xref: /aosp_15_r20/external/XNNPACK/src/f32-velu/gen/velu-sse41-rr2-lut16-p3-x24.c (revision 4bdc94577ba0e567308109d787f7fec7b531ce36)

// Auto-generated file. Do not edit!
//   Template: src/f32-velu/sse-rr2-lut16-p3.c.in
//   Generator: tools/xngen
//
// Copyright 2020 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 <smmintrin.h>

#include <xnnpack/vunary.h>
#include <xnnpack/common.h>


extern XNN_INTERNAL const float xnn_table_exp2minus_k_over_16[16];

void xnn_f32_velu_ukernel__sse41_rr2_lut16_p3_x24(
    size_t n,
    const float* x,
    float* y,
    const union xnn_f32_elu_params params[restrict XNN_MIN_ELEMENTS(1)]) XNN_OOB_READS
{
  assert(n != 0);
  assert(n % sizeof(float) == 0);
  assert(x != NULL);
  assert(y != NULL);

  const __m128 vprescale = _mm_load_ps(params->sse2_rr2_lut16_p3.prescale);
  const __m128 valpha = _mm_load_ps(params->sse2_rr2_lut16_p3.alpha);
  const __m128 vbeta = _mm_load_ps(params->sse2_rr2_lut16_p3.beta);
  const __m128 vsat_cutoff = _mm_load_ps(params->sse2_rr2_lut16_p3.sat_cutoff);
  const __m128 vmagic_bias = _mm_load_ps(params->sse2_rr2_lut16_p3.magic_bias);
  const __m128 vlog2e = _mm_load_ps(params->sse2_rr2_lut16_p3.log2e);
  const __m128i vindex_mask = _mm_load_si128((const __m128i*) params->sse2_rr2_lut16_p3.index_mask);
  const __m128 vminus_ln2_hi = _mm_load_ps(params->sse2_rr2_lut16_p3.minus_ln2_hi);
  const __m128 vminus_ln2_lo = _mm_load_ps(params->sse2_rr2_lut16_p3.minus_ln2_lo);
  const __m128 vc3 = _mm_load_ps(params->sse2_rr2_lut16_p3.c3);
  const __m128 vc2 = _mm_load_ps(params->sse2_rr2_lut16_p3.c2);
  const __m128 vone = _mm_load_ps(params->sse2_rr2_lut16_p3.one);

  for (; n >= 24 * sizeof(float); n -= 24 * sizeof(float)) {
    __m128 vx0123 = _mm_loadu_ps(x);
    __m128 vx4567 = _mm_loadu_ps(x + 4);
    __m128 vx89AB = _mm_loadu_ps(x + 8);
    __m128 vxCDEF = _mm_loadu_ps(x + 12);
    __m128 vxGHIJ = _mm_loadu_ps(x + 16);
    __m128 vxKLMN = _mm_loadu_ps(x + 20);
    x += 24;

    const __m128 vz0123 = _mm_max_ps(vsat_cutoff, _mm_mul_ps(vx0123, vprescale));
    const __m128 vz4567 = _mm_max_ps(vsat_cutoff, _mm_mul_ps(vx4567, vprescale));
    const __m128 vz89AB = _mm_max_ps(vsat_cutoff, _mm_mul_ps(vx89AB, vprescale));
    const __m128 vzCDEF = _mm_max_ps(vsat_cutoff, _mm_mul_ps(vxCDEF, vprescale));
    const __m128 vzGHIJ = _mm_max_ps(vsat_cutoff, _mm_mul_ps(vxGHIJ, vprescale));
    const __m128 vzKLMN = _mm_max_ps(vsat_cutoff, _mm_mul_ps(vxKLMN, vprescale));

