src/f16-dwconv/gen/f16-dwconv-5f5m5l16c8s4r-minmax-fma3-acc2.c

// Auto-generated file. Do not edit!
//   Template: src/f16-dwconv/multipass-fma3.c.in
//   Generator: tools/xngen
//
// Copyright 2022 Google LLC
//
// This source code is licensed under the BSD-style license found in the
// LICENSE file in the root directory of this source tree.

#include <assert.h>
#include <stddef.h>
#include <stdint.h>

#include <immintrin.h>

#include <xnnpack/dwconv.h>
#include <xnnpack/intrinsics-polyfill.h>
#include <xnnpack/math.h>


void xnn_f16_dwconv_minmax_ukernel_5f5m5l16c8s4r__fma3_acc2(
    size_t channels,
    size_t output_width,
    const void** input,
    const void* weights,
    void* output,
    intptr_t input_stride,
    size_t output_increment,
    size_t input_offset,
    const void* zero,
    size_t kernel_size,
    void* buffer,
    const union xnn_f16_minmax_params params[restrict XNN_MIN_ELEMENTS(1)]) XNN_OOB_READS
{
  assert(channels != 0);
  assert(output_width != 0);
  assert(kernel_size > 5);

  const __m256 vmax = _mm256_load_ps(params->avx.max);
  const __m256 vmin = _mm256_load_ps(params->avx.min);
  do {
    const uint16_t* w = weights;

    // First pass to process 5 inputs.
    {
      uint16_t* b = buffer;
      const uint16_t* i0 = input[0];
      assert(i0 != NULL);
      if XNN_UNPREDICTABLE(i0 != zero) {
        i0 = (const uint16_t*) ((uintptr_t) i0 + input_offset);
      }
      const uint16_t* i1 = input[1];
      assert(i1 != NULL);
      if XNN_UNPREDICTABLE(i1 != zero) {
        i1 = (const uint16_t*) ((uintptr_t) i1 + input_offset);
      }
      const uint16_t* i2 = input[2];
      assert(i2 != NULL);
      if XNN_UNPREDICTABLE(i2 != zero) {
        i2 = (const uint16_t*) ((uintptr_t) i2 + input_offset);
      }
      const uint16_t* i3 = input[3];
      assert(i3 != NULL);
      if XNN_UNPREDICTABLE(i3 != zero) {
        i3 = (const uint16_t*) ((uintptr_t) i3 + input_offset);
      }
      const uint16_t* i4 = input[4];
      assert(i4 != NULL);
      if XNN_UNPREDICTABLE(i4 != zero) {
        i4 = (const uint16_t*) ((uintptr_t) i4 + input_offset);
      }
      input += 5;

      // Process c channels and write to buffer.
      size_t c = round_up_po2(channels, 4);
      for (; c >= 16; c -= 16) {
        __m256 vacc01234567p0 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) w));
        __m256 vacc89ABCDEFp0 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 8)));

        const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i0)));
        const __m256 vi0x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i0 + 8)));
        i0 += 16;

        const __m256 vk0x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 16)));
        const __m256 vk0x89ABCDEF = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 24)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi0x01234567, vk0x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));
        vacc89ABCDEFp0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi0x89ABCDEF, vk0x89ABCDEF, vacc89ABCDEFp0), _MM_FROUND_TO_NEAREST_INT));
        const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i1)));
        const __m256 vi1x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i1 + 8)));
        i1 += 16;

        const __m256 vk1x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 32)));
        const __m256 vk1x89ABCDEF = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 40)));
        __m256 vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_mul_ps(vi1x01234567, vk1x01234567), _MM_FROUND_TO_NEAREST_INT));
        __m256 vacc89ABCDEFp1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_mul_ps(vi1x89ABCDEF, vk1x89ABCDEF), _MM_FROUND_TO_NEAREST_INT));
        const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i2)));
        const __m256 vi2x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i2 + 8)));
        i2 += 16;

        const __m256 vk2x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 48)));
        const __m256 vk2x89ABCDEF = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 56)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi2x01234567, vk2x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));
        vacc89ABCDEFp0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi2x89ABCDEF, vk2x89ABCDEF, vacc89ABCDEFp0), _MM_FROUND_TO_NEAREST_INT));
        const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i3)));
        const __m256 vi3x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i3 + 8)));
        i3 += 16;

