtering is not needed */
            if (mask)
            {
                /* filtering */
                vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);

                /* unpack processed 4x4 neighborhood
                 * don't use transpose on output data
                 * because memory isn't aligned
                 */
                __asm__ __volatile__ (
                    "sb         %[p4],  1(%[s4])    \n\t"
                    "sb         %[p3],  0(%[s4])    \n\t"
                    "sb         %[p2], -1(%[s4])    \n\t"
                    "sb         %[p1], -2(%[s4])    \n\t"
                    :
                    : [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
                      [p2] "r" (p2), [p1] "r" (p1)
                );

                __asm__ __volatile__ (
                    "srl        %[p4], %[p4], 8     \n\t"
                    "srl        %[p3], %[p3], 8     \n\t"
                    "srl        %[p2], %[p2], 8     \n\t"
                    "srl        %[p1], %[p1], 8     \n\t"
                    : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
                    :
                );

                __asm__ __volatile__ (
                    "sb         %[p4],  1(%[s3])    \n\t"
                    "sb         %[p3],  0(%[s3])    \n\t"
                    "sb         %[p2], -1(%[s3])    \n\t"
                    "sb         %[p1], -2(%[s3])    \n\t"
                    : [p1] "+r" (p1)
                    : [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3), [p2] "r" (p2)
                );

                __asm__ __volatile__ (
                    "srl        %[p4], %[p4], 8     \n\t"
                    "srl        %[p3], %[p3], 8     \n\t"
                    "srl        %[p2], %[p2], 8     \n\t"
                    "srl        %[p1], %[p1], 8     \n\t"
                    : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
                    :
                );

                __asm__ __volatile__ (
                    "sb         %[p4],  1(%[s2])    \n\t"
                    "sb         %[p3],  0(%[s2])    \n\t"
                    "sb         %[p2], -1(%[s2])    \n\t"
                    "sb         %[p1], -2(%[s2])    \n\t"
                    :
                    : [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
                      [p2] "r" (p2), [p1] "r" (p1)
                );

                __asm__ __volatile__ (
                    "srl        %[p4], %[p4], 8     \n\t"
                    "srl        %[p3], %[p3], 8     \n\t"
                    "srl        %[p2], %[p2], 8     \n\t"
                    "srl        %[p1], %[p1], 8     \n\t"
                    : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
                    :
                );

                __asm__ __volatile__ (
                    "sb         %[p4],  1(%[s1])    \n\t"
                    "sb         %[p3],  0(%[s1])    \n\t"
                    "sb         %[p2], -1(%[s1])    \n\t"
                    "sb         %[p1], -2(%[s1])    \n\t"
                    :
                    : [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
                      [p2] "r" (p2), [p1] "r" (p1)
                );
            }
        }

        s1 = s;
        s2 = s + p;
        s3 = s2 + p;
        s4 = s3 + p;
        s  = s4 + p;

        /* load quad-byte vectors
         * memory is 4 byte aligned
         */
        p2  = *((uint32_t *)(s1 - 4));
        p6  = *((uint32_t *)(s1));
        p1  = *((uint32_t *)(s2 - 4));
        p5  = *((uint32_t *)(s2));
        p0  = *((uint32_t *)(s3 - 4));
        p4  = *((uint32_t *)(s3));
        pm1 = *((uint32_t *)(s4 - 4));
        p3  = *((uint32_t *)(s4));

        /* transpose pm1, p0, p1, p2 */
        __asm__ __volatile__ (
            "precrq.qb.ph   %[prim1],   %[p2],      %[p1]       \n\t"
            "precr.qb.ph    %[prim2],   %[p2],      %[p1]       \n\t"
            "precrq.qb.ph   %[prim3],   %[p0],      %[pm1]      \n\t"
            "precr.qb.ph    %[prim4],   %[p0],      %[pm1]      \n\t"

            "precrq.qb.ph   %[p1],      %[prim1],   %[prim2]    \n\t"
            "precr.qb.ph    %[pm1],     %[prim1],   %[prim2]    \n\t"
            "precrq.qb.ph   %[sec3],    %[prim3],   %[prim4]    \n\t"
            "precr.qb.ph    %[sec4],    %[prim3],   %[prim4]    \n\t"

            "precrq.ph.w    %[p2],      %[p1],      %[sec3]     \n\t"
            "precrq.ph.w    %[p0],      %[pm1],     %[sec4]     \n\t"
            "append         %[p1],      %[sec3],    16          \n\t"
            "append         %[pm1],     %[sec4],    16          \n\t"

            : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
              [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
              [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
              [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
            :
        );

        /* transpose p3, p4, p5, p6 */
        __asm__ __volatile__ (
            "precrq.qb.ph   %[prim1],   %[p6],      %[p5]       \n\t"
            "precr.qb.ph    %[prim2],   %[p6],      %[p5]       \n\t"
            "precrq.qb.ph   %[prim3],   %[p4],      %[p3]       \n\t"
            "precr.qb.ph    %[prim4],   %[p4],      %[p3]       \n\t"

            "precrq.qb.ph   %[p5],      %[prim1],   %[prim2]    \n\t"
            "precr.qb.ph    %[p3],      %[prim1],   %[prim2]    \n\t"
            "precrq.qb.ph   %[sec3],    %[prim3],   %[prim4]    \n\t"
            "precr.qb.ph    %[sec4],    %[prim3],   %[prim4]    \n\t"

            "precrq.ph.w    %[p6],      %[p5],      %[sec3]     \n\t"
            "precrq.ph.w    %[p4],      %[p3],      %[sec4]     \n\t"
            "append         %[p5],      %[sec3],    16          \n\t"
            "append         %[p3],      %[sec4],    16          \n\t"

            : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
              [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
              [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
              [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
            :
        );

        /* if (p1 - p4 == 0) and (p2 - p3 == 0)
         * mask will be zero and filtering is not needed
         */
        if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
        {

            vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
                                     thresh, &hev, &mask);

            /* if mask == 0 do filtering is not needed */
            if (mask)
            {
                /* filtering */
                vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);

                /* unpack processed 4x4 neighborhood
                 * don't use transpose on output data
                 * because memory isn't aligned
                 */
                __asm__ __volatile__ (
                    "sb         %[p4],  1(%[s4])    \n\t"
                    "sb         %[p3],  0(%[s4])    \n\t"
                    "sb         %[p2], -1(%[s4])    \n\t"
                    "sb         %[p1], -2(%[s4])    \n\t"
                    :
                    : [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
                      [p2] "r" (p2), [p1] "r" (p1)
                );

                __asm__ __volatile__ (
                    "srl        %[p4], %[p4], 8     \n\t"
                    "srl        %[p3], %[p3], 8     \n\t"
                    "srl        %[p2], %[p2], 8     \n\t"
                    "srl        %[p1], %[p1], 8     \n\t"
                    : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
                    :
                );

                __asm__ __volatile__ (
                    "sb         %[p4],  1(%[s3])    \n\t"
                    "sb         %[p3],  0(%[s3])    \n\t"
                    "sb         %[p2], -1(%[s3])    \n\t"
                    "sb         %[p1], -2(%[s3])    \n\t"
                    : [p1] "+r" (p1)
                    : [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3), [p2] "r" (p2)
                );

                __asm__ __volatile__ (
                    "srl        %[p4], %[p4], 8     \n\t"
                    "srl        %[p3], %[p3], 8     \n\t"
                    "srl        %[p2], %[p2], 8     \n\t"
                    "srl        %[p1], %[p1], 8     \n\t"
                    : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
                    :
                );

                __asm__ __volatile__ (
                    "sb         %[p4],  1(%[s2])    \n\t"
                    "sb         %[p3],  0(%[s2])    \n\t"
                    "sb         %[p2], -1(%[s2])    \n\t"
                    "sb         %[p1], -2(%[s2])    \n\t"
                    :
                    : [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
                      [p2] "r" (p2), [p1] "r" (p1)
                );

                __asm__ __volatile__ (
                    "srl        %[p4], %[p4], 8     \n\t"
                    "srl        %[p3], %[p3], 8     \n\t"
                    "srl        %[p2], %[p2], 8     \n\t"
                    "srl        %[p1], %[p1], 8     \n\t"
                    : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
                    :
                );

                __asm__ __volatile__ (
                    "sb         %[p4],  1(%[s1])    \n\t"
                    "sb         %[p3],  0(%[s1])    \n\t"
                    "sb         %[p2], -1(%[s1])    \n\t"
                    "sb         %[p1], -2(%[s1])    \n\t"
                    :
                    : [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
                      [p2] "r" (p2), [p1] "r" (p1)
                );
            }
        }

        i += 8;
    }

    while (i < count);
}

void vp8_loop_filter_uvvertical_edge_mips
(
    unsigned char *s,
    int p,
    unsigned int flimit,
    unsigned int limit,
    unsigned int thresh,
    int count
)
{
    uint32_t mask, hev;
    uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
    unsigned char *s1, *s2, *s3, *s4;
    uint32_t prim1, prim2, sec3, sec4, prim3, prim4;