    __m128 vn0123 = _mm_add_ps(_mm_mul_ps(vz0123, vlog2e), vmagic_bias);
    __m128 vn4567 = _mm_add_ps(_mm_mul_ps(vz4567, vlog2e), vmagic_bias);
    __m128 vn89AB = _mm_add_ps(_mm_mul_ps(vz89AB, vlog2e), vmagic_bias);
    __m128 vnCDEF = _mm_add_ps(_mm_mul_ps(vzCDEF, vlog2e), vmagic_bias);
    __m128 vnGHIJ = _mm_add_ps(_mm_mul_ps(vzGHIJ, vlog2e), vmagic_bias);
    __m128 vnKLMN = _mm_add_ps(_mm_mul_ps(vzKLMN, vlog2e), vmagic_bias);

    const __m128i vidx0123 = _mm_slli_epi32(_mm_and_si128(_mm_castps_si128(vn0123), vindex_mask), 2);
    const __m128i ven0123 = _mm_slli_epi32(_mm_castps_si128(vn0123), 19);
    const __m128i vidx4567 = _mm_slli_epi32(_mm_and_si128(_mm_castps_si128(vn4567), vindex_mask), 2);
    const __m128i ven4567 = _mm_slli_epi32(_mm_castps_si128(vn4567), 19);
    const __m128i vidx89AB = _mm_slli_epi32(_mm_and_si128(_mm_castps_si128(vn89AB), vindex_mask), 2);
    const __m128i ven89AB = _mm_slli_epi32(_mm_castps_si128(vn89AB), 19);
    const __m128i vidxCDEF = _mm_slli_epi32(_mm_and_si128(_mm_castps_si128(vnCDEF), vindex_mask), 2);
    const __m128i venCDEF = _mm_slli_epi32(_mm_castps_si128(vnCDEF), 19);
    const __m128i vidxGHIJ = _mm_slli_epi32(_mm_and_si128(_mm_castps_si128(vnGHIJ), vindex_mask), 2);
    const __m128i venGHIJ = _mm_slli_epi32(_mm_castps_si128(vnGHIJ), 19);
    const __m128i vidxKLMN = _mm_slli_epi32(_mm_and_si128(_mm_castps_si128(vnKLMN), vindex_mask), 2);
    const __m128i venKLMN = _mm_slli_epi32(_mm_castps_si128(vnKLMN), 19);