        const __m256 vk3x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 64)));
        const __m256 vk3x89ABCDEF = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 72)));
        vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi3x01234567, vk3x01234567, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));
        vacc89ABCDEFp1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi3x89ABCDEF, vk3x89ABCDEF, vacc89ABCDEFp1), _MM_FROUND_TO_NEAREST_INT));
        const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i4)));
        const __m256 vi4x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i4 + 8)));
        i4 += 16;

        const __m256 vk4x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 80)));
        const __m256 vk4x89ABCDEF = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 88)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi4x01234567, vk4x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));
        vacc89ABCDEFp0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi4x89ABCDEF, vk4x89ABCDEF, vacc89ABCDEFp0), _MM_FROUND_TO_NEAREST_INT));

        w += 96;

        // Add up all accumulators to vacc01234567p0
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_add_ps(vacc01234567p0, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));
        vacc89ABCDEFp0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_add_ps(vacc89ABCDEFp0, vacc89ABCDEFp1), _MM_FROUND_TO_NEAREST_INT));

        _mm_store_si128((__m128i*) b, _mm256_cvtps_ph(vacc01234567p0, _MM_FROUND_TO_NEAREST_INT));
        _mm_store_si128((__m128i*) (b + 8), _mm256_cvtps_ph(vacc89ABCDEFp0, _MM_FROUND_TO_NEAREST_INT));
        b += 16;
      }

      for (; c >= 8; c -= 8) {
        __m256 vacc01234567p0 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) w));


        const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i0)));
        i0 += 8;

        const __m256 vk0x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 8)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi0x01234567, vk0x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i1)));
        i1 += 8;

        const __m256 vk1x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 16)));
        __m256 vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_mul_ps(vi1x01234567, vk1x01234567), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i2)));
        i2 += 8;

        const __m256 vk2x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 24)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi2x01234567, vk2x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i3)));
        i3 += 8;

        const __m256 vk3x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 32)));
        vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi3x01234567, vk3x01234567, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i4)));
        i4 += 8;

        const __m256 vk4x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 40)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi4x01234567, vk4x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        w += 48;

        // Add up all accumulators to vacc01234567p0
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_add_ps(vacc01234567p0, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));

        _mm_store_si128((__m128i*) b, _mm256_cvtps_ph(vacc01234567p0, _MM_FROUND_TO_NEAREST_INT));
        b += 8;
      }

      if (c != 0) {
        assert(c >= 1);
        assert(c <= 7);
        __m256 vacc01234567p0 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) w));


        const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i0));

        const __m256 vk0x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 8)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi0x01234567, vk0x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i1));

        const __m256 vk1x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 16)));
        __m256 vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_mul_ps(vi1x01234567, vk1x01234567), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i2));

        const __m256 vk2x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 24)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi2x01234567, vk2x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i3));

        const __m256 vk3x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 32)));
        vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi3x01234567, vk3x01234567, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i4));

        const __m256 vk4x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 40)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi4x01234567, vk4x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        w += 48;

        // Add up all accumulators to vacc01234567p0
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_add_ps(vacc01234567p0, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));

        _mm_store_si128((__m128i*) b, _mm256_cvtps_ph(vacc01234567p0, _MM_FROUND_TO_NEAREST_INT));
      }
    }

    // Middle pass to process 5 inputs in each iteration.
    for (size_t ks = kernel_size - 5; ks > 5; ks -= 5) {
      uint16_t* b = buffer;
      const uint16_t* i0 = input[0];
      assert(i0 != NULL);
      if XNN_UNPREDICTABLE(i0 != zero) {
        i0 = (const uint16_t*) ((uintptr_t) i0 + input_offset);
      }
      const uint16_t* i1 = input[1];
      assert(i1 != NULL);
      if XNN_UNPREDICTABLE(i1 != zero) {
        i1 = (const uint16_t*) ((uintptr_t) i1 + input_offset);
      }
      const uint16_t* i2 = input[2];
      assert(i2 != NULL);
      if XNN_UNPREDICTABLE(i2 != zero) {
        i2 = (const uint16_t*) ((uintptr_t) i2 + input_offset);
      }
      const uint16_t* i3 = input[3];
      assert(i3 != NULL);
      if XNN_UNPREDICTABLE(i3 != zero) {
        i3 = (const uint16_t*) ((uintptr_t) i3 + input_offset);
      }
      const uint16_t* i4 = input[4];
      assert(i4 != NULL);
      if XNN_UNPREDICTABLE(i4 != zero) {
        i4 = (const uint16_t*) ((uintptr_t) i4 + input_offset);
      }
      input += 5;