    /* loop filter designed to work using chars so that we can make maximum use
     * of 8 bit simd instructions.
     */

    /* apply filter on 4 pixesl at the same time */

    s1 = s;
    s2 = s + p;
    s3 = s2 + p;
    s4 = s3 + p;

    /* load quad-byte vectors
    * memory is 4 byte aligned
    */
    p2  = *((uint32_t *)(s1 - 4));
    p6  = *((uint32_t *)(s1));
    p1  = *((uint32_t *)(s2 - 4));
    p5  = *((uint32_t *)(s2));
    p0  = *((uint32_t *)(s3 - 4));
    p4  = *((uint32_t *)(s3));
    pm1 = *((uint32_t *)(s4 - 4));
    p3  = *((uint32_t *)(s4));

    /* transpose pm1, p0, p1, p2 */
    __asm__ __volatile__ (
        "precrq.qb.ph   %[prim1],   %[p2],      %[p1]       \n\t"
        "precr.qb.ph    %[prim2],   %[p2],      %[p1]       \n\t"
        "precrq.qb.ph   %[prim3],   %[p0],      %[pm1]      \n\t"
        "precr.qb.ph    %[prim4],   %[p0],      %[pm1]      \n\t"

        "precrq.qb.ph   %[p1],      %[prim1],   %[prim2]    \n\t"
        "precr.qb.ph    %[pm1],     %[prim1],   %[prim2]    \n\t"
        "precrq.qb.ph   %[sec3],    %[prim3],   %[prim4]    \n\t"
        "precr.qb.ph    %[sec4],    %[prim3],   %[prim4]    \n\t"

        "precrq.ph.w    %[p2],      %[p1],      %[sec3]     \n\t"
        "precrq.ph.w    %[p0],      %[pm1],     %[sec4]     \n\t"
        "append         %[p1],      %[sec3],    16          \n\t"
        "append         %[pm1],     %[sec4],    16          \n\t"

        : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
          [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
          [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
          [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
        :
    );

    /* transpose p3, p4, p5, p6 */
    __asm__ __volatile__ (
        "precrq.qb.ph   %[prim1],   %[p6],      %[p5]       \n\t"
        "precr.qb.ph    %[prim2],   %[p6],      %[p5]       \n\t"
        "precrq.qb.ph   %[prim3],   %[p4],      %[p3]       \n\t"
        "precr.qb.ph    %[prim4],   %[p4],      %[p3]       \n\t"

        "precrq.qb.ph   %[p5],      %[prim1],   %[prim2]    \n\t"
        "precr.qb.ph    %[p3],      %[prim1],   %[prim2]    \n\t"
        "precrq.qb.ph   %[sec3],    %[prim3],   %[prim4]    \n\t"
        "precr.qb.ph    %[sec4],    %[prim3],   %[prim4]    \n\t"

        "precrq.ph.w    %[p6],      %[p5],      %[sec3]     \n\t"
        "precrq.ph.w    %[p4],      %[p3],      %[sec4]     \n\t"
        "append         %[p5],      %[sec3],    16          \n\t"
        "append         %[p3],      %[sec4],    16          \n\t"

        : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
          [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
          [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
          [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
        :
    );

    /* if (p1 - p4 == 0) and (p2 - p3 == 0)
    * mask will be zero and filtering is not needed
    */
    if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
    {

        vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
                                 thresh, &hev, &mask);

        /* if mask == 0 do filtering is not needed */
        if (mask)
        {
            /* filtering */
            vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);

            /* unpack processed 4x4 neighborhood
             * don't use transpose on output data
             * because memory isn't aligned
             */
            __asm__ __volatile__ (
                "sb         %[p4],  1(%[s4])    \n\t"
                "sb         %[p3],  0(%[s4])    \n\t"
                "sb         %[p2], -1(%[s4])    \n\t"
                "sb         %[p1], -2(%[s4])    \n\t"
                :
                : [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
                  [p2] "r" (p2), [p1] "r" (p1)
            );

            __asm__ __volatile__ (
                "srl        %[p4], %[p4], 8     \n\t"
                "srl        %[p3], %[p3], 8     \n\t"
                "srl        %[p2], %[p2], 8     \n\t"
                "srl        %[p1], %[p1], 8     \n\t"
                : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
                :
            );

            __asm__ __volatile__ (
                "sb         %[p4],  1(%[s3])    \n\t"
                "sb         %[p3],  0(%[s3])    \n\t"
                "sb         %[p2], -1(%[s3])    \n\t"
                "sb         %[p1], -2(%[s3])    \n\t"
                : [p1] "+r" (p1)
                : [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3), [p2] "r" (p2)
            );

            __asm__ __volatile__ (
                "srl        %[p4], %[p4], 8     \n\t"
                "srl        %[p3], %[p3], 8     \n\t"
                "srl        %[p2], %[p2], 8     \n\t"
                "srl        %[p1], %[p1], 8     \n\t"
                : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
                :
            );

            __asm__ __volatile__ (
                "sb         %[p4],  1(%[s2])    \n\t"
                "sb         %[p3],  0(%[s2])    \n\t"
                "sb         %[p2], -1(%[s2])    \n\t"
                "sb         %[p1], -2(%[s2])    \n\t"
                :
                : [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
                  [p2] "r" (p2), [p1] "r" (p1)
            );

            __asm__ __volatile__ (
                "srl        %[p4], %[p4], 8     \n\t"
                "srl        %[p3], %[p3], 8     \n\t"
                "srl        %[p2], %[p2], 8     \n\t"
                "srl        %[p1], %[p1], 8     \n\t"
                : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
                :
            );

            __asm__ __volatile__ (
                "sb         %[p4],  1(%[s1])    \n\t"
                "sb         %[p3],  0(%[s1])    \n\t"
                "sb         %[p2], -1(%[s1])    \n\t"
                "sb         %[p1], -2(%[s1])    \n\t"
                :
                : [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1), [p2] "r" (p2), [p1] "r" (p1)
            );
        }
    }

    s1 = s4 + p;
    s2 = s1 + p;
    s3 = s2 + p;
    s4 = s3 + p;

    /* load quad-byte vectors
     * memory is 4 byte aligned
     */
    p2  = *((uint32_t *)(s1 - 4));
    p6  = *((uint32_t *)(s1));
    p1  = *((uint32_t *)(s2 - 4));
    p5  = *((uint32_t *)(s2));
    p0  = *((uint32_t *)(s3 - 4));
    p4  = *((uint32_t *)(s3));
    pm1 = *((uint32_t *)(s4 - 4));
    p3  = *((uint32_t *)(s4));

    /* transpose pm1, p0, p1, p2 */
    __asm__ __volatile__ (
        "precrq.qb.ph   %[prim1],   %[p2],      %[p1]       \n\t"
        "precr.qb.ph    %[prim2],   %[p2],      %[p1]       \n\t"
        "precrq.qb.ph   %[prim3],   %[p0],      %[pm1]      \n\t"
        "precr.qb.ph    %[prim4],   %[p0],      %[pm1]      \n\t"

        "precrq.qb.ph   %[p1],      %[prim1],   %[prim2]    \n\t"
        "precr.qb.ph    %[pm1],     %[prim1],   %[prim2]    \n\t"
        "precrq.qb.ph   %[sec3],    %[prim3],   %[prim4]    \n\t"
        "precr.qb.ph    %[sec4],    %[prim3],   %[prim4]    \n\t"

        "precrq.ph.w    %[p2],      %[p1],      %[sec3]     \n\t"
        "precrq.ph.w    %[p0],      %[pm1],     %[sec4]     \n\t"
        "append         %[p1],      %[sec3],    16          \n\t"
        "append         %[pm1],     %[sec4],    16          \n\t"

        : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
          [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
          [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
          [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
        :
    );

    /* transpose p3, p4, p5, p6 */
    __asm__ __volatile__ (
        "precrq.qb.ph   %[prim1],   %[p6],      %[p5]       \n\t"
        "precr.qb.ph    %[prim2],   %[p6],      %[p5]       \n\t"
        "precrq.qb.ph   %[prim3],   %[p4],      %[p3]       \n\t"
        "precr.qb.ph    %[prim4],   %[p4],      %[p3]       \n\t"

        "precrq.qb.ph   %[p5],      %[prim1],   %[prim2]    \n\t"
        "precr.qb.ph    %[p3],      %[prim1],   %[prim2]    \n\t"
        "precrq.qb.ph   %[sec3],    %[prim3],   %[prim4]    \n\t"
        "precr.qb.ph    %[sec4],    %[prim3],   %[prim4]    \n\t"

        "precrq.ph.w    %[p6],      %[p5],      %[sec3]     \n\t"
        "precrq.ph.w    %[p4],      %[p3],      %[sec4]     \n\t"
        "append         %[p5],      %[sec3],    16          \n\t"
        "append         %[p3],      %[sec4],    16          \n\t"

        : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
          [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
          [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
          [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
        :
    );

    /* if (p1 - p4 == 0) and (p2 - p3 == 0)
     * mask will be zero and filtering is not needed
     */
    if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
    {

        vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
                                 thresh, &hev, &mask);