    #if XNN_ARCH_X86_64
      const uint64_t vidx01 = (uint64_t) _mm_cvtsi128_si64(vidx0123);
      const uint64_t vidx23 = (uint64_t) _mm_extract_epi64(vidx0123, 1);
      const __m128i vl0   = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) vidx01)));
      const __m128i vl2 = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) vidx23)));
      const __m128i vl01 = _mm_insert_epi32(vl0, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) (vidx01 >> 32))), 1);
      const __m128i vl23 = _mm_insert_epi32(vl2, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) (vidx23 >> 32))), 1);
      const __m128i vl0123 = _mm_unpacklo_epi64(vl01, vl23);
      const uint64_t vidx45 = (uint64_t) _mm_cvtsi128_si64(vidx4567);
      const uint64_t vidx67 = (uint64_t) _mm_extract_epi64(vidx4567, 1);
      const __m128i vl4   = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) vidx45)));
      const __m128i vl6 = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) vidx67)));
      const __m128i vl45 = _mm_insert_epi32(vl4, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) (vidx45 >> 32))), 1);
      const __m128i vl67 = _mm_insert_epi32(vl6, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) (vidx67 >> 32))), 1);
      const __m128i vl4567 = _mm_unpacklo_epi64(vl45, vl67);
      const uint64_t vidx89 = (uint64_t) _mm_cvtsi128_si64(vidx89AB);
      const uint64_t vidxAB = (uint64_t) _mm_extract_epi64(vidx89AB, 1);
      const __m128i vl8   = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) vidx89)));
      const __m128i vlA = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) vidxAB)));
      const __m128i vl89 = _mm_insert_epi32(vl8, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) (vidx89 >> 32))), 1);
      const __m128i vlAB = _mm_insert_epi32(vlA, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) (vidxAB >> 32))), 1);
      const __m128i vl89AB = _mm_unpacklo_epi64(vl89, vlAB);
      const uint64_t vidxCD = (uint64_t) _mm_cvtsi128_si64(vidxCDEF);
      const uint64_t vidxEF = (uint64_t) _mm_extract_epi64(vidxCDEF, 1);
      const __m128i vlC   = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) vidxCD)));
      const __m128i vlE = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) vidxEF)));
      const __m128i vlCD = _mm_insert_epi32(vlC, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) (vidxCD >> 32))), 1);
      const __m128i vlEF = _mm_insert_epi32(vlE, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) (vidxEF >> 32))), 1);
      const __m128i vlCDEF = _mm_unpacklo_epi64(vlCD, vlEF);
      const uint64_t vidxGH = (uint64_t) _mm_cvtsi128_si64(vidxGHIJ);
      const uint64_t vidxIJ = (uint64_t) _mm_extract_epi64(vidxGHIJ, 1);
      const __m128i vlG   = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) vidxGH)));
      const __m128i vlI = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) vidxIJ)));
      const __m128i vlGH = _mm_insert_epi32(vlG, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) (vidxGH >> 32))), 1);
      const __m128i vlIJ = _mm_insert_epi32(vlI, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) (vidxIJ >> 32))), 1);
      const __m128i vlGHIJ = _mm_unpacklo_epi64(vlGH, vlIJ);
      const uint64_t vidxKL = (uint64_t) _mm_cvtsi128_si64(vidxKLMN);
      const uint64_t vidxMN = (uint64_t) _mm_extract_epi64(vidxKLMN, 1);
      const __m128i vlK   = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) vidxKL)));
      const __m128i vlM = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) vidxMN)));
      const __m128i vlKL = _mm_insert_epi32(vlK, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) (vidxKL >> 32))), 1);
      const __m128i vlMN = _mm_insert_epi32(vlM, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) (vidxMN >> 32))), 1);
      const __m128i vlKLMN = _mm_unpacklo_epi64(vlKL, vlMN);
    #else  // !XNN_ARCH_X86_64
      const uint32_t vidx0 = (uint32_t) _mm_cvtsi128_si32(vidx0123);
      const uint32_t vidx1 = (uint32_t) _mm_extract_epi16(vidx0123, 2);
      const uint32_t vidx2 = (uint32_t) _mm_extract_epi16(vidx0123, 4);
      const uint32_t vidx3 = (uint32_t) _mm_extract_epi16(vidx0123, 6);