      size_t c = round_up_po2(channels, 4);
      for (; c >= 16; c -= 16) {
        __m256 vacc01234567p0 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (b)));
        __m256 vacc89ABCDEFp0 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (b + 8)));


        const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i0)));
        const __m256 vi0x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i0 + 8)));
        i0 += 16;

        const __m256 vk0x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w)));
        const __m256 vk0x89ABCDEF = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 8)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi0x01234567, vk0x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));
        vacc89ABCDEFp0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi0x89ABCDEF, vk0x89ABCDEF, vacc89ABCDEFp0), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i1)));
        const __m256 vi1x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i1 + 8)));
        i1 += 16;

        const __m256 vk1x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 16)));
        const __m256 vk1x89ABCDEF = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 24)));
        __m256 vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_mul_ps(vi1x01234567, vk1x01234567), _MM_FROUND_TO_NEAREST_INT));
        __m256 vacc89ABCDEFp1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_mul_ps(vi1x89ABCDEF, vk1x89ABCDEF), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i2)));
        const __m256 vi2x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i2 + 8)));
        i2 += 16;

        const __m256 vk2x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 32)));
        const __m256 vk2x89ABCDEF = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 40)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi2x01234567, vk2x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));
        vacc89ABCDEFp0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi2x89ABCDEF, vk2x89ABCDEF, vacc89ABCDEFp0), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i3)));
        const __m256 vi3x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i3 + 8)));
        i3 += 16;

        const __m256 vk3x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 48)));
        const __m256 vk3x89ABCDEF = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 56)));
        vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi3x01234567, vk3x01234567, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));
        vacc89ABCDEFp1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi3x89ABCDEF, vk3x89ABCDEF, vacc89ABCDEFp1), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i4)));
        const __m256 vi4x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i4 + 8)));
        i4 += 16;

        const __m256 vk4x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 64)));
        const __m256 vk4x89ABCDEF = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 72)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi4x01234567, vk4x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));
        vacc89ABCDEFp0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi4x89ABCDEF, vk4x89ABCDEF, vacc89ABCDEFp0), _MM_FROUND_TO_NEAREST_INT));

        w += 80;

        // Add up all accumulators to vacc01234567p0
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_add_ps(vacc01234567p0, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));
        vacc89ABCDEFp0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_add_ps(vacc89ABCDEFp0, vacc89ABCDEFp1), _MM_FROUND_TO_NEAREST_INT));

        _mm_store_si128((__m128i*) b, _mm256_cvtps_ph(vacc01234567p0, _MM_FROUND_TO_NEAREST_INT));
        _mm_store_si128((__m128i*) (b + 8), _mm256_cvtps_ph(vacc89ABCDEFp0, _MM_FROUND_TO_NEAREST_INT));
        b += 16;
      }

      for (; c >= 8; c -= 8) {
        __m256 vacc01234567p0 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (b)));


        const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i0)));
        i0 += 8;

        const __m256 vk0x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi0x01234567, vk0x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i1)));
        i1 += 8;

        const __m256 vk1x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 8)));
        __m256 vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_mul_ps(vi1x01234567, vk1x01234567), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i2)));
        i2 += 8;

        const __m256 vk2x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 16)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi2x01234567, vk2x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i3)));
        i3 += 8;

        const __m256 vk3x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 24)));
        vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi3x01234567, vk3x01234567, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i4)));
        i4 += 8;

        const __m256 vk4x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 32)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi4x01234567, vk4x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        w += 40;

        // Add up all accumulators to vacc01234567p0
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_add_ps(vacc01234567p0, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));