        /* if mask == 0 do filtering is not needed */
        if (mask)
        {
            /* filtering */
            vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);

            /* unpack processed 4x4 neighborhood
             * don't use transpose on output data
             * because memory isn't aligned
             */
            __asm__ __volatile__ (
                "sb         %[p4],  1(%[s4])    \n\t"
                "sb         %[p3],  0(%[s4])    \n\t"
                "sb         %[p2], -1(%[s4])    \n\t"
                "sb         %[p1], -2(%[s4])    \n\t"
                :
                : [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
                  [p2] "r" (p2), [p1] "r" (p1)
            );

            __asm__ __volatile__ (
                "srl        %[p4], %[p4], 8     \n\t"
                "srl        %[p3], %[p3], 8     \n\t"
                "srl        %[p2], %[p2], 8     \n\t"
                "srl        %[p1], %[p1], 8     \n\t"
                : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
                :
            );

            __asm__ __volatile__ (
                "sb         %[p4],  1(%[s3])    \n\t"
                "sb         %[p3],  0(%[s3])    \n\t"
                "sb         %[p2], -1(%[s3])    \n\t"
                "sb         %[p1], -2(%[s3])    \n\t"
                : [p1] "+r" (p1)
                : [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3), [p2] "r" (p2)
            );

            __asm__ __volatile__ (
                "srl        %[p4], %[p4], 8     \n\t"
                "srl        %[p3], %[p3], 8     \n\t"
                "srl        %[p2], %[p2], 8     \n\t"
                "srl        %[p1], %[p1], 8     \n\t"
                : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
                :
            );

            __asm__ __volatile__ (
                "sb         %[p4],  1(%[s2])    \n\t"
                "sb         %[p3],  0(%[s2])    \n\t"
                "sb         %[p2], -1(%[s2])    \n\t"
                "sb         %[p1], -2(%[s2])    \n\t"
                :
                : [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
                  [p2] "r" (p2), [p1] "r" (p1)
            );

            __asm__ __volatile__ (
                "srl        %[p4], %[p4], 8     \n\t"
                "srl        %[p3], %[p3], 8     \n\t"
                "srl        %[p2], %[p2], 8     \n\t"
                "srl        %[p1], %[p1], 8     \n\t"
                : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
                :
            );

            __asm__ __volatile__ (
                "sb         %[p4],  1(%[s1])    \n\t"
                "sb         %[p3],  0(%[s1])    \n\t"
                "sb         %[p2], -1(%[s1])    \n\t"
                "sb         %[p1], -2(%[s1])    \n\t"
                :
                : [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
                  [p2] "r" (p2), [p1] "r" (p1)
            );
        }
    }
}

/* inputs & outputs are quad-byte vectors */
static __inline void vp8_mbfilter_mips
(
    uint32_t mask,
    uint32_t hev,
    uint32_t *ps2,
    uint32_t *ps1,
    uint32_t *ps0,
    uint32_t *qs0,
    uint32_t *qs1,
    uint32_t *qs2
)
{
    int32_t vps2, vps1, vps0, vqs0, vqs1, vqs2;
    int32_t vps2_l, vps1_l, vps0_l, vqs0_l, vqs1_l, vqs2_l;
    int32_t vps2_r, vps1_r, vps0_r, vqs0_r, vqs1_r, vqs2_r;
    uint32_t HWM, vp8_filter_l, vp8_filter_r, mask_l, mask_r, hev_l, hev_r, subr_r, subr_l;
    uint32_t Filter2_l, Filter2_r, t1, t2, Filter1_l, Filter1_r, invhev_l, invhev_r;
    uint32_t N128, R63;
    uint32_t u1_l, u1_r, u2_l, u2_r, u3_l, u3_r;

    R63  = 0x003F003F;
    HWM  = 0xFF00FF00;
    N128 = 0x80808080;
    t1   = 0x03000300;
    t2   = 0x04000400;

    vps0 = (*ps0) ^ N128;
    vps1 = (*ps1) ^ N128;
    vps2 = (*ps2) ^ N128;
    vqs0 = (*qs0) ^ N128;
    vqs1 = (*qs1) ^ N128;
    vqs2 = (*qs2) ^ N128;

    /* use halfword pairs instead quad-bytes because of accuracy */
    vps0_l = vps0 & HWM;
    vps0_r = vps0 << 8;
    vps0_r = vps0_r & HWM;

    vqs0_l = vqs0 & HWM;
    vqs0_r = vqs0 << 8;
    vqs0_r = vqs0_r & HWM;

    vps1_l = vps1 & HWM;
    vps1_r = vps1 << 8;
    vps1_r = vps1_r & HWM;

    vqs1_l = vqs1 & HWM;
    vqs1_r = vqs1 << 8;
    vqs1_r = vqs1_r & HWM;

    vqs2_l = vqs2 & HWM;
    vqs2_r = vqs2 << 8;
    vqs2_r = vqs2_r & HWM;

    __asm__ __volatile__ (
        /* qs0 - ps0 */
        "subq_s.ph    %[subr_l],       %[vqs0_l],       %[vps0_l]       \n\t"
        "subq_s.ph    %[subr_r],       %[vqs0_r],       %[vps0_r]       \n\t"

        /* vp8_filter = vp8_signed_char_clamp(ps1 - qs1); */
        "subq_s.ph    %[vp8_filter_l], %[vps1_l],       %[vqs1_l]       \n\t"
        "subq_s.ph    %[vp8_filter_r], %[vps1_r],       %[vqs1_r]       \n\t"

        : [vp8_filter_l] "=&r" (vp8_filter_l), [vp8_filter_r] "=r" (vp8_filter_r),
          [subr_l] "=&r" (subr_l), [subr_r] "=&r" (subr_r)
        : [vps0_l] "r" (vps0_l), [vps0_r] "r" (vps0_r), [vps1_l] "r" (vps1_l),
          [vps1_r] "r" (vps1_r), [vqs0_l] "r" (vqs0_l), [vqs0_r] "r" (vqs0_r),
          [vqs1_l] "r" (vqs1_l), [vqs1_r] "r" (vqs1_r)
    );

    vps2_l = vps2 & HWM;
    vps2_r = vps2 << 8;
    vps2_r = vps2_r & HWM;

    /* add outer taps if we have high edge variance */
    __asm__ __volatile__ (
        /* vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * (qs0 - ps0)); */
        "addq_s.ph    %[vp8_filter_l], %[vp8_filter_l], %[subr_l]       \n\t"
        "addq_s.ph    %[vp8_filter_r], %[vp8_filter_r], %[subr_r]       \n\t"
        "and          %[mask_l],       %[HWM],          %[mask]         \n\t"
        "sll          %[mask_r],       %[mask],         8               \n\t"
        "and          %[mask_r],       %[HWM],          %[mask_r]       \n\t"
        "addq_s.ph    %[vp8_filter_l], %[vp8_filter_l], %[subr_l]       \n\t"
        "addq_s.ph    %[vp8_filter_r], %[vp8_filter_r], %[subr_r]       \n\t"
        "and          %[hev_l],        %[HWM],          %[hev]          \n\t"
        "sll          %[hev_r],        %[hev],          8               \n\t"
        "and          %[hev_r],        %[HWM],          %[hev_r]        \n\t"
        "addq_s.ph    %[vp8_filter_l], %[vp8_filter_l], %[subr_l]       \n\t"
        "addq_s.ph    %[vp8_filter_r], %[vp8_filter_r], %[subr_r]       \n\t"

        /* vp8_filter &= mask; */
        "and          %[vp8_filter_l], %[vp8_filter_l], %[mask_l]       \n\t"
        "and          %[vp8_filter_r], %[vp8_filter_r], %[mask_r]       \n\t"

        /* Filter2 = vp8_filter & hev; */
        "and          %[Filter2_l],    %[vp8_filter_l], %[hev_l]        \n\t"
        "and          %[Filter2_r],    %[vp8_filter_r], %[hev_r]        \n\t"

        : [vp8_filter_l] "+r" (vp8_filter_l), [vp8_filter_r] "+r" (vp8_filter_r),
          [hev_l] "=&r" (hev_l), [hev_r] "=&r" (hev_r),
          [mask_l] "=&r" (mask_l), [mask_r] "=&r" (mask_r),
          [Filter2_l] "=&r" (Filter2_l), [Filter2_r] "=&r" (Filter2_r)
        : [subr_l] "r" (subr_l), [subr_r] "r" (subr_r),
          [HWM] "r" (HWM), [hev]  "r" (hev), [mask] "r" (mask)
    );

    /* save bottom 3 bits so that we round one side +4 and the other +3 */
    __asm__ __volatile__ (
        /* Filter1 = vp8_signed_char_clamp(Filter2 + 4) >>= 3; */
        "addq_s.ph    %[Filter1_l],    %[Filter2_l],    %[t2]           \n\t"
        "xor          %[invhev_l],     %[hev_l],        %[HWM]          \n\t"
        "addq_s.ph    %[Filter1_r],    %[Filter2_r],    %[t2]           \n\t"

