Adding upstream version 124.0.1.upstream/124.0.1

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

1 files changed, 1113 insertions, 0 deletions

diff --git a/third_party/dav1d/src/arm/looprestoration.h b/third_party/dav1d/src/arm/looprestoration.h
new file mode 100644
index 0000000000..1ac6d5fb5e
--- /dev/null
+++ b/third_party/dav1d/src/arm/looprestoration.h
@@ -0,0 +1,1113 @@
+/*
+ * Copyright © 2018, VideoLAN and dav1d authors
+ * Copyright © 2018, Two Orioles, LLC
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice, this
+ *    list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright notice,
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+ * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "src/cpu.h"
+#include "src/looprestoration.h"
+
+#if ARCH_AARCH64
+void BF(dav1d_wiener_filter7, neon)(pixel *p, const ptrdiff_t stride,
+                                    const pixel (*left)[4], const pixel *lpf,
+                                    const int w, int h,
+                                    const LooprestorationParams *const params,
+                                    const enum LrEdgeFlags edges
+                                    HIGHBD_DECL_SUFFIX);
+void BF(dav1d_wiener_filter5, neon)(pixel *p, const ptrdiff_t stride,
+                                    const pixel (*left)[4], const pixel *lpf,
+                                    const int w, int h,
+                                    const LooprestorationParams *const params,
+                                    const enum LrEdgeFlags edges
+                                    HIGHBD_DECL_SUFFIX);
+#else
+
+// The 8bpc version calculates things slightly differently than the reference
+// C version. That version calculates roughly this:
+// int16_t sum = 0;
+// for (int i = 0; i < 7; i++)
+//     sum += src[idx] * fh[i];
+// int16_t sum2 = (src[x] << 7) - (1 << (bitdepth + 6)) + rounding_off_h;
+// sum = iclip(sum + sum2, INT16_MIN, INT16_MAX) >> round_bits_h;
+// sum += 1 << (bitdepth + 6 - round_bits_h);
+// Compared to the reference C version, this is the output of the first pass
+// _subtracted_ by 1 << (bitdepth + 6 - round_bits_h) = 2048, i.e.
+// with round_offset precompensated.
+// The 16bpc version calculates things pretty much the same way as the
+// reference C version, but with the end result subtracted by
+// 1 << (bitdepth + 6 - round_bits_h).
+void BF(dav1d_wiener_filter_h, neon)(int16_t *dst, const pixel (*left)[4],
+                                     const pixel *src, ptrdiff_t stride,
+                                     const int16_t fh[8], intptr_t w,
+                                     int h, enum LrEdgeFlags edges
+                                     HIGHBD_DECL_SUFFIX);
+// This calculates things slightly differently than the reference C version.
+// This version calculates roughly this:
+// int32_t sum = 0;
+// for (int i = 0; i < 7; i++)
+//     sum += mid[idx] * fv[i];
+// sum = (sum + rounding_off_v) >> round_bits_v;
+// This function assumes that the width is a multiple of 8.
+void BF(dav1d_wiener_filter_v, neon)(pixel *dst, ptrdiff_t stride,
+                                     const int16_t *mid, int w, int h,
+                                     const int16_t fv[8], enum LrEdgeFlags edges,
+                                     ptrdiff_t mid_stride HIGHBD_DECL_SUFFIX);
+
+static void wiener_filter_neon(pixel *const dst, const ptrdiff_t stride,
+                               const pixel (*const left)[4], const pixel *lpf,
+                               const int w, const int h,
+                               const LooprestorationParams *const params,
+                               const enum LrEdgeFlags edges HIGHBD_DECL_SUFFIX)
+{
+    const int16_t (*const filter)[8] = params->filter;
+    ALIGN_STK_16(int16_t, mid, 68 * 384,);
+    int mid_stride = (w + 7) & ~7;
+
+    // Horizontal filter
+    BF(dav1d_wiener_filter_h, neon)(&mid[2 * mid_stride], left, dst, stride,
+                                    filter[0], w, h, edges HIGHBD_TAIL_SUFFIX);
+    if (edges & LR_HAVE_TOP)
+        BF(dav1d_wiener_filter_h, neon)(mid, NULL, lpf, stride,
+                                        filter[0], w, 2, edges
+                                        HIGHBD_TAIL_SUFFIX);
+    if (edges & LR_HAVE_BOTTOM)
+        BF(dav1d_wiener_filter_h, neon)(&mid[(2 + h) * mid_stride], NULL,
+                                        lpf + 6 * PXSTRIDE(stride),
+                                        stride, filter[0], w, 2, edges
+                                        HIGHBD_TAIL_SUFFIX);
+
+    // Vertical filter
+    BF(dav1d_wiener_filter_v, neon)(dst, stride, &mid[2*mid_stride],
+                                    w, h, filter[1], edges,
+                                    mid_stride * sizeof(*mid)
+                                    HIGHBD_TAIL_SUFFIX);
+}
+#endif
+
+#if ARCH_ARM
+void BF(dav1d_sgr_box3_h, neon)(int32_t *sumsq, int16_t *sum,
+                                const pixel (*left)[4],
+                                const pixel *src, const ptrdiff_t stride,
+                                const int w, const int h,
+                                const enum LrEdgeFlags edges);
+void dav1d_sgr_box3_v_neon(int32_t *sumsq, int16_t *sum,
+                           const int w, const int h,
+                           const enum LrEdgeFlags edges);
+void dav1d_sgr_calc_ab1_neon(int32_t *a, int16_t *b,
+                             const int w, const int h, const int strength,
+                             const int bitdepth_max);
+void BF(dav1d_sgr_finish_filter1, neon)(int16_t *tmp,
+                                        const pixel *src, const ptrdiff_t stride,
+                                        const int32_t *a, const int16_t *b,
+                                        const int w, const int h);
+
+/* filter with a 3x3 box (radius=1) */
+static void dav1d_sgr_filter1_neon(int16_t *tmp,
+                                   const pixel *src, const ptrdiff_t stride,