      const __m128i vl0   = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidx0)));
      const __m128i vl2 = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidx2)));
      const __m128i vl01 = _mm_insert_epi32(vl0, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidx1)), 1);
      const __m128i vl23 = _mm_insert_epi32(vl2, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidx3)), 1);
      const __m128i vl0123 = _mm_unpacklo_epi64(vl01, vl23);
      const uint32_t vidx4 = (uint32_t) _mm_cvtsi128_si32(vidx4567);
      const uint32_t vidx5 = (uint32_t) _mm_extract_epi16(vidx4567, 2);
      const uint32_t vidx6 = (uint32_t) _mm_extract_epi16(vidx4567, 4);
      const uint32_t vidx7 = (uint32_t) _mm_extract_epi16(vidx4567, 6);
      const __m128i vl4   = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidx4)));
      const __m128i vl6 = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidx6)));
      const __m128i vl45 = _mm_insert_epi32(vl4, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidx5)), 1);
      const __m128i vl67 = _mm_insert_epi32(vl6, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidx7)), 1);
      const __m128i vl4567 = _mm_unpacklo_epi64(vl45, vl67);
      const uint32_t vidx8 = (uint32_t) _mm_cvtsi128_si32(vidx89AB);
      const uint32_t vidx9 = (uint32_t) _mm_extract_epi16(vidx89AB, 2);
      const uint32_t vidxA = (uint32_t) _mm_extract_epi16(vidx89AB, 4);
      const uint32_t vidxB = (uint32_t) _mm_extract_epi16(vidx89AB, 6);
      const __m128i vl8   = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidx8)));
      const __m128i vlA = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidxA)));
      const __m128i vl89 = _mm_insert_epi32(vl8, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidx9)), 1);
      const __m128i vlAB = _mm_insert_epi32(vlA, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidxB)), 1);
      const __m128i vl89AB = _mm_unpacklo_epi64(vl89, vlAB);
      const uint32_t vidxC = (uint32_t) _mm_cvtsi128_si32(vidxCDEF);
      const uint32_t vidxD = (uint32_t) _mm_extract_epi16(vidxCDEF, 2);
      const uint32_t vidxE = (uint32_t) _mm_extract_epi16(vidxCDEF, 4);
      const uint32_t vidxF = (uint32_t) _mm_extract_epi16(vidxCDEF, 6);
      const __m128i vlC   = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidxC)));
      const __m128i vlE = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidxE)));
      const __m128i vlCD = _mm_insert_epi32(vlC, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidxD)), 1);
      const __m128i vlEF = _mm_insert_epi32(vlE, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidxF)), 1);
      const __m128i vlCDEF = _mm_unpacklo_epi64(vlCD, vlEF);
      const uint32_t vidxG = (uint32_t) _mm_cvtsi128_si32(vidxGHIJ);
      const uint32_t vidxH = (uint32_t) _mm_extract_epi16(vidxGHIJ, 2);
      const uint32_t vidxI = (uint32_t) _mm_extract_epi16(vidxGHIJ, 4);
      const uint32_t vidxJ = (uint32_t) _mm_extract_epi16(vidxGHIJ, 6);
      const __m128i vlG   = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidxG)));
      const __m128i vlI = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidxI)));
      const __m128i vlGH = _mm_insert_epi32(vlG, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidxH)), 1);
      const __m128i vlIJ = _mm_insert_epi32(vlI, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidxJ)), 1);
      const __m128i vlGHIJ = _mm_unpacklo_epi64(vlGH, vlIJ);
      const uint32_t vidxK = (uint32_t) _mm_cvtsi128_si32(vidxKLMN);
      const uint32_t vidxL = (uint32_t) _mm_extract_epi16(vidxKLMN, 2);
      const uint32_t vidxM = (uint32_t) _mm_extract_epi16(vidxKLMN, 4);
      const uint32_t vidxN = (uint32_t) _mm_extract_epi16(vidxKLMN, 6);
      const __m128i vlK   = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidxK)));
      const __m128i vlM = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidxM)));
      const __m128i vlKL = _mm_insert_epi32(vlK, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidxL)), 1);
      const __m128i vlMN = _mm_insert_epi32(vlM, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidxN)), 1);
      const __m128i vlKLMN = _mm_unpacklo_epi64(vlKL, vlMN);
    #endif  // XNN_ARCH_X86_64