        _mm_store_si128((__m128i*) b, _mm256_cvtps_ph(vacc01234567p0, _MM_FROUND_TO_NEAREST_INT));
        b += 8;
      }

      if (c != 0) {
        assert(c >= 1);
        assert(c <= 7);
        __m256 vacc01234567p0 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (b)));


        const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i0));

        const __m256 vk0x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi0x01234567, vk0x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i1));

        const __m256 vk1x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 8)));
        __m256 vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_mul_ps(vi1x01234567, vk1x01234567), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i2));

        const __m256 vk2x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 16)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi2x01234567, vk2x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i3));

        const __m256 vk3x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 24)));
        vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi3x01234567, vk3x01234567, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i4));

        const __m256 vk4x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 32)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi4x01234567, vk4x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        w += 40;

        // Add up all accumulators to vacc01234567p0
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_add_ps(vacc01234567p0, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));

        _mm_store_si128((__m128i*) b, _mm256_cvtps_ph(vacc01234567p0, _MM_FROUND_TO_NEAREST_INT));
      }
    }

    // Last pass to process up to 5 inputs.
    {
      uint16_t* b = buffer;
      const uint16_t* i0 = input[0];
      assert(i0 != NULL);
      if XNN_UNPREDICTABLE(i0 != zero) {
        i0 = (const uint16_t*) ((uintptr_t) i0 + input_offset);
      }
      const uint16_t* i1 = input[1];
      assert(i1 != NULL);
      if XNN_UNPREDICTABLE(i1 != zero) {
        i1 = (const uint16_t*) ((uintptr_t) i1 + input_offset);
      }
      const uint16_t* i2 = input[2];
      assert(i2 != NULL);
      if XNN_UNPREDICTABLE(i2 != zero) {
        i2 = (const uint16_t*) ((uintptr_t) i2 + input_offset);
      }
      const uint16_t* i3 = input[3];
      assert(i3 != NULL);
      if XNN_UNPREDICTABLE(i3 != zero) {
        i3 = (const uint16_t*) ((uintptr_t) i3 + input_offset);
      }
      const uint16_t* i4 = input[4];
      assert(i4 != NULL);
      if XNN_UNPREDICTABLE(i4 != zero) {
        i4 = (const uint16_t*) ((uintptr_t) i4 + input_offset);
      }

      size_t c = channels;
      for (; c >= 16; c -= 16) {
        __m256 vacc01234567p0 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (b)));
        __m256 vacc89ABCDEFp0 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (b + 8)));
        b += 16;


        const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i0)));
        const __m256 vi0x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i0 + 8)));
        i0 += 16;

        __m256 vk0x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w)));
        __m256 vk0x89ABCDEF = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 8)));

        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi0x01234567, vk0x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));
        vacc89ABCDEFp0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi0x89ABCDEF, vk0x89ABCDEF, vacc89ABCDEFp0), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i1)));
        const __m256 vi1x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i1 + 8)));
        i1 += 16;

        __m256 vk1x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 16)));
        __m256 vk1x89ABCDEF = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 24)));

        __m256 vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_mul_ps(vi1x01234567, vk1x01234567), _MM_FROUND_TO_NEAREST_INT));
        __m256 vacc89ABCDEFp1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_mul_ps(vi1x89ABCDEF, vk1x89ABCDEF), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i2)));
        const __m256 vi2x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i2 + 8)));
        i2 += 16;

        __m256 vk2x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 32)));
        __m256 vk2x89ABCDEF = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 40)));

        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi2x01234567, vk2x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));
        vacc89ABCDEFp0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi2x89ABCDEF, vk2x89ABCDEF, vacc89ABCDEFp0), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i3)));
        const __m256 vi3x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i3 + 8)));
        i3 += 16;

        __m256 vk3x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 48)));
        __m256 vk3x89ABCDEF = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 56)));

        vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi3x01234567, vk3x01234567, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));
        vacc89ABCDEFp1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi3x89ABCDEF, vk3x89ABCDEF, vacc89ABCDEFp1), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i4)));
        const __m256 vi4x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i4 + 8)));
        i4 += 16;

        __m256 vk4x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 64)));
        __m256 vk4x89ABCDEF = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 72)));

        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi4x01234567, vk4x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));
        vacc89ABCDEFp0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi4x89ABCDEF, vk4x89ABCDEF, vacc89ABCDEFp0), _MM_FROUND_TO_NEAREST_INT));

        w += 80;

        // Add up all accumulators to vacc01234567p0
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_add_ps(vacc01234567p0, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));
        vacc89ABCDEFp0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_add_ps(vacc89ABCDEFp0, vacc89ABCDEFp1), _MM_FROUND_TO_NEAREST_INT));