+                                   const pixel (*left)[4], const pixel *lpf,
+                                   const int w, const int h, const int strength,
+                                   const enum LrEdgeFlags edges
+                                   HIGHBD_DECL_SUFFIX)
+{
+    ALIGN_STK_16(int32_t, sumsq_mem, (384 + 16) * 68 + 8,);
+    int32_t *const sumsq = &sumsq_mem[(384 + 16) * 2 + 8], *const a = sumsq;
+    ALIGN_STK_16(int16_t, sum_mem, (384 + 16) * 68 + 16,);
+    int16_t *const sum = &sum_mem[(384 + 16) * 2 + 16], *const b = sum;
+
+    BF(dav1d_sgr_box3_h, neon)(sumsq, sum, left, src, stride, w, h, edges);
+    if (edges & LR_HAVE_TOP)
+        BF(dav1d_sgr_box3_h, neon)(&sumsq[-2 * (384 + 16)], &sum[-2 * (384 + 16)],
+                                   NULL, lpf, stride, w, 2, edges);
+
+    if (edges & LR_HAVE_BOTTOM)
+        BF(dav1d_sgr_box3_h, neon)(&sumsq[h * (384 + 16)], &sum[h * (384 + 16)],
+                                   NULL, lpf + 6 * PXSTRIDE(stride),
+                                   stride, w, 2, edges);
+
+    dav1d_sgr_box3_v_neon(sumsq, sum, w, h, edges);
+    dav1d_sgr_calc_ab1_neon(a, b, w, h, strength, BITDEPTH_MAX);
+    BF(dav1d_sgr_finish_filter1, neon)(tmp, src, stride, a, b, w, h);
+}
+
+void BF(dav1d_sgr_box5_h, neon)(int32_t *sumsq, int16_t *sum,
+                                const pixel (*left)[4],
+                                const pixel *src, const ptrdiff_t stride,
+                                const int w, const int h,
+                                const enum LrEdgeFlags edges);
+void dav1d_sgr_box5_v_neon(int32_t *sumsq, int16_t *sum,
+                           const int w, const int h,
+                           const enum LrEdgeFlags edges);
+void dav1d_sgr_calc_ab2_neon(int32_t *a, int16_t *b,
+                             const int w, const int h, const int strength,
+                             const int bitdepth_max);
+void BF(dav1d_sgr_finish_filter2, neon)(int16_t *tmp,
+                                        const pixel *src, const ptrdiff_t stride,
+                                        const int32_t *a, const int16_t *b,
+                                        const int w, const int h);
+
+/* filter with a 5x5 box (radius=2) */
+static void dav1d_sgr_filter2_neon(int16_t *tmp,
+                                   const pixel *src, const ptrdiff_t stride,
+                                   const pixel (*left)[4], const pixel *lpf,
+                                   const int w, const int h, const int strength,
+                                   const enum LrEdgeFlags edges
+                                   HIGHBD_DECL_SUFFIX)
+{
+    ALIGN_STK_16(int32_t, sumsq_mem, (384 + 16) * 68 + 8,);
+    int32_t *const sumsq = &sumsq_mem[(384 + 16) * 2 + 8], *const a = sumsq;
+    ALIGN_STK_16(int16_t, sum_mem, (384 + 16) * 68 + 16,);
+    int16_t *const sum = &sum_mem[(384 + 16) * 2 + 16], *const b = sum;
+
+    BF(dav1d_sgr_box5_h, neon)(sumsq, sum, left, src, stride, w, h, edges);
+    if (edges & LR_HAVE_TOP)
+        BF(dav1d_sgr_box5_h, neon)(&sumsq[-2 * (384 + 16)], &sum[-2 * (384 + 16)],
+                                   NULL, lpf, stride, w, 2, edges);
+
+    if (edges & LR_HAVE_BOTTOM)
+        BF(dav1d_sgr_box5_h, neon)(&sumsq[h * (384 + 16)], &sum[h * (384 + 16)],
+                                   NULL, lpf + 6 * PXSTRIDE(stride),
+                                   stride, w, 2, edges);
+
+    dav1d_sgr_box5_v_neon(sumsq, sum, w, h, edges);
+    dav1d_sgr_calc_ab2_neon(a, b, w, h, strength, BITDEPTH_MAX);
+    BF(dav1d_sgr_finish_filter2, neon)(tmp, src, stride, a, b, w, h);
+}
+
+void BF(dav1d_sgr_weighted1, neon)(pixel *dst, const ptrdiff_t dst_stride,
+                                   const pixel *src, const ptrdiff_t src_stride,
+                                   const int16_t *t1, const int w, const int h,
+                                   const int wt HIGHBD_DECL_SUFFIX);
+void BF(dav1d_sgr_weighted2, neon)(pixel *dst, const ptrdiff_t dst_stride,
+                                   const pixel *src, const ptrdiff_t src_stride,
+                                   const int16_t *t1, const int16_t *t2,
+                                   const int w, const int h,
+                                   const int16_t wt[2] HIGHBD_DECL_SUFFIX);
+
+static void sgr_filter_5x5_neon(pixel *const dst, const ptrdiff_t stride,
+                                const pixel (*const left)[4], const pixel *lpf,
+                                const int w, const int h,
+                                const LooprestorationParams *const params,
+                                const enum LrEdgeFlags edges HIGHBD_DECL_SUFFIX)
+{
+    ALIGN_STK_16(int16_t, tmp, 64 * 384,);
+    dav1d_sgr_filter2_neon(tmp, dst, stride, left, lpf,
+                           w, h, params->sgr.s0, edges HIGHBD_TAIL_SUFFIX);
+    BF(dav1d_sgr_weighted1, neon)(dst, stride, dst, stride,
+                                  tmp, w, h, params->sgr.w0 HIGHBD_TAIL_SUFFIX);
+}
+
+static void sgr_filter_3x3_neon(pixel *const dst, const ptrdiff_t stride,
+                                const pixel (*const left)[4], const pixel *lpf,
+                                const int w, const int h,
+                                const LooprestorationParams *const params,
+                                const enum LrEdgeFlags edges HIGHBD_DECL_SUFFIX)
+{
+    ALIGN_STK_16(int16_t, tmp, 64 * 384,);
+    dav1d_sgr_filter1_neon(tmp, dst, stride, left, lpf,
+                           w, h, params->sgr.s1, edges HIGHBD_TAIL_SUFFIX);
+    BF(dav1d_sgr_weighted1, neon)(dst, stride, dst, stride,
+                                  tmp, w, h, params->sgr.w1 HIGHBD_TAIL_SUFFIX);
+}
+
+static void sgr_filter_mix_neon(pixel *const dst, const ptrdiff_t stride,
+                                const pixel (*const left)[4], const pixel *lpf,
+                                const int w, const int h,
+                                const LooprestorationParams *const params,