    vn0123 = _mm_sub_ps(vn0123, vmagic_bias);
    __m128 vs0123 = _mm_castsi128_ps(_mm_add_epi32(vl0123, ven0123));
    vn4567 = _mm_sub_ps(vn4567, vmagic_bias);
    __m128 vs4567 = _mm_castsi128_ps(_mm_add_epi32(vl4567, ven4567));
    vn89AB = _mm_sub_ps(vn89AB, vmagic_bias);
    __m128 vs89AB = _mm_castsi128_ps(_mm_add_epi32(vl89AB, ven89AB));
    vnCDEF = _mm_sub_ps(vnCDEF, vmagic_bias);
    __m128 vsCDEF = _mm_castsi128_ps(_mm_add_epi32(vlCDEF, venCDEF));
    vnGHIJ = _mm_sub_ps(vnGHIJ, vmagic_bias);
    __m128 vsGHIJ = _mm_castsi128_ps(_mm_add_epi32(vlGHIJ, venGHIJ));
    vnKLMN = _mm_sub_ps(vnKLMN, vmagic_bias);
    __m128 vsKLMN = _mm_castsi128_ps(_mm_add_epi32(vlKLMN, venKLMN));

    __m128 vt0123 = _mm_add_ps(_mm_mul_ps(vn0123, vminus_ln2_hi), vz0123);
    __m128 vt4567 = _mm_add_ps(_mm_mul_ps(vn4567, vminus_ln2_hi), vz4567);
    __m128 vt89AB = _mm_add_ps(_mm_mul_ps(vn89AB, vminus_ln2_hi), vz89AB);
    __m128 vtCDEF = _mm_add_ps(_mm_mul_ps(vnCDEF, vminus_ln2_hi), vzCDEF);
    __m128 vtGHIJ = _mm_add_ps(_mm_mul_ps(vnGHIJ, vminus_ln2_hi), vzGHIJ);
    __m128 vtKLMN = _mm_add_ps(_mm_mul_ps(vnKLMN, vminus_ln2_hi), vzKLMN);

    vt0123 = _mm_add_ps(_mm_mul_ps(vn0123, vminus_ln2_lo), vt0123);
    vt4567 = _mm_add_ps(_mm_mul_ps(vn4567, vminus_ln2_lo), vt4567);
    vt89AB = _mm_add_ps(_mm_mul_ps(vn89AB, vminus_ln2_lo), vt89AB);
    vtCDEF = _mm_add_ps(_mm_mul_ps(vnCDEF, vminus_ln2_lo), vtCDEF);
    vtGHIJ = _mm_add_ps(_mm_mul_ps(vnGHIJ, vminus_ln2_lo), vtGHIJ);
    vtKLMN = _mm_add_ps(_mm_mul_ps(vnKLMN, vminus_ln2_lo), vtKLMN);

    __m128 vp0123 = _mm_add_ps(_mm_mul_ps(vc3, vt0123), vc2);
    __m128 vp4567 = _mm_add_ps(_mm_mul_ps(vc3, vt4567), vc2);
    __m128 vp89AB = _mm_add_ps(_mm_mul_ps(vc3, vt89AB), vc2);
    __m128 vpCDEF = _mm_add_ps(_mm_mul_ps(vc3, vtCDEF), vc2);
    __m128 vpGHIJ = _mm_add_ps(_mm_mul_ps(vc3, vtGHIJ), vc2);
    __m128 vpKLMN = _mm_add_ps(_mm_mul_ps(vc3, vtKLMN), vc2);

    vp0123 = _mm_mul_ps(vp0123, vt0123);
    vp4567 = _mm_mul_ps(vp4567, vt4567);
    vp89AB = _mm_mul_ps(vp89AB, vt89AB);
    vpCDEF = _mm_mul_ps(vpCDEF, vtCDEF);
    vpGHIJ = _mm_mul_ps(vpGHIJ, vtGHIJ);
    vpKLMN = _mm_mul_ps(vpKLMN, vtKLMN);

    vt0123 = _mm_mul_ps(vt0123, vs0123);
    vs0123 = _mm_sub_ps(vs0123, vone);
    vt4567 = _mm_mul_ps(vt4567, vs4567);
    vs4567 = _mm_sub_ps(vs4567, vone);
    vt89AB = _mm_mul_ps(vt89AB, vs89AB);
    vs89AB = _mm_sub_ps(vs89AB, vone);
    vtCDEF = _mm_mul_ps(vtCDEF, vsCDEF);
    vsCDEF = _mm_sub_ps(vsCDEF, vone);
    vtGHIJ = _mm_mul_ps(vtGHIJ, vsGHIJ);
    vsGHIJ = _mm_sub_ps(vsGHIJ, vone);
    vtKLMN = _mm_mul_ps(vtKLMN, vsKLMN);
    vsKLMN = _mm_sub_ps(vsKLMN, vone);

    vp0123 = _mm_add_ps(_mm_mul_ps(vp0123, vt0123), vt0123);
    vp4567 = _mm_add_ps(_mm_mul_ps(vp4567, vt4567), vt4567);
    vp89AB = _mm_add_ps(_mm_mul_ps(vp89AB, vt89AB), vt89AB);
    vpCDEF = _mm_add_ps(_mm_mul_ps(vpCDEF, vtCDEF), vtCDEF);
    vpGHIJ = _mm_add_ps(_mm_mul_ps(vpGHIJ, vtGHIJ), vtGHIJ);
    vpKLMN = _mm_add_ps(_mm_mul_ps(vpKLMN, vtKLMN), vtKLMN);