        /* Filter2 = vp8_signed_char_clamp(Filter2 + 3) >>= 3; */
        "addq_s.ph    %[Filter2_l],    %[Filter2_l],    %[t1]           \n\t"
        "addq_s.ph    %[Filter2_r],    %[Filter2_r],    %[t1]           \n\t"

        "shra.ph      %[Filter1_l],    %[Filter1_l],    3               \n\t"
        "shra.ph      %[Filter1_r],    %[Filter1_r],    3               \n\t"

        "shra.ph      %[Filter2_l],    %[Filter2_l],    3               \n\t"
        "shra.ph      %[Filter2_r],    %[Filter2_r],    3               \n\t"
        "and          %[Filter1_l],    %[Filter1_l],    %[HWM]          \n\t"
        "and          %[Filter1_r],    %[Filter1_r],    %[HWM]          \n\t"
        "xor          %[invhev_r],     %[hev_r],        %[HWM]          \n\t"

        /* qs0 = vp8_signed_char_clamp(qs0 - Filter1); */
        "subq_s.ph    %[vqs0_l],       %[vqs0_l],       %[Filter1_l]    \n\t"
        "subq_s.ph    %[vqs0_r],       %[vqs0_r],       %[Filter1_r]    \n\t"

        /* ps0 = vp8_signed_char_clamp(ps0 + Filter2); */
        "addq_s.ph    %[vps0_l],       %[vps0_l],       %[Filter2_l]    \n\t"
        "addq_s.ph    %[vps0_r],       %[vps0_r],       %[Filter2_r]    \n\t"

        : [invhev_l] "=&r" (invhev_l), [invhev_r] "=&r" (invhev_r),
          [Filter1_l] "=&r" (Filter1_l), [Filter1_r] "=&r" (Filter1_r),
          [Filter2_l] "+r" (Filter2_l), [Filter2_r] "+r" (Filter2_r),
          [vps0_l] "+r" (vps0_l), [vps0_r] "+r" (vps0_r),
          [vqs0_l] "+r" (vqs0_l), [vqs0_r] "+r" (vqs0_r)
        : [t1] "r" (t1), [t2] "r" (t2), [HWM] "r" (HWM),
          [hev_l] "r" (hev_l), [hev_r] "r" (hev_r)
    );

    /* only apply wider filter if not high edge variance */
    __asm__ __volatile__ (
        /* vp8_filter &= ~hev; */
        "and          %[Filter2_l],    %[vp8_filter_l], %[invhev_l]     \n\t"
        "and          %[Filter2_r],    %[vp8_filter_r], %[invhev_r]     \n\t"

        "shra.ph      %[Filter2_l],    %[Filter2_l],    8               \n\t"
        "shra.ph      %[Filter2_r],    %[Filter2_r],    8               \n\t"

        : [Filter2_l] "=&r" (Filter2_l), [Filter2_r] "=&r" (Filter2_r)
        : [vp8_filter_l] "r" (vp8_filter_l), [vp8_filter_r] "r" (vp8_filter_r),
          [invhev_l] "r" (invhev_l), [invhev_r] "r" (invhev_r)
    );

    /* roughly 3/7th difference across boundary */
    __asm__ __volatile__ (
        "shll.ph      %[u3_l],         %[Filter2_l],    3               \n\t"
        "shll.ph      %[u3_r],         %[Filter2_r],    3               \n\t"

        "addq.ph      %[u3_l],         %[u3_l],         %[Filter2_l]    \n\t"
        "addq.ph      %[u3_r],         %[u3_r],         %[Filter2_r]    \n\t"

        "shll.ph      %[u2_l],         %[u3_l],         1               \n\t"
        "shll.ph      %[u2_r],         %[u3_r],         1               \n\t"

        "addq.ph      %[u1_l],         %[u3_l],         %[u2_l]         \n\t"
        "addq.ph      %[u1_r],         %[u3_r],         %[u2_r]         \n\t"

        "addq.ph      %[u2_l],         %[u2_l],         %[R63]          \n\t"
        "addq.ph      %[u2_r],         %[u2_r],         %[R63]          \n\t"

        "addq.ph      %[u3_l],         %[u3_l],         %[R63]          \n\t"
        "addq.ph      %[u3_r],         %[u3_r],         %[R63]          \n\t"

        /* vp8_signed_char_clamp((63 + Filter2 * 27) >> 7)
         * vp8_signed_char_clamp((63 + Filter2 * 18) >> 7)
         */
        "addq.ph      %[u1_l],         %[u1_l],         %[R63]          \n\t"
        "addq.ph      %[u1_r],         %[u1_r],         %[R63]          \n\t"
        "shra.ph      %[u1_l],         %[u1_l],         7               \n\t"
        "shra.ph      %[u1_r],         %[u1_r],         7               \n\t"
        "shra.ph      %[u2_l],         %[u2_l],         7               \n\t"
        "shra.ph      %[u2_r],         %[u2_r],         7               \n\t"
        "shll.ph      %[u1_l],         %[u1_l],         8               \n\t"
        "shll.ph      %[u1_r],         %[u1_r],         8               \n\t"
        "shll.ph      %[u2_l],         %[u2_l],         8               \n\t"
        "shll.ph      %[u2_r],         %[u2_r],         8               \n\t"

        /* vqs0 = vp8_signed_char_clamp(qs0 - u); */
        "subq_s.ph    %[vqs0_l],       %[vqs0_l],       %[u1_l]         \n\t"
        "subq_s.ph    %[vqs0_r],       %[vqs0_r],       %[u1_r]         \n\t"

        /* vps0 = vp8_signed_char_clamp(ps0 + u); */
        "addq_s.ph    %[vps0_l],       %[vps0_l],       %[u1_l]         \n\t"
        "addq_s.ph    %[vps0_r],       %[vps0_r],       %[u1_r]         \n\t"

        : [u1_l] "=&r" (u1_l), [u1_r] "=&r" (u1_r), [u2_l] "=&r" (u2_l),
          [u2_r] "=&r" (u2_r), [u3_l] "=&r" (u3_l), [u3_r] "=&r" (u3_r),
          [vps0_l] "+r" (vps0_l), [vps0_r] "+r" (vps0_r),
          [vqs0_l] "+r" (vqs0_l), [vqs0_r] "+r" (vqs0_r)
        : [R63]  "r" (R63),
          [Filter2_l] "r" (Filter2_l), [Filter2_r] "r" (Filter2_r)
    );

    __asm__ __volatile__ (
        /* vqs1 = vp8_signed_char_clamp(qs1 - u); */
        "subq_s.ph    %[vqs1_l],       %[vqs1_l],       %[u2_l]         \n\t"
        "addq_s.ph    %[vps1_l],       %[vps1_l],       %[u2_l]         \n\t"

        /* vps1 = vp8_signed_char_clamp(ps1 + u); */
        "addq_s.ph    %[vps1_r],       %[vps1_r],       %[u2_r]         \n\t"
        "subq_s.ph    %[vqs1_r],       %[vqs1_r],       %[u2_r]         \n\t"

        : [vps1_l] "+r" (vps1_l), [vps1_r] "+r" (vps1_r),
          [vqs1_l] "+r" (vqs1_l), [vqs1_r] "+r" (vqs1_r)
        : [u2_l] "r" (u2_l), [u2_r] "r" (u2_r)
    );

    /* roughly 1/7th difference across boundary */
    __asm__ __volatile__ (
        /* u = vp8_signed_char_clamp((63 + Filter2 * 9) >> 7); */
        "shra.ph      %[u3_l],         %[u3_l],         7               \n\t"
        "shra.ph      %[u3_r],         %[u3_r],         7               \n\t"
        "shll.ph      %[u3_l],         %[u3_l],         8               \n\t"
        "shll.ph      %[u3_r],         %[u3_r],         8               \n\t"

        /* vqs2 = vp8_signed_char_clamp(qs2 - u); */
        "subq_s.ph    %[vqs2_l],       %[vqs2_l],       %[u3_l]         \n\t"
        "subq_s.ph    %[vqs2_r],       %[vqs2_r],       %[u3_r]         \n\t"

        /* vps2 = vp8_signed_char_clamp(ps2 + u); */
        "addq_s.ph    %[vps2_l],       %[vps2_l],       %[u3_l]         \n\t"
        "addq_s.ph    %[vps2_r],       %[vps2_r],       %[u3_r]         \n\t"

        : [u3_l] "+r" (u3_l), [u3_r] "+r" (u3_r), [vps2_l] "+r" (vps2_l),
          [vps2_r] "+r" (vps2_r), [vqs2_l] "+r" (vqs2_l), [vqs2_r] "+r" (vqs2_r)
        :
    );

    /* Create quad-bytes from halfword pairs */
    __asm__ __volatile__ (
        "and          %[vqs0_l],       %[vqs0_l],       %[HWM]          \n\t"
        "shrl.ph      %[vqs0_r],       %[vqs0_r],       8               \n\t"

        "and          %[vps0_l],       %[vps0_l],       %[HWM]          \n\t"
        "shrl.ph      %[vps0_r],       %[vps0_r],       8               \n\t"