+                                const enum LrEdgeFlags edges HIGHBD_DECL_SUFFIX)
+{
+    ALIGN_STK_16(int16_t, tmp1, 64 * 384,);
+    ALIGN_STK_16(int16_t, tmp2, 64 * 384,);
+    dav1d_sgr_filter2_neon(tmp1, dst, stride, left, lpf,
+                           w, h, params->sgr.s0, edges HIGHBD_TAIL_SUFFIX);
+    dav1d_sgr_filter1_neon(tmp2, dst, stride, left, lpf,
+                           w, h, params->sgr.s1, edges HIGHBD_TAIL_SUFFIX);
+    const int16_t wt[2] = { params->sgr.w0, params->sgr.w1 };
+    BF(dav1d_sgr_weighted2, neon)(dst, stride, dst, stride,
+                                  tmp1, tmp2, w, h, wt HIGHBD_TAIL_SUFFIX);
+}
+
+#else
+static void rotate(int32_t **sumsq_ptrs, int16_t **sum_ptrs, int n) {
+    int32_t *tmp32 = sumsq_ptrs[0];
+    int16_t *tmp16 = sum_ptrs[0];
+    for (int i = 0; i < n - 1; i++) {
+        sumsq_ptrs[i] = sumsq_ptrs[i+1];
+        sum_ptrs[i] = sum_ptrs[i+1];
+    }
+    sumsq_ptrs[n - 1] = tmp32;
+    sum_ptrs[n - 1] = tmp16;
+}
+static void rotate5_x2(int32_t **sumsq_ptrs, int16_t **sum_ptrs) {
+    int32_t *tmp32[2];
+    int16_t *tmp16[2];
+    for (int i = 0; i < 2; i++) {
+        tmp32[i] = sumsq_ptrs[i];
+        tmp16[i] = sum_ptrs[i];
+    }
+    for (int i = 0; i < 3; i++) {
+        sumsq_ptrs[i] = sumsq_ptrs[i+2];
+        sum_ptrs[i] = sum_ptrs[i+2];
+    }
+    for (int i = 0; i < 2; i++) {
+        sumsq_ptrs[3 + i] = tmp32[i];
+        sum_ptrs[3 + i] = tmp16[i];
+    }
+}
+
+static void rotate_ab_3(int32_t **A_ptrs, int16_t **B_ptrs) {
+    rotate(A_ptrs, B_ptrs, 3);
+}
+
+static void rotate_ab_2(int32_t **A_ptrs, int16_t **B_ptrs) {
+    rotate(A_ptrs, B_ptrs, 2);
+}
+
+static void rotate_ab_4(int32_t **A_ptrs, int16_t **B_ptrs) {
+    rotate(A_ptrs, B_ptrs, 4);
+}
+
+void BF(dav1d_sgr_box3_row_h, neon)(int32_t *sumsq, int16_t *sum,
+                                    const pixel (*left)[4],
+                                    const pixel *src, const int w,
+                                    const enum LrEdgeFlags edges);
+void BF(dav1d_sgr_box5_row_h, neon)(int32_t *sumsq, int16_t *sum,
+                                    const pixel (*left)[4],
+                                    const pixel *src, const int w,
+                                    const enum LrEdgeFlags edges);
+void BF(dav1d_sgr_box35_row_h, neon)(int32_t *sumsq3, int16_t *sum3,
+                                     int32_t *sumsq5, int16_t *sum5,
+                                     const pixel (*left)[4],
+                                     const pixel *src, const int w,
+                                     const enum LrEdgeFlags edges);
+
+void dav1d_sgr_box3_vert_neon(int32_t **sumsq, int16_t **sum,
+                              int32_t *AA, int16_t *BB,
+                              const int w, const int s,
+                              const int bitdepth_max);
+void dav1d_sgr_box5_vert_neon(int32_t **sumsq, int16_t **sum,
+                              int32_t *AA, int16_t *BB,
+                              const int w, const int s,
+                              const int bitdepth_max);
+
+void BF(dav1d_sgr_finish_weighted1, neon)(pixel *dst,
+                                          int32_t **A_ptrs, int16_t **B_ptrs,
+                                          const int w, const int w1
+                                          HIGHBD_DECL_SUFFIX);
+void BF(dav1d_sgr_finish_weighted2, neon)(pixel *dst, const ptrdiff_t stride,
+                                          int32_t **A_ptrs, int16_t **B_ptrs,
+                                          const int w, const int h,
+                                          const int w1 HIGHBD_DECL_SUFFIX);
+
+void BF(dav1d_sgr_finish_filter1_2rows, neon)(int16_t *tmp, const pixel *src,
+                                              const ptrdiff_t src_stride,
+                                              int32_t **A_ptrs,
+                                              int16_t **B_ptrs,
+                                              const int w, const int h);
+void BF(dav1d_sgr_finish_filter2_2rows, neon)(int16_t *tmp, const pixel *src,
+                                              const ptrdiff_t src_stride,
+                                              int32_t **A_ptrs, int16_t **B_ptrs,
+                                              const int w, const int h);
+void BF(dav1d_sgr_weighted2, neon)(pixel *dst, const ptrdiff_t dst_stride,
+                                   const pixel *src, const ptrdiff_t src_stride,
+                                   const int16_t *t1, const int16_t *t2,
+                                   const int w, const int h,
+                                   const int16_t wt[2] HIGHBD_DECL_SUFFIX);
+
+static void sgr_box3_vert_neon(int32_t **sumsq, int16_t **sum,
+                               int32_t *sumsq_out, int16_t *sum_out,
+                               const int w, int s, int bitdepth_max) {
+    // box3_v + calc_ab1
+    dav1d_sgr_box3_vert_neon(sumsq, sum, sumsq_out, sum_out, w, s, bitdepth_max);
+    rotate(sumsq, sum, 3);
+}
+
+static void sgr_box5_vert_neon(int32_t **sumsq, int16_t **sum,
+                               int32_t *sumsq_out, int16_t *sum_out,
+                               const int w, int s, int bitdepth_max) {
+    // box5_v + calc_ab2
+    dav1d_sgr_box5_vert_neon(sumsq, sum, sumsq_out, sum_out, w, s, bitdepth_max);
+    rotate5_x2(sumsq, sum);
+}
+
+static void sgr_box3_hv_neon(int32_t **sumsq, int16_t **sum,
+                             int32_t *AA, int16_t *BB,
+                             const pixel (*left)[4],
+                             const pixel *src, const int w,
+                             const int s,
+                             const enum LrEdgeFlags edges,
+                             const int bitdepth_max) {
+    BF(dav1d_sgr_box3_row_h, neon)(sumsq[2], sum[2], left, src, w, edges);
+    sgr_box3_vert_neon(sumsq, sum, AA, BB, w, s, bitdepth_max);
+}
+
+