    const __m128 ve0123 = _mm_mul_ps(_mm_add_ps(vp0123, vs0123), valpha);
    const __m128 ve4567 = _mm_mul_ps(_mm_add_ps(vp4567, vs4567), valpha);
    const __m128 ve89AB = _mm_mul_ps(_mm_add_ps(vp89AB, vs89AB), valpha);
    const __m128 veCDEF = _mm_mul_ps(_mm_add_ps(vpCDEF, vsCDEF), valpha);
    const __m128 veGHIJ = _mm_mul_ps(_mm_add_ps(vpGHIJ, vsGHIJ), valpha);
    const __m128 veKLMN = _mm_mul_ps(_mm_add_ps(vpKLMN, vsKLMN), valpha);

    vx0123 = _mm_mul_ps(vx0123, vbeta);
    vx4567 = _mm_mul_ps(vx4567, vbeta);
    vx89AB = _mm_mul_ps(vx89AB, vbeta);
    vxCDEF = _mm_mul_ps(vxCDEF, vbeta);
    vxGHIJ = _mm_mul_ps(vxGHIJ, vbeta);
    vxKLMN = _mm_mul_ps(vxKLMN, vbeta);

    const __m128 vy0123 = _mm_blendv_ps(vx0123, ve0123, vx0123);
    const __m128 vy4567 = _mm_blendv_ps(vx4567, ve4567, vx4567);
    const __m128 vy89AB = _mm_blendv_ps(vx89AB, ve89AB, vx89AB);
    const __m128 vyCDEF = _mm_blendv_ps(vxCDEF, veCDEF, vxCDEF);
    const __m128 vyGHIJ = _mm_blendv_ps(vxGHIJ, veGHIJ, vxGHIJ);
    const __m128 vyKLMN = _mm_blendv_ps(vxKLMN, veKLMN, vxKLMN);

    _mm_storeu_ps(y, vy0123);
    _mm_storeu_ps(y + 4, vy4567);
    _mm_storeu_ps(y + 8, vy89AB);
    _mm_storeu_ps(y + 12, vyCDEF);
    _mm_storeu_ps(y + 16, vyGHIJ);
    _mm_storeu_ps(y + 20, vyKLMN);
    y += 24;
  }
  for (; n >= 4 * sizeof(float); n -= 4 * sizeof(float)) {
    __m128 vx = _mm_loadu_ps(x);
    x += 4;

    const __m128 vz = _mm_max_ps(vsat_cutoff, _mm_mul_ps(vx, vprescale));

    __m128 vn = _mm_add_ps(_mm_mul_ps(vz, vlog2e), vmagic_bias);

    const __m128i ven = _mm_slli_epi32(_mm_castps_si128(vn), 19);
    const __m128i vidx = _mm_slli_epi32(_mm_and_si128(_mm_castps_si128(vn), vindex_mask), 2);
    #if XNN_ARCH_X86_64
      const uint64_t vidx_lo = (uint64_t) _mm_cvtsi128_si64(vidx);
      const uint64_t vidx_hi = (uint64_t) _mm_extract_epi64(vidx, 1);
      const __m128i vl_ll   = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) vidx_lo)));
      const __m128i vl_hl = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) vidx_hi)));
      const __m128i vl_lo = _mm_insert_epi32(vl_ll, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) (vidx_lo >> 32))), 1);
      const __m128i vl_hi = _mm_insert_epi32(vl_hl, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) (vidx_hi >> 32))), 1);
    #else  // !XNN_ARCH_X86_64
      const __m128i vl_ll = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) _mm_cvtsi128_si32(vidx))));
      const __m128i vl_hl = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) _mm_extract_epi16(vidx, 4))));
      const __m128i vl_lo = _mm_insert_epi32(vl_ll, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) _mm_extract_epi16(vidx, 2))), 1);
      const __m128i vl_hi = _mm_insert_epi32(vl_hl, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) _mm_extract_epi16(vidx, 6))), 1);
    #endif  // XNN_ARCH_X86_64
    const __m128i vl = _mm_unpacklo_epi64(vl_lo, vl_hi);
    __m128 vs = _mm_castsi128_ps(_mm_add_epi32(vl, ven));
    vn = _mm_sub_ps(vn, vmagic_bias);