        "and          %[vqs1_l],       %[vqs1_l],       %[HWM]          \n\t"
        "shrl.ph      %[vqs1_r],       %[vqs1_r],       8               \n\t"

        "and          %[vps1_l],       %[vps1_l],       %[HWM]          \n\t"
        "shrl.ph      %[vps1_r],       %[vps1_r],       8               \n\t"

        "and          %[vqs2_l],       %[vqs2_l],       %[HWM]          \n\t"
        "shrl.ph      %[vqs2_r],       %[vqs2_r],       8               \n\t"

        "and          %[vps2_l],       %[vps2_l],       %[HWM]          \n\t"
        "shrl.ph      %[vps2_r],       %[vps2_r],       8               \n\t"

        "or           %[vqs0_r],       %[vqs0_l],       %[vqs0_r]       \n\t"
        "or           %[vps0_r],       %[vps0_l],       %[vps0_r]       \n\t"
        "or           %[vqs1_r],       %[vqs1_l],       %[vqs1_r]       \n\t"
        "or           %[vps1_r],       %[vps1_l],       %[vps1_r]       \n\t"
        "or           %[vqs2_r],       %[vqs2_l],       %[vqs2_r]       \n\t"
        "or           %[vps2_r],       %[vps2_l],       %[vps2_r]       \n\t"

        : [vps1_l] "+r" (vps1_l), [vps1_r] "+r" (vps1_r), [vqs1_l] "+r" (vqs1_l),
          [vqs1_r] "+r" (vqs1_r), [vps0_l] "+r" (vps0_l), [vps0_r] "+r" (vps0_r),
          [vqs0_l] "+r" (vqs0_l), [vqs0_r] "+r" (vqs0_r), [vqs2_l] "+r" (vqs2_l),
          [vqs2_r] "+r" (vqs2_r), [vps2_r] "+r" (vps2_r), [vps2_l] "+r" (vps2_l)
        : [HWM] "r" (HWM)
    );

    *ps0 = vps0_r ^ N128;
    *ps1 = vps1_r ^ N128;
    *ps2 = vps2_r ^ N128;
    *qs0 = vqs0_r ^ N128;
    *qs1 = vqs1_r ^ N128;
    *qs2 = vqs2_r ^ N128;
}

void vp8_mbloop_filter_horizontal_edge_mips
(
    unsigned char *s,
    int p,
    unsigned int flimit,
    unsigned int limit,
    unsigned int thresh,
    int count
)
{
    int i;
    uint32_t mask, hev;
    uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
    unsigned char *sm1, *s0, *s1, *s2, *s3, *s4, *s5, *s6;

    mask = 0;
    hev = 0;
    i = 0;
    p1 = 0;
    p2 = 0;
    p3 = 0;
    p4 = 0;

    /* loop filter designed to work using chars so that we can make maximum use
     * of 8 bit simd instructions.
     */

    sm1 = s - (p << 2);
    s0  = s - p - p - p;
    s1  = s - p - p;
    s2  = s - p;
    s3  = s;
    s4  = s + p;
    s5  = s + p + p;
    s6  = s + p + p + p;

    /* prefetch data for load */
    prefetch_load_lf(s + p);

    /* apply filter on 4 pixesl at the same time */
    do
    {
        /* load quad-byte vectors
         * memory is 4 byte aligned
         */
        p1 = *((uint32_t *)(s1));
        p2 = *((uint32_t *)(s2));
        p3 = *((uint32_t *)(s3));
        p4 = *((uint32_t *)(s4));

        /* if (p1 - p4 == 0) and (p2 - p3 == 0)
         * mask will be zero and filtering is not needed
         */
        if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
        {

            pm1 = *((uint32_t *)(sm1));
            p0  = *((uint32_t *)(s0));
            p5  = *((uint32_t *)(s5));
            p6  = *((uint32_t *)(s6));

            vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
                                     thresh, &hev, &mask);

            /* if mask == 0 do filtering is not needed */
            if (mask)
            {
                /* filtering */
                vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);

                /* unpack processed 4x4 neighborhood
                 * memory is 4 byte aligned
                 */
                *((uint32_t *)s0) = p0;
                *((uint32_t *)s1) = p1;
                *((uint32_t *)s2) = p2;
                *((uint32_t *)s3) = p3;
                *((uint32_t *)s4) = p4;
                *((uint32_t *)s5) = p5;
            }
        }

        sm1 += 4;
        s0  += 4;
        s1  += 4;
        s2  += 4;
        s3  += 4;
        s4  += 4;
        s5  += 4;
        s6  += 4;

        /* load quad-byte vectors
         * memory is 4 byte aligned
         */
        p1 = *((uint32_t *)(s1));
        p2 = *((uint32_t *)(s2));
        p3 = *((uint32_t *)(s3));
        p4 = *((uint32_t *)(s4));

        /* if (p1 - p4 == 0) and (p2 - p3 == 0)
         * mask will be zero and filtering is not needed
         */
        if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
        {

            pm1 = *((uint32_t *)(sm1));
            p0  = *((uint32_t *)(s0));
            p5  = *((uint32_t *)(s5));
            p6  = *((uint32_t *)(s6));

            vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
                                     thresh, &hev, &mask);

            /* if mask == 0 do filtering is not needed */
            if (mask)
            {
                /* filtering */
                vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);

                /* unpack processed 4x4 neighborhood
                 * memory is 4 byte aligned
                 */
                *((uint32_t *)s0) = p0;
                *((uint32_t *)s1) = p1;
                *((uint32_t *)s2) = p2;
                *((uint32_t *)s3) = p3;
                *((uint32_t *)s4) = p4;
                *((uint32_t *)s5) = p5;
            }
        }

        sm1 += 4;
        s0  += 4;
        s1  += 4;
        s2  += 4;
        s3  += 4;
        s4  += 4;
        s5  += 4;
        s6  += 4;

        i += 8;
    }

    while (i < count);
}

void vp8_mbloop_filter_uvhorizontal_edge_mips
(
    unsigned char *s,
    int p,
    unsigned int flimit,
    unsigned int limit,
    unsigned int thresh,
    int count
)
{
    uint32_t mask, hev;
    uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
    unsigned char *sm1, *s0, *s1, *s2, *s3, *s4, *s5, *s6;

    mask = 0;
    hev = 0;
    p1 = 0;
    p2 = 0;
    p3 = 0;
    p4 = 0;

    /* loop filter designed to work using chars so that we can make maximum use
     * of 8 bit simd instructions.
     */

    sm1 = s - (p << 2);
    s0  = s - p - p - p;
    s1  = s - p - p;
    s2  = s - p;
    s3  = s;
    s4  = s + p;
    s5  = s + p + p;
    s6  = s + p + p + p;

    /* load quad-byte vectors
     * memory is 4 byte aligned
     */
    p1 = *((uint32_t *)(s1));
    p2 = *((uint32_t *)(s2));
    p3 = *((uint32_t *)(s3));
    p4 = *((uint32_t *)(s4));

    /* if (p1 - p4 == 0) and (p2 - p3 == 0)
     * mask will be zero and filtering is not needed
     */
    if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
    {

        pm1 = *((uint32_t *)(sm1));
        p0  = *((uint32_t *)(s0));
        p5  = *((uint32_t *)(s5));
        p6  = *((uint32_t *)(s6));

        /* if mask == 0 do filtering is not needed */
        vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
                                 thresh, &hev, &mask);

        if (mask)
        {
            /* filtering */
            vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);

            /* unpack processed 4x4 neighborhood
             * memory is 4 byte aligned
             */
            *((uint32_t *)s0) = p0;
            *((uint32_t *)s1) = p1;
            *((uint32_t *)s2) = p2;
            *((uint32_t *)s3) = p3;
            *((uint32_t *)s4) = p4;
            *((uint32_t *)s5) = p5;
        }
    }

    sm1 += 4;
    s0  += 4;
    s1  += 4;
    s2  += 4;
    s3  += 4;
    s4  += 4;
    s5  += 4;
    s6  += 4;

    /* load quad-byte vectors
     * memory is 4 byte aligned
     */
    p1 = *((uint32_t *)(s1));
    p2 = *((uint32_t *)(s2));
    p3 = *((uint32_t *)(s3));
    p4 = *((uint32_t *)(s4));

    /* if (p1 - p4 == 0) and (p2 - p3 == 0)
     * mask will be zero and filtering is not needed
     */
    if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
    {

        pm1 = *((uint32_t *)(sm1));
        p0  = *((uint32_t *)(s0));
        p5  = *((uint32_t *)(s5));
        p6  = *((uint32_t *)(s6));

        vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
                                 thresh, &hev, &mask);

        /* if mask == 0 do filtering is not needed */
        if (mask)
        {
            /* filtering */
            vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);