+static void sgr_finish1_neon(pixel **dst, const ptrdiff_t stride,
+                             int32_t **A_ptrs, int16_t **B_ptrs, const int w,
+                             const int w1 HIGHBD_DECL_SUFFIX) {
+    BF(dav1d_sgr_finish_weighted1, neon)(*dst, A_ptrs, B_ptrs,
+                                         w, w1 HIGHBD_TAIL_SUFFIX);
+    *dst += PXSTRIDE(stride);
+    rotate_ab_3(A_ptrs, B_ptrs);
+}
+
+static void sgr_finish2_neon(pixel **dst, const ptrdiff_t stride,
+                             int32_t **A_ptrs, int16_t **B_ptrs,
+                             const int w, const int h, const int w1
+                             HIGHBD_DECL_SUFFIX) {
+    BF(dav1d_sgr_finish_weighted2, neon)(*dst, stride, A_ptrs, B_ptrs,
+                                         w, h, w1 HIGHBD_TAIL_SUFFIX);
+    *dst += 2*PXSTRIDE(stride);
+    rotate_ab_2(A_ptrs, B_ptrs);
+}
+
+static void sgr_finish_mix_neon(pixel **dst, const ptrdiff_t stride,
+                                int32_t **A5_ptrs, int16_t **B5_ptrs,
+                                int32_t **A3_ptrs, int16_t **B3_ptrs,
+                                const int w, const int h,
+                                const int w0, const int w1 HIGHBD_DECL_SUFFIX) {
+#define FILTER_OUT_STRIDE 384
+    ALIGN_STK_16(int16_t, tmp5, 2*FILTER_OUT_STRIDE,);
+    ALIGN_STK_16(int16_t, tmp3, 2*FILTER_OUT_STRIDE,);
+
+    BF(dav1d_sgr_finish_filter2_2rows, neon)(tmp5, *dst, stride,
+                                             A5_ptrs, B5_ptrs, w, h);
+    BF(dav1d_sgr_finish_filter1_2rows, neon)(tmp3, *dst, stride,
+                                             A3_ptrs, B3_ptrs, w, h);
+    const int16_t wt[2] = { w0, w1 };
+    BF(dav1d_sgr_weighted2, neon)(*dst, stride, *dst, stride,
+                                  tmp5, tmp3, w, h, wt HIGHBD_TAIL_SUFFIX);
+    *dst += h*PXSTRIDE(stride);
+    rotate_ab_2(A5_ptrs, B5_ptrs);
+    rotate_ab_4(A3_ptrs, B3_ptrs);
+}
+
+
+static void sgr_filter_3x3_neon(pixel *dst, const ptrdiff_t stride,
+                                const pixel (*left)[4], const pixel *lpf,
+                                const int w, int h,
+                                const LooprestorationParams *const params,
+                                const enum LrEdgeFlags edges HIGHBD_DECL_SUFFIX)
+{
+#define BUF_STRIDE (384 + 16)
+    ALIGN_STK_16(int32_t, sumsq_buf, BUF_STRIDE * 3 + 16,);
+    ALIGN_STK_16(int16_t, sum_buf, BUF_STRIDE * 3 + 16,);
+    int32_t *sumsq_ptrs[3], *sumsq_rows[3];
+    int16_t *sum_ptrs[3], *sum_rows[3];
+    for (int i = 0; i < 3; i++) {
+        sumsq_rows[i] = &sumsq_buf[i * BUF_STRIDE];
+        sum_rows[i] = &sum_buf[i * BUF_STRIDE];
+    }
+
+    ALIGN_STK_16(int32_t, A_buf, BUF_STRIDE * 3 + 16,);
+    ALIGN_STK_16(int16_t, B_buf, BUF_STRIDE * 3 + 16,);
+    int32_t *A_ptrs[3];
+    int16_t *B_ptrs[3];
+    for (int i = 0; i < 3; i++) {
+        A_ptrs[i] = &A_buf[i * BUF_STRIDE];
+        B_ptrs[i] = &B_buf[i * BUF_STRIDE];
+    }
+    const pixel *src = dst;
+    const pixel *lpf_bottom = lpf + 6*PXSTRIDE(stride);
+
+    if (edges & LR_HAVE_TOP) {
+        sumsq_ptrs[0] = sumsq_rows[0];
+        sumsq_ptrs[1] = sumsq_rows[1];
+        sumsq_ptrs[2] = sumsq_rows[2];
+        sum_ptrs[0] = sum_rows[0];
+        sum_ptrs[1] = sum_rows[1];
+        sum_ptrs[2] = sum_rows[2];
+
+        BF(dav1d_sgr_box3_row_h, neon)(sumsq_rows[0], sum_rows[0],
+                                       NULL, lpf, w, edges);
+        lpf += PXSTRIDE(stride);
+        BF(dav1d_sgr_box3_row_h, neon)(sumsq_rows[1], sum_rows[1],
+                                       NULL, lpf, w, edges);
+
+        sgr_box3_hv_neon(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2],
+                         left, src, w, params->sgr.s1, edges, BITDEPTH_MAX);
+        left++;
+        src += PXSTRIDE(stride);
+        rotate_ab_3(A_ptrs, B_ptrs);
+
+        if (--h <= 0)
+            goto vert_1;
+
+        sgr_box3_hv_neon(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2], left, src, w, params->sgr.s1, edges, BITDEPTH_MAX);
+        left++;
+        src += PXSTRIDE(stride);
+        rotate_ab_3(A_ptrs, B_ptrs);
+
+        if (--h <= 0)
+            goto vert_2;
+    } else {
+        sumsq_ptrs[0] = sumsq_rows[0];
+        sumsq_ptrs[1] = sumsq_rows[0];
+        sumsq_ptrs[2] = sumsq_rows[0];
+        sum_ptrs[0] = sum_rows[0];
+        sum_ptrs[1] = sum_rows[0];
+        sum_ptrs[2] = sum_rows[0];
+
+        BF(dav1d_sgr_box3_row_h, neon)(sumsq_rows[0], sum_rows[0],
+                                       left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+
+        sgr_box3_vert_neon(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2],
+                           w, params->sgr.s1, BITDEPTH_MAX);
+        rotate_ab_3(A_ptrs, B_ptrs);
+
+        if (--h <= 0)
+            goto vert_1;
+
+        sumsq_ptrs[2] = sumsq_rows[1];
+        sum_ptrs[2] = sum_rows[1];
+
+        sgr_box3_hv_neon(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2],
+                         left, src, w, params->sgr.s1, edges, BITDEPTH_MAX);
+        left++;
+        src += PXSTRIDE(stride);
+        rotate_ab_3(A_ptrs, B_ptrs);
+
+        if (--h <= 0)
+            goto vert_2;
+
+        sumsq_ptrs[2] = sumsq_rows[2];
+        sum_ptrs[2] = sum_rows[2];
+    }
+
+    do {
+        sgr_box3_hv_neon(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2],
+                         left, src, w, params->sgr.s1, edges, BITDEPTH_MAX);
+        left++;
+        src += PXSTRIDE(stride);
+
+        sgr_finish1_neon(&dst, stride, A_ptrs, B_ptrs,
+                         w, params->sgr.w1 HIGHBD_TAIL_SUFFIX);
+    } while (--h > 0);
+
+    if (!(edges & LR_HAVE_BOTTOM))
+        goto vert_2;
+
+    sgr_box3_hv_neon(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2],
+                     NULL, lpf_bottom, w, params->sgr.s1, edges, BITDEPTH_MAX);
+    lpf_bottom += PXSTRIDE(stride);
+
+    sgr_finish1_neon(&dst, stride, A_ptrs, B_ptrs,
