src/f32-vbinary/gen/vadd-minmax-wasmsimd-x86-x16.c

// Auto-generated file. Do not edit!
//   Template: src/f32-vbinary/vop-wasmsimd.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 <wasm_simd128.h>

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


void xnn_f32_vadd_minmax_ukernel__wasmsimd_x86_x16(
    size_t n,
    const float* a,
    const float* b,
    float* y,
    const union xnn_f32_minmax_params params[restrict XNN_MIN_ELEMENTS(1)]) XNN_DISABLE_TSAN
{
  assert(n != 0);
  assert(n % sizeof(float) == 0);
  assert(a != NULL);
  assert(b != NULL);
  assert(y != NULL);

  const v128_t vy_min = wasm_v128_load32_splat(&params->scalar.min);
  const v128_t vy_max = wasm_v128_load32_splat(&params->scalar.max);

  for (; n >= 16 * sizeof(float); n -= 16 * sizeof(float)) {
    const v128_t va0123 = wasm_v128_load(a);
    const v128_t va4567 = wasm_v128_load(a + 4);
    const v128_t va89AB = wasm_v128_load(a + 8);
    const v128_t vaCDEF = wasm_v128_load(a + 12);
    a += 16;

    const v128_t vb0123 = wasm_v128_load(b);
    const v128_t vb4567 = wasm_v128_load(b + 4);
    const v128_t vb89AB = wasm_v128_load(b + 8);
    const v128_t vbCDEF = wasm_v128_load(b + 12);
    b += 16;

    v128_t vy0123 = wasm_f32x4_add(va0123, vb0123);
    v128_t vy4567 = wasm_f32x4_add(va4567, vb4567);
    v128_t vy89AB = wasm_f32x4_add(va89AB, vb89AB);
    v128_t vyCDEF = wasm_f32x4_add(vaCDEF, vbCDEF);


    const v128_t vltmask0123 = wasm_f32x4_lt(vy0123, vy_min);
    const v128_t vltmask4567 = wasm_f32x4_lt(vy4567, vy_min);
    const v128_t vltmask89AB = wasm_f32x4_lt(vy89AB, vy_min);
    const v128_t vltmaskCDEF = wasm_f32x4_lt(vyCDEF, vy_min);

    const v128_t vngtmask0123 = wasm_f32x4_le(vy0123, vy_max);
    vy0123 = wasm_v128_bitselect(vy_min, vy0123, vltmask0123);
    const v128_t vngtmask4567 = wasm_f32x4_le(vy4567, vy_max);
    vy4567 = wasm_v128_bitselect(vy_min, vy4567, vltmask4567);
    const v128_t vngtmask89AB = wasm_f32x4_le(vy89AB, vy_max);
    vy89AB = wasm_v128_bitselect(vy_min, vy89AB, vltmask89AB);
    const v128_t vngtmaskCDEF = wasm_f32x4_le(vyCDEF, vy_max);
    vyCDEF = wasm_v128_bitselect(vy_min, vyCDEF, vltmaskCDEF);

    vy0123 = wasm_v128_bitselect(vy0123, vy_max, vngtmask0123);
    vy4567 = wasm_v128_bitselect(vy4567, vy_max, vngtmask4567);
    vy89AB = wasm_v128_bitselect(vy89AB, vy_max, vngtmask89AB);
    vyCDEF = wasm_v128_bitselect(vyCDEF, vy_max, vngtmaskCDEF);

    wasm_v128_store(y, vy0123);
    wasm_v128_store(y + 4, vy4567);
    wasm_v128_store(y + 8, vy89AB);
    wasm_v128_store(y + 12, vyCDEF);
    y += 16;
  }
  for (; n >= 4 * sizeof(float); n -= 4 * sizeof(float)) {
    const v128_t va = wasm_v128_load(a);
    a += 4;

    const v128_t vb = wasm_v128_load(b);
    b += 4;

    v128_t vy = wasm_f32x4_add(va, vb);

    const v128_t vltmask = wasm_f32x4_lt(vy, vy_min);
    const v128_t vngtmask = wasm_f32x4_le(vy, vy_max);
    vy = wasm_v128_bitselect(vy_min, vy, vltmask);
    vy = wasm_v128_bitselect(vy, vy_max, vngtmask);

    wasm_v128_store(y, vy);
    y += 4;
  }
  if XNN_UNLIKELY(n != 0) {
    const v128_t va = wasm_v128_load(a);
    const v128_t vb = wasm_v128_load(b);

    v128_t vy = wasm_f32x4_add(va, vb);

    const v128_t vltmask = wasm_f32x4_lt(vy, vy_min);
    const v128_t vngtmask = wasm_f32x4_le(vy, vy_max);
    vy = wasm_v128_bitselect(vy_min, vy, vltmask);
    vy = wasm_v128_bitselect(vy, vy_max, vngtmask);

    if (n & (2 * sizeof(float))) {
      *((double*) y) = wasm_f64x2_extract_lane(vy, 0);
      vy = wasm_v32x4_shuffle(vy, vy, 2, 3, 2, 3);
      y += 2;
    }
    if (n & (1 * sizeof(float))) {
      *y = wasm_f32x4_extract_lane(vy, 0);
    }
  }
}