            /* unpack processed 4x4 neighborhood
             * memory is 4 byte aligned
             */
            *((uint32_t *)s0) = p0;
            *((uint32_t *)s1) = p1;
            *((uint32_t *)s2) = p2;
            *((uint32_t *)s3) = p3;
            *((uint32_t *)s4) = p4;
            *((uint32_t *)s5) = p5;
        }
    }
}


void vp8_mbloop_filter_vertical_edge_mips
(
    unsigned char *s,
    int p,
    unsigned int flimit,
    unsigned int limit,
    unsigned int thresh,
    int count
)
{

    int i;
    uint32_t mask, hev;
    uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
    unsigned char *s1, *s2, *s3, *s4;
    uint32_t prim1, prim2, sec3, sec4, prim3, prim4;

    mask = 0;
    hev = 0;
    i = 0;
    pm1 = 0;
    p0 = 0;
    p1 = 0;
    p2 = 0;
    p3 = 0;
    p4 = 0;
    p5 = 0;
    p6 = 0;

    /* loop filter designed to work using chars so that we can make maximum use
     * of 8 bit simd instructions.
     */

    /* apply filter on 4 pixesl at the same time */
    do
    {
        s1 = s;
        s2 = s + p;
        s3 = s2 + p;
        s4 = s3 + p;
        s  = s4 + p;

        /* load quad-byte vectors
         * memory is 4 byte aligned
         */
        p2  = *((uint32_t *)(s1 - 4));
        p6  = *((uint32_t *)(s1));
        p1  = *((uint32_t *)(s2 - 4));
        p5  = *((uint32_t *)(s2));
        p0  = *((uint32_t *)(s3 - 4));
        p4  = *((uint32_t *)(s3));
        pm1 = *((uint32_t *)(s4 - 4));
        p3  = *((uint32_t *)(s4));

        /* transpose pm1, p0, p1, p2 */
        __asm__ __volatile__ (
            "precrq.qb.ph   %[prim1],   %[p2],      %[p1]       \n\t"
            "precr.qb.ph    %[prim2],   %[p2],      %[p1]       \n\t"
            "precrq.qb.ph   %[prim3],   %[p0],      %[pm1]      \n\t"
            "precr.qb.ph    %[prim4],   %[p0],      %[pm1]      \n\t"

            "precrq.qb.ph   %[p1],      %[prim1],   %[prim2]    \n\t"
            "precr.qb.ph    %[pm1],     %[prim1],   %[prim2]    \n\t"
            "precrq.qb.ph   %[sec3],    %[prim3],   %[prim4]    \n\t"
            "precr.qb.ph    %[sec4],    %[prim3],   %[prim4]    \n\t"

            "precrq.ph.w    %[p2],      %[p1],      %[sec3]     \n\t"
            "precrq.ph.w    %[p0],      %[pm1],     %[sec4]     \n\t"
            "append         %[p1],      %[sec3],    16          \n\t"
            "append         %[pm1],     %[sec4],    16          \n\t"

            : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
              [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
              [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
              [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
            :
        );

        /* transpose p3, p4, p5, p6 */
        __asm__ __volatile__ (
            "precrq.qb.ph   %[prim1],   %[p6],      %[p5]       \n\t"
            "precr.qb.ph    %[prim2],   %[p6],      %[p5]       \n\t"
            "precrq.qb.ph   %[prim3],   %[p4],      %[p3]       \n\t"
            "precr.qb.ph    %[prim4],   %[p4],      %[p3]       \n\t"

            "precrq.qb.ph   %[p5],      %[prim1],   %[prim2]    \n\t"
            "precr.qb.ph    %[p3],      %[prim1],   %[prim2]    \n\t"
            "precrq.qb.ph   %[sec3],    %[prim3],   %[prim4]    \n\t"
            "precr.qb.ph    %[sec4],    %[prim3],   %[prim4]    \n\t"

            "precrq.ph.w    %[p6],      %[p5],      %[sec3]     \n\t"
            "precrq.ph.w    %[p4],      %[p3],      %[sec4]     \n\t"
            "append         %[p5],      %[sec3],    16          \n\t"
            "append         %[p3],      %[sec4],    16          \n\t"

            : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
              [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
              [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
              [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
            :
        );

        /* if (p1 - p4 == 0) and (p2 - p3 == 0)
         * mask will be zero and filtering is not needed
         */
        if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
        {

            vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
                                     thresh, &hev, &mask);

            /* if mask == 0 do filtering is not needed */
            if (mask)
            {
                /* filtering */
                vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);

                /* don't use transpose on output data
                 * because memory isn't aligned
                 */
                __asm__ __volatile__ (
                    "sb         %[p5],  2(%[s4])        \n\t"
                    "sb         %[p4],  1(%[s4])        \n\t"
                    "sb         %[p3],  0(%[s4])        \n\t"
                    "sb         %[p2], -1(%[s4])        \n\t"
                    "sb         %[p1], -2(%[s4])        \n\t"
                    "sb         %[p0], -3(%[s4])        \n\t"
                    :
                    : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
                      [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
                );

                __asm__ __volatile__ (
                    "srl        %[p5], %[p5], 8         \n\t"
                    "srl        %[p4], %[p4], 8         \n\t"
                    "srl        %[p3], %[p3], 8         \n\t"
                    "srl        %[p2], %[p2], 8         \n\t"
                    "srl        %[p1], %[p1], 8         \n\t"
                    "srl        %[p0], %[p0], 8         \n\t"
                    : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
                      [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
                    :
                );

                __asm__ __volatile__ (
                    "sb         %[p5],  2(%[s3])        \n\t"
                    "sb         %[p4],  1(%[s3])        \n\t"
                    "sb         %[p3],  0(%[s3])        \n\t"
                    "sb         %[p2], -1(%[s3])        \n\t"
                    "sb         %[p1], -2(%[s3])        \n\t"
                    "sb         %[p0], -3(%[s3])        \n\t"
                    :
                    : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3),
                      [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
                );

                __asm__ __volatile__ (
                    "srl        %[p5], %[p5], 8         \n\t"
                    "srl        %[p4], %[p4], 8         \n\t"
                    "srl        %[p3], %[p3], 8         \n\t"
                    "srl        %[p2], %[p2], 8         \n\t"
                    "srl        %[p1], %[p1], 8         \n\t"
                    "srl        %[p0], %[p0], 8         \n\t"
                    : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
                      [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
                    :
                );

                __asm__ __volatile__ (
                    "sb         %[p5],  2(%[s2])        \n\t"
                    "sb         %[p4],  1(%[s2])        \n\t"
                    "sb         %[p3],  0(%[s2])        \n\t"
                    "sb         %[p2], -1(%[s2])        \n\t"
                    "sb         %[p1], -2(%[s2])        \n\t"
                    "sb         %[p0], -3(%[s2])        \n\t"
                    :
                    : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
                      [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
                );

                __asm__ __volatile__ (
                    "srl        %[p5], %[p5], 8         \n\t"
                    "srl        %[p4], %[p4], 8         \n\t"
                    "srl        %[p3], %[p3], 8         \n\t"
                    "srl        %[p2], %[p2], 8         \n\t"
                    "srl        %[p1], %[p1], 8         \n\t"
                    "srl        %[p0], %[p0], 8         \n\t"
                    : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
                      [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
                    :
                );

                __asm__ __volatile__ (
                    "sb         %[p5],  2(%[s1])        \n\t"
                    "sb         %[p4],  1(%[s1])        \n\t"
                    "sb         %[p3],  0(%[s1])        \n\t"
                    "sb         %[p2], -1(%[s1])        \n\t"
                    "sb         %[p1], -2(%[s1])        \n\t"
                    "sb         %[p0], -3(%[s1])        \n\t"
                    :
                    : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
                      [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
                );
            }
        }

        i += 4;
    }

    while (i < count);
}

void vp8_mbloop_filter_uvvertical_edge_mips
(
    unsigned char *s,
    int p,
    unsigned int flimit,
    unsigned int limit,
    unsigned int thresh,
    int count
)
{
    uint32_t mask, hev;
    uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
    unsigned char *s1, *s2, *s3, *s4;
    uint32_t prim1, prim2, sec3, sec4, prim3, prim4;

    mask = 0;
    hev = 0;
    pm1 = 0;
    p0 = 0;
    p1 = 0;
    p2 = 0;
    p3 = 0;
    p4 = 0;
    p5 = 0;
    p6 = 0;

    /* loop filter designed to work using chars so that we can make maximum use
     * of 8 bit simd instructions.
     */

    /* apply filter on 4 pixesl at the same time */

    s1 = s;
    s2 = s + p;
    s3 = s2 + p;
    s4 = s3 + p;

    /* prefetch data for load */
    prefetch_load_lf(s + 2 * p);

    /* load quad-byte vectors
     * memory is 4 byte aligned
     */
    p2  = *((uint32_t *)(s1 - 4));
    p6  = *((uint32_t *)(s1));
    p1  = *((uint32_t *)(s2 - 4));
    p5  = *((uint32_t *)(s2));
    p0  = *((uint32_t *)(s3 - 4));
    p4  = *((uint32_t *)(s3));
    pm1 = *((uint32_t *)(s4 - 4));
    p3  = *((uint32_t *)(s4));