+                     w, params->sgr.w1 HIGHBD_TAIL_SUFFIX);
+
+    sgr_box3_hv_neon(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2],
+                     NULL, lpf_bottom, w, params->sgr.s1, edges, BITDEPTH_MAX);
+
+    sgr_finish1_neon(&dst, stride, A_ptrs, B_ptrs,
+                     w, params->sgr.w1 HIGHBD_TAIL_SUFFIX);
+    return;
+
+vert_2:
+    sumsq_ptrs[2] = sumsq_ptrs[1];
+    sum_ptrs[2] = sum_ptrs[1];
+    sgr_box3_vert_neon(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2],
+                       w, params->sgr.s1, BITDEPTH_MAX);
+
+    sgr_finish1_neon(&dst, stride, A_ptrs, B_ptrs,
+                     w, params->sgr.w1 HIGHBD_TAIL_SUFFIX);
+
+output_1:
+    sumsq_ptrs[2] = sumsq_ptrs[1];
+    sum_ptrs[2] = sum_ptrs[1];
+    sgr_box3_vert_neon(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2],
+                       w, params->sgr.s1, BITDEPTH_MAX);
+
+    sgr_finish1_neon(&dst, stride, A_ptrs, B_ptrs,
+                     w, params->sgr.w1 HIGHBD_TAIL_SUFFIX);
+    return;
+
+vert_1:
+    sumsq_ptrs[2] = sumsq_ptrs[1];
+    sum_ptrs[2] = sum_ptrs[1];
+    sgr_box3_vert_neon(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2],
+                       w, params->sgr.s1, BITDEPTH_MAX);
+    rotate_ab_3(A_ptrs, B_ptrs);
+    goto output_1;
+}
+
+static void sgr_filter_5x5_neon(pixel *dst, const ptrdiff_t stride,
+                                const pixel (*left)[4], const pixel *lpf,
+                                const int w, int h,
+                                const LooprestorationParams *const params,
+                                const enum LrEdgeFlags edges HIGHBD_DECL_SUFFIX)
+{
+    ALIGN_STK_16(int32_t, sumsq_buf, BUF_STRIDE * 5 + 16,);
+    ALIGN_STK_16(int16_t, sum_buf, BUF_STRIDE * 5 + 16,);
+    int32_t *sumsq_ptrs[5], *sumsq_rows[5];
+    int16_t *sum_ptrs[5], *sum_rows[5];
+    for (int i = 0; i < 5; i++) {
+        sumsq_rows[i] = &sumsq_buf[i * BUF_STRIDE];
+        sum_rows[i] = &sum_buf[i * BUF_STRIDE];
+    }
+
+    ALIGN_STK_16(int32_t, A_buf, BUF_STRIDE * 2 + 16,);
+    ALIGN_STK_16(int16_t, B_buf, BUF_STRIDE * 2 + 16,);
+    int32_t *A_ptrs[2];
+    int16_t *B_ptrs[2];
+    for (int i = 0; i < 2; i++) {
+        A_ptrs[i] = &A_buf[i * BUF_STRIDE];
+        B_ptrs[i] = &B_buf[i * BUF_STRIDE];
+    }
+    const pixel *src = dst;
+    const pixel *lpf_bottom = lpf + 6*PXSTRIDE(stride);
+
+    if (edges & LR_HAVE_TOP) {
+        sumsq_ptrs[0] = sumsq_rows[0];
+        sumsq_ptrs[1] = sumsq_rows[0];
+        sumsq_ptrs[2] = sumsq_rows[1];
+        sumsq_ptrs[3] = sumsq_rows[2];
+        sumsq_ptrs[4] = sumsq_rows[3];
+        sum_ptrs[0] = sum_rows[0];
+        sum_ptrs[1] = sum_rows[0];
+        sum_ptrs[2] = sum_rows[1];
+        sum_ptrs[3] = sum_rows[2];
+        sum_ptrs[4] = sum_rows[3];
+
+        BF(dav1d_sgr_box5_row_h, neon)(sumsq_rows[0], sum_rows[0],
+                                       NULL, lpf, w, edges);
+        lpf += PXSTRIDE(stride);
+        BF(dav1d_sgr_box5_row_h, neon)(sumsq_rows[1], sum_rows[1],
+                                       NULL, lpf, w, edges);
+
+        BF(dav1d_sgr_box5_row_h, neon)(sumsq_rows[2], sum_rows[2],
+                                       left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+
+        if (--h <= 0)
+            goto vert_1;
+
+        BF(dav1d_sgr_box5_row_h, neon)(sumsq_rows[3], sum_rows[3],
+                                       left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+        sgr_box5_vert_neon(sumsq_ptrs, sum_ptrs, A_ptrs[1], B_ptrs[1],
+                           w, params->sgr.s0, BITDEPTH_MAX);
+        rotate_ab_2(A_ptrs, B_ptrs);
+
+        if (--h <= 0)
+            goto vert_2;
+
+        // ptrs are rotated by 2; both [3] and [4] now point at rows[0]; set
+        // one of them to point at the previously unused rows[4].
+        sumsq_ptrs[3] = sumsq_rows[4];
+        sum_ptrs[3] = sum_rows[4];
+    } else {
+        sumsq_ptrs[0] = sumsq_rows[0];
+        sumsq_ptrs[1] = sumsq_rows[0];
+        sumsq_ptrs[2] = sumsq_rows[0];
+        sumsq_ptrs[3] = sumsq_rows[0];
+        sumsq_ptrs[4] = sumsq_rows[0];
+        sum_ptrs[0] = sum_rows[0];
+        sum_ptrs[1] = sum_rows[0];
+        sum_ptrs[2] = sum_rows[0];
+        sum_ptrs[3] = sum_rows[0];
+        sum_ptrs[4] = sum_rows[0];
+
+        BF(dav1d_sgr_box5_row_h, neon)(sumsq_rows[0], sum_rows[0],
+                                       left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+
+        if (--h <= 0)
+            goto vert_1;
+
+        sumsq_ptrs[4] = sumsq_rows[1];
+        sum_ptrs[4] = sum_rows[1];
+
+        BF(dav1d_sgr_box5_row_h, neon)(sumsq_rows[1], sum_rows[1],
+                                       left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+
+        sgr_box5_vert_neon(sumsq_ptrs, sum_ptrs, A_ptrs[1], B_ptrs[1],
+                           w, params->sgr.s0, BITDEPTH_MAX);
+        rotate_ab_2(A_ptrs, B_ptrs);
+
+        if (--h <= 0)
+            goto vert_2;
+
+        sumsq_ptrs[3] = sumsq_rows[2];
+        sumsq_ptrs[4] = sumsq_rows[3];
+        sum_ptrs[3] = sum_rows[2];
+        sum_ptrs[4] = sum_rows[3];
+
+        BF(dav1d_sgr_box5_row_h, neon)(sumsq_rows[2], sum_rows[2],
+                                       left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+
+        if (--h <= 0)
+            goto odd;
+
+        BF(dav1d_sgr_box5_row_h, neon)(sumsq_rows[3], sum_rows[3],
+                                       left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+
+        sgr_box5_vert_neon(sumsq_ptrs, sum_ptrs, A_ptrs[1], B_ptrs[1],
+                           w, params->sgr.s0, BITDEPTH_MAX);
+        sgr_finish2_neon(&dst, stride, A_ptrs, B_ptrs,
+                         w, 2, params->sgr.w0 HIGHBD_TAIL_SUFFIX);
+
+        if (--h <= 0)
+            goto vert_2;
+
+        // ptrs are rotated by 2; both [3] and [4] now point at rows[0]; set
+        // one of them to point at the previously unused rows[4].