        __m256 vacc01234567 = _mm256_max_ps(vacc01234567p0, vmin);
        __m256 vacc89ABCDEF = _mm256_max_ps(vacc89ABCDEFp0, vmin);

        vacc01234567 = _mm256_min_ps(vacc01234567, vmax);
        vacc89ABCDEF = _mm256_min_ps(vacc89ABCDEF, vmax);

        _mm_storeu_si128((__m128i*) output, _mm256_cvtps_ph(vacc01234567, _MM_FROUND_TO_NEAREST_INT));
        _mm_storeu_si128((__m128i*) ((uint16_t*) output + 8), _mm256_cvtps_ph(vacc89ABCDEF, _MM_FROUND_TO_NEAREST_INT));
        output = (uint16_t*) output + 16;
      }


      for (; c >= 8; c -= 8) {
        __m256 vacc01234567p0 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (b)));
        b += 8;


        const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i0)));
        i0 += 8;

        __m256 vk0x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w)));

        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi0x01234567, vk0x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i1)));
        i1 += 8;

        __m256 vk1x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 8)));

        __m256 vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_mul_ps(vi1x01234567, vk1x01234567), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i2)));
        i2 += 8;

        __m256 vk2x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 16)));

        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi2x01234567, vk2x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i3)));
        i3 += 8;

        __m256 vk3x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 24)));

        vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi3x01234567, vk3x01234567, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i4)));
        i4 += 8;

        __m256 vk4x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 32)));

        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi4x01234567, vk4x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        w += 40;


        // Add up all accumulators to vacc01234567p0
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_add_ps(vacc01234567p0, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));

        __m256 vacc01234567 = _mm256_max_ps(vacc01234567p0, vmin);

        vacc01234567 = _mm256_min_ps(vacc01234567, vmax);

        _mm_storeu_si128((__m128i*) output, _mm256_cvtps_ph(vacc01234567, _MM_FROUND_TO_NEAREST_INT));
        output = (uint16_t*) output + 8;
      }

      if XNN_UNLIKELY(c != 0) {
        assert(c >= 1);
        assert(c <= 7);
        __m256 vacc01234567p0 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (b)));

        const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i0));
        __m256 vk0x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi0x01234567, vk0x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i1));
        __m256 vk1x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 8)));
        __m256 vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_mul_ps(vi1x01234567, vk1x01234567), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i2));
        __m256 vk2x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 16)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi2x01234567, vk2x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i3));
        __m256 vk3x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 24)));
        vacc01234567p1 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi3x01234567, vk3x01234567, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));

        const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i4));
        __m256 vk4x01234567 = _mm256_cvtph_ps(_mm_load_si128((const __m128i*) (w + 32)));
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_fmadd_ps(vi4x01234567, vk4x01234567, vacc01234567p0), _MM_FROUND_TO_NEAREST_INT));

        // Add up all accumulators to vacc01234567p0
        vacc01234567p0 = _mm256_cvtph_ps(_mm256_cvtps_ph(_mm256_add_ps(vacc01234567p0, vacc01234567p1), _MM_FROUND_TO_NEAREST_INT));

        __m256 vacc01234567 = _mm256_max_ps(vacc01234567p0, vmin);
        vacc01234567 = _mm256_min_ps(vacc01234567, vmax);

        __m128i vh01234567 = _mm256_cvtps_ph(vacc01234567, _MM_FROUND_TO_NEAREST_INT);
        if (c & 4) {
          _mm_storel_epi64((__m128i*) output, vh01234567);
          vh01234567 = _mm_unpackhi_epi64(vh01234567, vh01234567);
          output = (uint16_t*) output + 4;
        }
        if (c & 2) {
          _mm_storeu_si32(output, vh01234567);
          vh01234567 = _mm_srli_epi64(vh01234567, 32);
          output = (uint16_t*) output + 2;
        }
        if (c & 1) {
          *((uint16_t*) output) = (uint16_t) _mm_extract_epi16(vh01234567, 0);
          output = (uint16_t*) output + 1;
        }
      }

    }
    input = (const void**) (const uint16_t**) ((uintptr_t) input + input_stride);
    output = (uint16_t*) ((uintptr_t) output + output_increment);
  } while (--output_width != 0);
}