    /* transpose pm1, p0, p1, p2 */
    __asm__ __volatile__ (
        "precrq.qb.ph   %[prim1],   %[p2],      %[p1]       \n\t"
        "precr.qb.ph    %[prim2],   %[p2],      %[p1]       \n\t"
        "precrq.qb.ph   %[prim3],   %[p0],      %[pm1]      \n\t"
        "precr.qb.ph    %[prim4],   %[p0],      %[pm1]      \n\t"

        "precrq.qb.ph   %[p1],      %[prim1],   %[prim2]    \n\t"
        "precr.qb.ph    %[pm1],     %[prim1],   %[prim2]    \n\t"
        "precrq.qb.ph   %[sec3],    %[prim3],   %[prim4]    \n\t"
        "precr.qb.ph    %[sec4],    %[prim3],   %[prim4]    \n\t"

        "precrq.ph.w    %[p2],      %[p1],      %[sec3]     \n\t"
        "precrq.ph.w    %[p0],      %[pm1],     %[sec4]     \n\t"
        "append         %[p1],      %[sec3],    16          \n\t"
        "append         %[pm1],     %[sec4],    16          \n\t"

        : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
          [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
          [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
          [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
        :
    );

    /* transpose p3, p4, p5, p6 */
    __asm__ __volatile__ (
        "precrq.qb.ph   %[prim1],   %[p6],      %[p5]       \n\t"
        "precr.qb.ph    %[prim2],   %[p6],      %[p5]       \n\t"
        "precrq.qb.ph   %[prim3],   %[p4],      %[p3]       \n\t"
        "precr.qb.ph    %[prim4],   %[p4],      %[p3]       \n\t"

        "precrq.qb.ph   %[p5],      %[prim1],   %[prim2]    \n\t"
        "precr.qb.ph    %[p3],      %[prim1],   %[prim2]    \n\t"
        "precrq.qb.ph   %[sec3],    %[prim3],   %[prim4]    \n\t"
        "precr.qb.ph    %[sec4],    %[prim3],   %[prim4]    \n\t"

        "precrq.ph.w    %[p6],      %[p5],      %[sec3]     \n\t"
        "precrq.ph.w    %[p4],      %[p3],      %[sec4]     \n\t"
        "append         %[p5],      %[sec3],    16          \n\t"
        "append         %[p3],      %[sec4],    16          \n\t"

        : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
          [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
          [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
          [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
        :
    );

    /* if (p1 - p4 == 0) and (p2 - p3 == 0)
     * mask will be zero and filtering is not needed
     */
    if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
    {

        vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
                                 thresh, &hev, &mask);

        /* if mask == 0 do filtering is not needed */
        if (mask)
        {
            /* filtering */
            vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);

            /* don't use transpose on output data
             * because memory isn't aligned
             */
            __asm__ __volatile__ (
                "sb         %[p5],  2(%[s4])        \n\t"
                "sb         %[p4],  1(%[s4])        \n\t"
                "sb         %[p3],  0(%[s4])        \n\t"
                "sb         %[p2], -1(%[s4])        \n\t"
                "sb         %[p1], -2(%[s4])        \n\t"
                "sb         %[p0], -3(%[s4])        \n\t"
                :
                : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
                  [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
            );

            __asm__ __volatile__ (
                "srl        %[p5], %[p5], 8         \n\t"
                "srl        %[p4], %[p4], 8         \n\t"
                "srl        %[p3], %[p3], 8         \n\t"
                "srl        %[p2], %[p2], 8         \n\t"
                "srl        %[p1], %[p1], 8         \n\t"
                "srl        %[p0], %[p0], 8         \n\t"
                : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
                  [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
                :
            );

            __asm__ __volatile__ (
                "sb         %[p5],  2(%[s3])        \n\t"
                "sb         %[p4],  1(%[s3])        \n\t"
                "sb         %[p3],  0(%[s3])        \n\t"
                "sb         %[p2], -1(%[s3])        \n\t"
                "sb         %[p1], -2(%[s3])        \n\t"
                "sb         %[p0], -3(%[s3])        \n\t"
                :
                : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3),
                  [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
            );

            __asm__ __volatile__ (
                "srl        %[p5], %[p5], 8         \n\t"
                "srl        %[p4], %[p4], 8         \n\t"
                "srl        %[p3], %[p3], 8         \n\t"
                "srl        %[p2], %[p2], 8         \n\t"
                "srl        %[p1], %[p1], 8         \n\t"
                "srl        %[p0], %[p0], 8         \n\t"
                : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
                  [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
                :
            );

            __asm__ __volatile__ (
                "sb         %[p5],  2(%[s2])        \n\t"
                "sb         %[p4],  1(%[s2])        \n\t"
                "sb         %[p3],  0(%[s2])        \n\t"
                "sb         %[p2], -1(%[s2])        \n\t"
                "sb         %[p1], -2(%[s2])        \n\t"
                "sb         %[p0], -3(%[s2])        \n\t"
                :
                : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
                  [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
            );

            __asm__ __volatile__ (
                "srl        %[p5], %[p5], 8         \n\t"
                "srl        %[p4], %[p4], 8         \n\t"
                "srl        %[p3], %[p3], 8         \n\t"
                "srl        %[p2], %[p2], 8         \n\t"
                "srl        %[p1], %[p1], 8         \n\t"
                "srl        %[p0], %[p0], 8         \n\t"
                : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
                  [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
                :
            );

            __asm__ __volatile__ (
                "sb         %[p5],  2(%[s1])        \n\t"
                "sb         %[p4],  1(%[s1])        \n\t"
                "sb         %[p3],  0(%[s1])        \n\t"
                "sb         %[p2], -1(%[s1])        \n\t"
                "sb         %[p1], -2(%[s1])        \n\t"
                "sb         %[p0], -3(%[s1])        \n\t"
                :
                : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
                  [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
            );
        }
    }

    s1 = s4 + p;
    s2 = s1 + p;
    s3 = s2 + p;
    s4 = s3 + p;

    /* load quad-byte vectors
    * memory is 4 byte aligned
    */
    p2  = *((uint32_t *)(s1 - 4));
    p6  = *((uint32_t *)(s1));
    p1  = *((uint32_t *)(s2 - 4));
    p5  = *((uint32_t *)(s2));
    p0  = *((uint32_t *)(s3 - 4));
    p4  = *((uint32_t *)(s3));
    pm1 = *((uint32_t *)(s4 - 4));
    p3  = *((uint32_t *)(s4));

    /* transpose pm1, p0, p1, p2 */
    __asm__ __volatile__ (
        "precrq.qb.ph   %[prim1],   %[p2],      %[p1]       \n\t"
        "precr.qb.ph    %[prim2],   %[p2],      %[p1]       \n\t"
        "precrq.qb.ph   %[prim3],   %[p0],      %[pm1]      \n\t"
        "precr.qb.ph    %[prim4],   %[p0],      %[pm1]      \n\t"

        "precrq.qb.ph   %[p1],      %[prim1],   %[prim2]    \n\t"
        "precr.qb.ph    %[pm1],     %[prim1],   %[prim2]    \n\t"
        "precrq.qb.ph   %[sec3],    %[prim3],   %[prim4]    \n\t"
        "precr.qb.ph    %[sec4],    %[prim3],   %[prim4]    \n\t"

        "precrq.ph.w    %[p2],      %[p1],      %[sec3]     \n\t"
        "precrq.ph.w    %[p0],      %[pm1],     %[sec4]     \n\t"
        "append         %[p1],      %[sec3],    16          \n\t"
        "append         %[pm1],     %[sec4],    16          \n\t"

        : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
          [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
          [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
          [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
        :
    );

    /* transpose p3, p4, p5, p6 */
    __asm__ __volatile__ (
        "precrq.qb.ph   %[prim1],   %[p6],      %[p5]       \n\t"
        "precr.qb.ph    %[prim2],   %[p6],      %[p5]       \n\t"
        "precrq.qb.ph   %[prim3],   %[p4],      %[p3]       \n\t"
        "precr.qb.ph    %[prim4],   %[p4],      %[p3]       \n\t"

        "precrq.qb.ph   %[p5],      %[prim1],   %[prim2]    \n\t"
        "precr.qb.ph    %[p3],      %[prim1],   %[prim2]    \n\t"
        "precrq.qb.ph   %[sec3],    %[prim3],   %[prim4]    \n\t"
        "precr.qb.ph    %[sec4],    %[prim3],   %[prim4]    \n\t"

        "precrq.ph.w    %[p6],      %[p5],      %[sec3]     \n\t"
        "precrq.ph.w    %[p4],      %[p3],      %[sec4]     \n\t"
        "append         %[p5],      %[sec3],    16          \n\t"
        "append         %[p3],      %[sec4],    16          \n\t"

        : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
          [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
          [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
          [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
        :
    );

    /* if (p1 - p4 == 0) and (p2 - p3 == 0)
     * mask will be zero and filtering is not needed
     */
    if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
    {

        vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6, thresh, &hev, &mask);

        /* if mask == 0 do filtering is not needed */
        if (mask)
        {
            /* filtering */
            vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);