+        sumsq_ptrs[3] = sumsq_rows[4];
+        sum_ptrs[3] = sum_rows[4];
+    }
+
+    do {
+        BF(dav1d_sgr_box5_row_h, neon)(sumsq_ptrs[3], sum_ptrs[3],
+                                       left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+
+        if (--h <= 0)
+            goto odd;
+
+        BF(dav1d_sgr_box5_row_h, neon)(sumsq_ptrs[4], sum_ptrs[4],
+                                       left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+
+        sgr_box5_vert_neon(sumsq_ptrs, sum_ptrs, A_ptrs[1], B_ptrs[1],
+                           w, params->sgr.s0, BITDEPTH_MAX);
+        sgr_finish2_neon(&dst, stride, A_ptrs, B_ptrs,
+                         w, 2, params->sgr.w0 HIGHBD_TAIL_SUFFIX);
+    } while (--h > 0);
+
+    if (!(edges & LR_HAVE_BOTTOM))
+        goto vert_2;
+
+    BF(dav1d_sgr_box5_row_h, neon)(sumsq_ptrs[3], sum_ptrs[3],
+                                   NULL, lpf_bottom, w, edges);
+    lpf_bottom += PXSTRIDE(stride);
+    BF(dav1d_sgr_box5_row_h, neon)(sumsq_ptrs[4], sum_ptrs[4],
+                                   NULL, lpf_bottom, w, edges);
+
+output_2:
+    sgr_box5_vert_neon(sumsq_ptrs, sum_ptrs, A_ptrs[1], B_ptrs[1],
+                       w, params->sgr.s0, BITDEPTH_MAX);
+    sgr_finish2_neon(&dst, stride, A_ptrs, B_ptrs,
+                     w, 2, params->sgr.w0 HIGHBD_TAIL_SUFFIX);
+    return;
+
+vert_2:
+    // Duplicate the last row twice more
+    sumsq_ptrs[3] = sumsq_ptrs[2];
+    sumsq_ptrs[4] = sumsq_ptrs[2];
+    sum_ptrs[3] = sum_ptrs[2];
+    sum_ptrs[4] = sum_ptrs[2];
+    goto output_2;
+
+odd:
+    // Copy the last row as padding once
+    sumsq_ptrs[4] = sumsq_ptrs[3];
+    sum_ptrs[4] = sum_ptrs[3];
+
+    sgr_box5_vert_neon(sumsq_ptrs, sum_ptrs, A_ptrs[1], B_ptrs[1],
+                       w, params->sgr.s0, BITDEPTH_MAX);
+    sgr_finish2_neon(&dst, stride, A_ptrs, B_ptrs,
+                     w, 2, params->sgr.w0 HIGHBD_TAIL_SUFFIX);
+
+output_1:
+    // Duplicate the last row twice more
+    sumsq_ptrs[3] = sumsq_ptrs[2];
+    sumsq_ptrs[4] = sumsq_ptrs[2];
+    sum_ptrs[3] = sum_ptrs[2];
+    sum_ptrs[4] = sum_ptrs[2];
+
+    sgr_box5_vert_neon(sumsq_ptrs, sum_ptrs, A_ptrs[1], B_ptrs[1],
+                       w, params->sgr.s0, BITDEPTH_MAX);
+    // Output only one row
+    sgr_finish2_neon(&dst, stride, A_ptrs, B_ptrs,
+                     w, 1, params->sgr.w0 HIGHBD_TAIL_SUFFIX);
+    return;
+
+vert_1:
+    // Copy the last row as padding once
+    sumsq_ptrs[4] = sumsq_ptrs[3];
+    sum_ptrs[4] = sum_ptrs[3];
+
+    sgr_box5_vert_neon(sumsq_ptrs, sum_ptrs, A_ptrs[1], B_ptrs[1],
+                       w, params->sgr.s0, BITDEPTH_MAX);
+    rotate_ab_2(A_ptrs, B_ptrs);
+
+    goto output_1;
+}
+
+static void sgr_filter_mix_neon(pixel *dst, const ptrdiff_t stride,
+                                const pixel (*left)[4], const pixel *lpf,
+                                const int w, int h,
+                                const LooprestorationParams *const params,
+                                const enum LrEdgeFlags edges HIGHBD_DECL_SUFFIX)
+{
+    ALIGN_STK_16(int32_t, sumsq5_buf, BUF_STRIDE * 5 + 16,);
+    ALIGN_STK_16(int16_t, sum5_buf, BUF_STRIDE * 5 + 16,);
+    int32_t *sumsq5_ptrs[5], *sumsq5_rows[5];
+    int16_t *sum5_ptrs[5], *sum5_rows[5];
+    for (int i = 0; i < 5; i++) {
+        sumsq5_rows[i] = &sumsq5_buf[i * BUF_STRIDE];
+        sum5_rows[i] = &sum5_buf[i * BUF_STRIDE];
+    }
+    ALIGN_STK_16(int32_t, sumsq3_buf, BUF_STRIDE * 3 + 16,);
+    ALIGN_STK_16(int16_t, sum3_buf, BUF_STRIDE * 3 + 16,);
+    int32_t *sumsq3_ptrs[3], *sumsq3_rows[3];
+    int16_t *sum3_ptrs[3], *sum3_rows[3];
+    for (int i = 0; i < 3; i++) {
+        sumsq3_rows[i] = &sumsq3_buf[i * BUF_STRIDE];
+        sum3_rows[i] = &sum3_buf[i * BUF_STRIDE];
+    }
+
+    ALIGN_STK_16(int32_t, A5_buf, BUF_STRIDE * 2 + 16,);
+    ALIGN_STK_16(int16_t, B5_buf, BUF_STRIDE * 2 + 16,);
+    int32_t *A5_ptrs[2];
+    int16_t *B5_ptrs[2];
+    for (int i = 0; i < 2; i++) {
+        A5_ptrs[i] = &A5_buf[i * BUF_STRIDE];
+        B5_ptrs[i] = &B5_buf[i * BUF_STRIDE];
+    }
+    ALIGN_STK_16(int32_t, A3_buf, BUF_STRIDE * 4 + 16,);
+    ALIGN_STK_16(int16_t, B3_buf, BUF_STRIDE * 4 + 16,);
+    int32_t *A3_ptrs[4];
+    int16_t *B3_ptrs[4];
+    for (int i = 0; i < 4; i++) {
+        A3_ptrs[i] = &A3_buf[i * BUF_STRIDE];
+        B3_ptrs[i] = &B3_buf[i * BUF_STRIDE];
+    }
+    const pixel *src = dst;
+    const pixel *lpf_bottom = lpf + 6*PXSTRIDE(stride);
+
+    if (edges & LR_HAVE_TOP) {
+        sumsq5_ptrs[0] = sumsq5_rows[0];
+        sumsq5_ptrs[1] = sumsq5_rows[0];
+        sumsq5_ptrs[2] = sumsq5_rows[1];
+        sumsq5_ptrs[3] = sumsq5_rows[2];
+        sumsq5_ptrs[4] = sumsq5_rows[3];
+        sum5_ptrs[0] = sum5_rows[0];
+        sum5_ptrs[1] = sum5_rows[0];
+        sum5_ptrs[2] = sum5_rows[1];
+        sum5_ptrs[3] = sum5_rows[2];
+        sum5_ptrs[4] = sum5_rows[3];
+
+        sumsq3_ptrs[0] = sumsq3_rows[0];
+        sumsq3_ptrs[1] = sumsq3_rows[1];
+        sumsq3_ptrs[2] = sumsq3_rows[2];
+        sum3_ptrs[0] = sum3_rows[0];
+        sum3_ptrs[1] = sum3_rows[1];
+        sum3_ptrs[2] = sum3_rows[2];
+
+        BF(dav1d_sgr_box35_row_h, neon)(sumsq3_rows[0], sum3_rows[0],
+                                        sumsq5_rows[0], sum5_rows[0],
+                                        NULL, lpf, w, edges);
+        lpf += PXSTRIDE(stride);
+        BF(dav1d_sgr_box35_row_h, neon)(sumsq3_rows[1], sum3_rows[1],
+                                        sumsq5_rows[1], sum5_rows[1],
+                                        NULL, lpf, w, edges);
+
+        BF(dav1d_sgr_box35_row_h, neon)(sumsq3_rows[2], sum3_rows[2],
+                                        sumsq5_rows[2], sum5_rows[2],
+                                        left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+
+        sgr_box3_vert_neon(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3],
+                           w, params->sgr.s1, BITDEPTH_MAX);
+        rotate_ab_4(A3_ptrs, B3_ptrs);
+
+        if (--h <= 0)
+            goto vert_1;
+
+        BF(dav1d_sgr_box35_row_h, neon)(sumsq3_ptrs[2], sum3_ptrs[2],
+                                        sumsq5_rows[3], sum5_rows[3],
+                                        left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+        sgr_box5_vert_neon(sumsq5_ptrs, sum5_ptrs, A5_ptrs[1], B5_ptrs[1],
+                           w, params->sgr.s0, BITDEPTH_MAX);
+        rotate_ab_2(A5_ptrs, B5_ptrs);
+        sgr_box3_vert_neon(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3],
+                           w, params->sgr.s1, BITDEPTH_MAX);
+        rotate_ab_4(A3_ptrs, B3_ptrs);
+
+        if (--h <= 0)
+            goto vert_2;
+
+        // ptrs are rotated by 2; both [3] and [4] now point at rows[0]; set
+        // one of them to point at the previously unused rows[4].