    __m128 vt = _mm_add_ps(_mm_mul_ps(vn, vminus_ln2_hi), vz);
    vt = _mm_add_ps(_mm_mul_ps(vn, vminus_ln2_lo), vt);

    __m128 vp = _mm_add_ps(_mm_mul_ps(vc3, vt), vc2);
    vp = _mm_mul_ps(vp, vt);

    vt = _mm_mul_ps(vt, vs);
    vs = _mm_sub_ps(vs, vone);
    vp = _mm_add_ps(_mm_mul_ps(vp, vt), vt);
    const __m128 ve = _mm_mul_ps(_mm_add_ps(vp, vs), valpha);

    vx = _mm_mul_ps(vx, vbeta);
    const __m128 vy = _mm_blendv_ps(vx, ve, vx);

    _mm_storeu_ps(y, vy);
    y += 4;
  }
  if XNN_UNLIKELY(n != 0) {
    __m128 vx = _mm_loadu_ps(x);

    const __m128 vz = _mm_max_ps(vsat_cutoff, _mm_mul_ps(vx, vprescale));

    __m128 vn = _mm_add_ps(_mm_mul_ps(vz, vlog2e), vmagic_bias);

    const __m128i ven = _mm_slli_epi32(_mm_castps_si128(vn), 19);
    const __m128i vidx = _mm_slli_epi32(_mm_and_si128(_mm_castps_si128(vn), vindex_mask), 2);
    #if XNN_ARCH_X86_64
      const uint64_t vidx_lo = (uint64_t) _mm_cvtsi128_si64(vidx);
      const uint64_t vidx_hi = (uint64_t) _mm_extract_epi64(vidx, 1);
      const __m128i vl_ll   = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) vidx_lo)));
      const __m128i vl_hl = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) vidx_hi)));
      const __m128i vl_lo = _mm_insert_epi32(vl_ll, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) (vidx_lo >> 32))), 1);
      const __m128i vl_hi = _mm_insert_epi32(vl_hl, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) (vidx_hi >> 32))), 1);
    #else  // !XNN_ARCH_X86_64
      const __m128i vl_ll = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) _mm_cvtsi128_si32(vidx))));
      const __m128i vl_hl = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) _mm_extract_epi16(vidx, 4))));
      const __m128i vl_lo = _mm_insert_epi32(vl_ll, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) _mm_extract_epi16(vidx, 2))), 1);
      const __m128i vl_hi = _mm_insert_epi32(vl_hl, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + (uint32_t) _mm_extract_epi16(vidx, 6))), 1);
    #endif  // XNN_ARCH_X86_64
    const __m128i vl = _mm_unpacklo_epi64(vl_lo, vl_hi);
    __m128 vs = _mm_castsi128_ps(_mm_add_epi32(vl, ven));
    vn = _mm_sub_ps(vn, vmagic_bias);

    __m128 vt = _mm_add_ps(_mm_mul_ps(vn, vminus_ln2_hi), vz);
    vt = _mm_add_ps(_mm_mul_ps(vn, vminus_ln2_lo), vt);

    __m128 vp = _mm_add_ps(_mm_mul_ps(vc3, vt), vc2);
    vp = _mm_mul_ps(vp, vt);

    vt = _mm_mul_ps(vt, vs);
    vs = _mm_sub_ps(vs, vone);
    vp = _mm_add_ps(_mm_mul_ps(vp, vt), vt);
    const __m128 ve = _mm_mul_ps(_mm_add_ps(vp, vs), valpha);

    vx = _mm_mul_ps(vx, vbeta);
    __m128 vy = _mm_blendv_ps(vx, ve, vx);

    if (n & (2 * sizeof(float))) {
      _mm_storel_pi((__m64*) y, vy);
      vy = _mm_movehl_ps(vy, vy);
      y += 2;
    }
    if (n & (1 * sizeof(float))) {
      _mm_store_ss(y, vy);
    }
  }
}