            /* don't use transpose on output data
             * because memory isn't aligned
             */
            __asm__ __volatile__ (
                "sb         %[p5],  2(%[s4])        \n\t"
                "sb         %[p4],  1(%[s4])        \n\t"
                "sb         %[p3],  0(%[s4])        \n\t"
                "sb         %[p2], -1(%[s4])        \n\t"
                "sb         %[p1], -2(%[s4])        \n\t"
                "sb         %[p0], -3(%[s4])        \n\t"
                :
                : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
                  [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
            );

            __asm__ __volatile__ (
                "srl        %[p5], %[p5], 8         \n\t"
                "srl        %[p4], %[p4], 8         \n\t"
                "srl        %[p3], %[p3], 8         \n\t"
                "srl        %[p2], %[p2], 8         \n\t"
                "srl        %[p1], %[p1], 8         \n\t"
                "srl        %[p0], %[p0], 8         \n\t"
                : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
                  [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
                :
            );

            __asm__ __volatile__ (
                "sb         %[p5],  2(%[s3])        \n\t"
                "sb         %[p4],  1(%[s3])        \n\t"
                "sb         %[p3],  0(%[s3])        \n\t"
                "sb         %[p2], -1(%[s3])        \n\t"
                "sb         %[p1], -2(%[s3])        \n\t"
                "sb         %[p0], -3(%[s3])        \n\t"
                :
                : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3),
                  [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
            );

            __asm__ __volatile__ (
                "srl        %[p5], %[p5], 8         \n\t"
                "srl        %[p4], %[p4], 8         \n\t"
                "srl        %[p3], %[p3], 8         \n\t"
                "srl        %[p2], %[p2], 8         \n\t"
                "srl        %[p1], %[p1], 8         \n\t"
                "srl        %[p0], %[p0], 8         \n\t"
                : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
                  [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
                :
            );

            __asm__ __volatile__ (
                "sb         %[p5],  2(%[s2])        \n\t"
                "sb         %[p4],  1(%[s2])        \n\t"
                "sb         %[p3],  0(%[s2])        \n\t"
                "sb         %[p2], -1(%[s2])        \n\t"
                "sb         %[p1], -2(%[s2])        \n\t"
                "sb         %[p0], -3(%[s2])        \n\t"
                :
                : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
                  [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
            );

            __asm__ __volatile__ (
                "srl        %[p5], %[p5], 8         \n\t"
                "srl        %[p4], %[p4], 8         \n\t"
                "srl        %[p3], %[p3], 8         \n\t"
                "srl        %[p2], %[p2], 8         \n\t"
                "srl        %[p1], %[p1], 8         \n\t"
                "srl        %[p0], %[p0], 8         \n\t"
                : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
                  [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
                :
            );

            __asm__ __volatile__ (
                "sb         %[p5],  2(%[s1])        \n\t"
                "sb         %[p4],  1(%[s1])        \n\t"
                "sb         %[p3],  0(%[s1])        \n\t"
                "sb         %[p2], -1(%[s1])        \n\t"
                "sb         %[p1], -2(%[s1])        \n\t"
                "sb         %[p0], -3(%[s1])        \n\t"
                :
                : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
                  [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
            );
        }
    }
}

/* Horizontal MB filtering */
void vp8_loop_filter_mbh_dspr2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
                               int y_stride, int uv_stride, loop_filter_info *lfi)
{
    unsigned int thresh_vec, flimit_vec, limit_vec;
    unsigned char thresh, flimit, limit, flimit_temp;

    /* use direct value instead pointers */
    limit = *(lfi->lim);
    flimit_temp = *(lfi->mblim);
    thresh = *(lfi->hev_thr);
    flimit = flimit_temp;

    /* create quad-byte */
    __asm__ __volatile__ (
        "replv.qb       %[thresh_vec], %[thresh]    \n\t"
        "replv.qb       %[flimit_vec], %[flimit]    \n\t"
        "replv.qb       %[limit_vec],  %[limit]     \n\t"
        : [thresh_vec] "=&r" (thresh_vec), [flimit_vec] "=&r" (flimit_vec), [limit_vec] "=r" (limit_vec)
        : [thresh] "r" (thresh), [flimit] "r" (flimit), [limit] "r" (limit)
    );

    vp8_mbloop_filter_horizontal_edge_mips(y_ptr, y_stride, flimit_vec, limit_vec, thresh_vec, 16);

    if (u_ptr)
    {
        vp8_mbloop_filter_uvhorizontal_edge_mips(u_ptr, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
    }

    if (v_ptr)
    {
        vp8_mbloop_filter_uvhorizontal_edge_mips(v_ptr, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
    }
}


/* Vertical MB Filtering */
void vp8_loop_filter_mbv_dspr2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
                               int y_stride, int uv_stride, loop_filter_info *lfi)
{
    unsigned int thresh_vec, flimit_vec, limit_vec;
    unsigned char thresh, flimit, limit, flimit_temp;

    /* use direct value instead pointers */
    limit = *(lfi->lim);
    flimit_temp = *(lfi->mblim);
    thresh = *(lfi->hev_thr);
    flimit = flimit_temp;

    /* create quad-byte */
    __asm__ __volatile__ (
        "replv.qb       %[thresh_vec], %[thresh]    \n\t"
        "replv.qb       %[flimit_vec], %[flimit]    \n\t"
        "replv.qb       %[limit_vec],  %[limit]     \n\t"
        : [thresh_vec] "=&r" (thresh_vec), [flimit_vec] "=&r" (flimit_vec), [limit_vec] "=r" (limit_vec)
        : [thresh] "r" (thresh), [flimit] "r" (flimit), [limit] "r" (limit)
    );

    vp8_mbloop_filter_vertical_edge_mips(y_ptr, y_stride, flimit_vec, limit_vec, thresh_vec, 16);

    if (u_ptr)
        vp8_mbloop_filter_uvvertical_edge_mips(u_ptr, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);

    if (v_ptr)
        vp8_mbloop_filter_uvvertical_edge_mips(v_ptr, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
}


/* Horizontal B Filtering */
void vp8_loop_filter_bh_dspr2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
                              int y_stride, int uv_stride, loop_filter_info *lfi)
{
    unsigned int thresh_vec, flimit_vec, limit_vec;
    unsigned char thresh, flimit, limit, flimit_temp;

    /* use direct value instead pointers */
    limit = *(lfi->lim);
    flimit_temp = *(lfi->blim);
    thresh = *(lfi->hev_thr);
    flimit = flimit_temp;

    /* create quad-byte */
    __asm__ __volatile__ (
        "replv.qb       %[thresh_vec], %[thresh]    \n\t"
        "replv.qb       %[flimit_vec], %[flimit]    \n\t"
        "replv.qb       %[limit_vec],  %[limit]     \n\t"
        : [thresh_vec] "=&r" (thresh_vec), [flimit_vec] "=&r" (flimit_vec), [limit_vec] "=r" (limit_vec)
        : [thresh] "r" (thresh), [flimit] "r" (flimit), [limit] "r" (limit)
    );

    vp8_loop_filter_horizontal_edge_mips(y_ptr + 4 * y_stride, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
    vp8_loop_filter_horizontal_edge_mips(y_ptr + 8 * y_stride, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
    vp8_loop_filter_horizontal_edge_mips(y_ptr + 12 * y_stride, y_stride, flimit_vec, limit_vec, thresh_vec, 16);

    if (u_ptr)
        vp8_loop_filter_uvhorizontal_edge_mips(u_ptr + 4 * uv_stride, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);

    if (v_ptr)
        vp8_loop_filter_uvhorizontal_edge_mips(v_ptr + 4 * uv_stride, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
}


/* Vertical B Filtering */
void vp8_loop_filter_bv_dspr2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
                              int y_stride, int uv_stride, loop_filter_info *lfi)
{
    unsigned int thresh_vec, flimit_vec, limit_vec;
    unsigned char thresh, flimit, limit, flimit_temp;

    /* use direct value instead pointers */
    limit = *(lfi->lim);
    flimit_temp = *(lfi->blim);
    thresh = *(lfi->hev_thr);
    flimit = flimit_temp;

    /* create quad-byte */
    __asm__ __volatile__ (
        "replv.qb       %[thresh_vec], %[thresh]    \n\t"
        "replv.qb       %[flimit_vec], %[flimit]    \n\t"
        "replv.qb       %[limit_vec],  %[limit]     \n\t"
        : [thresh_vec] "=&r" (thresh_vec), [flimit_vec] "=&r" (flimit_vec), [limit_vec] "=r" (limit_vec)
        : [thresh] "r" (thresh), [flimit] "r" (flimit), [limit] "r" (limit)
    );

    vp8_loop_filter_vertical_edge_mips(y_ptr + 4, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
    vp8_loop_filter_vertical_edge_mips(y_ptr + 8, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
    vp8_loop_filter_vertical_edge_mips(y_ptr + 12, y_stride, flimit_vec, limit_vec, thresh_vec, 16);

    if (u_ptr)
        vp8_loop_filter_uvvertical_edge_mips(u_ptr + 4, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);

    if (v_ptr)
        vp8_loop_filter_uvvertical_edge_mips(v_ptr + 4, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
}

#endif