+        sumsq5_ptrs[3] = sumsq5_rows[4];
+        sum5_ptrs[3] = sum5_rows[4];
+    } else {
+        sumsq5_ptrs[0] = sumsq5_rows[0];
+        sumsq5_ptrs[1] = sumsq5_rows[0];
+        sumsq5_ptrs[2] = sumsq5_rows[0];
+        sumsq5_ptrs[3] = sumsq5_rows[0];
+        sumsq5_ptrs[4] = sumsq5_rows[0];
+        sum5_ptrs[0] = sum5_rows[0];
+        sum5_ptrs[1] = sum5_rows[0];
+        sum5_ptrs[2] = sum5_rows[0];
+        sum5_ptrs[3] = sum5_rows[0];
+        sum5_ptrs[4] = sum5_rows[0];
+
+        sumsq3_ptrs[0] = sumsq3_rows[0];
+        sumsq3_ptrs[1] = sumsq3_rows[0];
+        sumsq3_ptrs[2] = sumsq3_rows[0];
+        sum3_ptrs[0] = sum3_rows[0];
+        sum3_ptrs[1] = sum3_rows[0];
+        sum3_ptrs[2] = sum3_rows[0];
+
+        BF(dav1d_sgr_box35_row_h, neon)(sumsq3_rows[0], sum3_rows[0],
+                                        sumsq5_rows[0], sum5_rows[0],
+                                        left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+
+        sgr_box3_vert_neon(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3],
+                           w, params->sgr.s1, BITDEPTH_MAX);
+        rotate_ab_4(A3_ptrs, B3_ptrs);
+
+        if (--h <= 0)
+            goto vert_1;
+
+        sumsq5_ptrs[4] = sumsq5_rows[1];
+        sum5_ptrs[4] = sum5_rows[1];
+
+        sumsq3_ptrs[2] = sumsq3_rows[1];
+        sum3_ptrs[2] = sum3_rows[1];
+
+        BF(dav1d_sgr_box35_row_h, neon)(sumsq3_rows[1], sum3_rows[1],
+                                        sumsq5_rows[1], sum5_rows[1],
+                                        left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+
+        sgr_box5_vert_neon(sumsq5_ptrs, sum5_ptrs, A5_ptrs[1], B5_ptrs[1],
+                           w, params->sgr.s0, BITDEPTH_MAX);
+        rotate_ab_2(A5_ptrs, B5_ptrs);
+        sgr_box3_vert_neon(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3],
+                           w, params->sgr.s1, BITDEPTH_MAX);
+        rotate_ab_4(A3_ptrs, B3_ptrs);
+
+        if (--h <= 0)
+            goto vert_2;
+
+        sumsq5_ptrs[3] = sumsq5_rows[2];
+        sumsq5_ptrs[4] = sumsq5_rows[3];
+        sum5_ptrs[3] = sum5_rows[2];
+        sum5_ptrs[4] = sum5_rows[3];
+
+        sumsq3_ptrs[2] = sumsq3_rows[2];
+        sum3_ptrs[2] = sum3_rows[2];
+
+        BF(dav1d_sgr_box35_row_h, neon)(sumsq3_rows[2], sum3_rows[2],
+                                        sumsq5_rows[2], sum5_rows[2],
+                                        left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+
+        sgr_box3_vert_neon(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3],
+                           w, params->sgr.s1, BITDEPTH_MAX);
+        rotate_ab_4(A3_ptrs, B3_ptrs);
+
+        if (--h <= 0)
+            goto odd;
+
+        BF(dav1d_sgr_box35_row_h, neon)(sumsq3_ptrs[2], sum3_ptrs[2],
+                                        sumsq5_rows[3], sum5_rows[3],
+                                        left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+
+        sgr_box5_vert_neon(sumsq5_ptrs, sum5_ptrs, A5_ptrs[1], B5_ptrs[1],
+                           w, params->sgr.s0, BITDEPTH_MAX);
+        sgr_box3_vert_neon(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3],
+                           w, params->sgr.s1, BITDEPTH_MAX);
+        sgr_finish_mix_neon(&dst, stride, A5_ptrs, B5_ptrs, A3_ptrs, B3_ptrs,
+                            w, 2, params->sgr.w0, params->sgr.w1
+                            HIGHBD_TAIL_SUFFIX);
+
+        if (--h <= 0)
+            goto vert_2;
+
+        // ptrs are rotated by 2; both [3] and [4] now point at rows[0]; set
+        // one of them to point at the previously unused rows[4].
+        sumsq5_ptrs[3] = sumsq5_rows[4];
+        sum5_ptrs[3] = sum5_rows[4];
+    }
+
+    do {
+        BF(dav1d_sgr_box35_row_h, neon)(sumsq3_ptrs[2], sum3_ptrs[2],
+                                        sumsq5_ptrs[3], sum5_ptrs[3],
+                                        left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+
+        sgr_box3_vert_neon(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3],
+                           w, params->sgr.s1, BITDEPTH_MAX);
+        rotate_ab_4(A3_ptrs, B3_ptrs);
+
+        if (--h <= 0)
+            goto odd;
+
+        BF(dav1d_sgr_box35_row_h, neon)(sumsq3_ptrs[2], sum3_ptrs[2],
+                                        sumsq5_ptrs[4], sum5_ptrs[4],
+                                        left, src, w, edges);
+        left++;
+        src += PXSTRIDE(stride);
+
+        sgr_box5_vert_neon(sumsq5_ptrs, sum5_ptrs, A5_ptrs[1], B5_ptrs[1],
+                           w, params->sgr.s0, BITDEPTH_MAX);
+        sgr_box3_vert_neon(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3],
+                           w, params->sgr.s1, BITDEPTH_MAX);
+        sgr_finish_mix_neon(&dst, stride, A5_ptrs, B5_ptrs, A3_ptrs, B3_ptrs,
+                            w, 2, params->sgr.w0, params->sgr.w1
+                            HIGHBD_TAIL_SUFFIX);
+    } while (--h > 0);
+
+    if (!(edges & LR_HAVE_BOTTOM))
+        goto vert_2;
+
+    BF(dav1d_sgr_box35_row_h, neon)(sumsq3_ptrs[2], sum3_ptrs[2],
+                                    sumsq5_ptrs[3], sum5_ptrs[3],
+                                    NULL, lpf_bottom, w, edges);
+    lpf_bottom += PXSTRIDE(stride);
+    sgr_box3_vert_neon(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3],
+                       w, params->sgr.s1, BITDEPTH_MAX);
+    rotate_ab_4(A3_ptrs, B3_ptrs);
+
+    BF(dav1d_sgr_box35_row_h, neon)(sumsq3_ptrs[2], sum3_ptrs[2],
+                                    sumsq5_ptrs[4], sum5_ptrs[4],
+                                    NULL, lpf_bottom, w, edges);
+
+output_2:
+    sgr_box5_vert_neon(sumsq5_ptrs, sum5_ptrs, A5_ptrs[1], B5_ptrs[1],
+                       w, params->sgr.s0, BITDEPTH_MAX);
+    sgr_box3_vert_neon(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3],
+                       w, params->sgr.s1, BITDEPTH_MAX);
+    sgr_finish_mix_neon(&dst, stride, A5_ptrs, B5_ptrs, A3_ptrs, B3_ptrs,
+                        w, 2, params->sgr.w0, params->sgr.w1
+                        HIGHBD_TAIL_SUFFIX);
+    return;
+
+vert_2:
+    // Duplicate the last row twice more
+    sumsq5_ptrs[3] = sumsq5_ptrs[2];
+    sumsq5_ptrs[4] = sumsq5_ptrs[2];
+    sum5_ptrs[3] = sum5_ptrs[2];
+    sum5_ptrs[4] = sum5_ptrs[2];
+
+    sumsq3_ptrs[2] = sumsq3_ptrs[1];
+    sum3_ptrs[2] = sum3_ptrs[1];
+    sgr_box3_vert_neon(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3],
+                       w, params->sgr.s1, BITDEPTH_MAX);
+    rotate_ab_4(A3_ptrs, B3_ptrs);
+
+    sumsq3_ptrs[2] = sumsq3_ptrs[1];
+    sum3_ptrs[2] = sum3_ptrs[1];
+
+    goto output_2;
+
+odd:
+    // Copy the last row as padding once
+    sumsq5_ptrs[4] = sumsq5_ptrs[3];
+    sum5_ptrs[4] = sum5_ptrs[3];
+
+    sumsq3_ptrs[2] = sumsq3_ptrs[1];
+    sum3_ptrs[2] = sum3_ptrs[1];
+
+    sgr_box5_vert_neon(sumsq5_ptrs, sum5_ptrs, A5_ptrs[1], B5_ptrs[1],
+                       w, params->sgr.s0, BITDEPTH_MAX);
+    sgr_box3_vert_neon(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3],
+                       w, params->sgr.s1, BITDEPTH_MAX);
+    sgr_finish_mix_neon(&dst, stride, A5_ptrs, B5_ptrs, A3_ptrs, B3_ptrs,
+                        w, 2, params->sgr.w0, params->sgr.w1
+                        HIGHBD_TAIL_SUFFIX);
+
+output_1:
+    // Duplicate the last row twice more
+    sumsq5_ptrs[3] = sumsq5_ptrs[2];
+    sumsq5_ptrs[4] = sumsq5_ptrs[2];
+    sum5_ptrs[3] = sum5_ptrs[2];
+    sum5_ptrs[4] = sum5_ptrs[2];
+
+    sumsq3_ptrs[2] = sumsq3_ptrs[1];
+    sum3_ptrs[2] = sum3_ptrs[1];
+
+    sgr_box5_vert_neon(sumsq5_ptrs, sum5_ptrs, A5_ptrs[1], B5_ptrs[1],
+                       w, params->sgr.s0, BITDEPTH_MAX);
+    sgr_box3_vert_neon(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3],
+                       w, params->sgr.s1, BITDEPTH_MAX);
+    rotate_ab_4(A3_ptrs, B3_ptrs);
+    // Output only one row
+    sgr_finish_mix_neon(&dst, stride, A5_ptrs, B5_ptrs, A3_ptrs, B3_ptrs,
+                        w, 1, params->sgr.w0, params->sgr.w1
+                        HIGHBD_TAIL_SUFFIX);
+    return;
+
+vert_1:
+    // Copy the last row as padding once
+    sumsq5_ptrs[4] = sumsq5_ptrs[3];
+    sum5_ptrs[4] = sum5_ptrs[3];
+
+    sumsq3_ptrs[2] = sumsq3_ptrs[1];
+    sum3_ptrs[2] = sum3_ptrs[1];
+
+    sgr_box5_vert_neon(sumsq5_ptrs, sum5_ptrs, A5_ptrs[1], B5_ptrs[1],
+                       w, params->sgr.s0, BITDEPTH_MAX);
+    rotate_ab_2(A5_ptrs, B5_ptrs);
+    sgr_box3_vert_neon(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3],
+                       w, params->sgr.s1, BITDEPTH_MAX);
+    rotate_ab_4(A3_ptrs, B3_ptrs);
+
+    goto output_1;
+}
+
+#endif
+
+
+static ALWAYS_INLINE void loop_restoration_dsp_init_arm(Dav1dLoopRestorationDSPContext *const c, int bpc) {
+    const unsigned flags = dav1d_get_cpu_flags();
+
+    if (!(flags & DAV1D_ARM_CPU_FLAG_NEON)) return;
+
+#if ARCH_AARCH64
+    c->wiener[0] = BF(dav1d_wiener_filter7, neon);
+    c->wiener[1] = BF(dav1d_wiener_filter5, neon);
+#else
+    c->wiener[0] = c->wiener[1] = wiener_filter_neon;
+#endif
+    if (BITDEPTH == 8 || bpc == 10) {
+        c->sgr[0] = sgr_filter_5x5_neon;
+        c->sgr[1] = sgr_filter_3x3_neon;
+        c->sgr[2] = sgr_filter_mix_neon;
+    }
+}