src/qs8-gemm/gen/1x4c8-minmax-fp32-wasmsimd-mul16-ld128.c

// Auto-generated file. Do not edit!
//   Template: src/qs8-gemm/MRx4c8-wasmsimd-mul16.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/gemm.h>
#include <xnnpack/math.h>


void xnn_qs8_gemm_minmax_fp32_ukernel_1x4c8__wasmsimd_mul16_ld128(
    size_t mr,
    size_t nc,
    size_t kc,
    const int8_t* restrict a,
    size_t a_stride,
    const void* restrict w,
    int8_t* restrict c,
    size_t cm_stride,
    size_t cn_stride,
    const union xnn_qs8_conv_minmax_params params[restrict XNN_MIN_ELEMENTS(1)]) XNN_OOB_READS
{
  assert(mr != 0);
  assert(mr <= 1);
  assert(nc != 0);
  assert(kc != 0);
  assert(kc % sizeof(int8_t) == 0);
  assert(a != NULL);
  assert(w != NULL);
  assert(c != NULL);

  kc = round_up_po2(kc, 8);
  const int8_t* a0 = a;
  int8_t* c0 = c;

  do {
    v128_t vacc0x0 = wasm_f32x4_replace_lane(wasm_f32x4_const_splat(0.0f), 0, ((const float*) w)[0]);
    v128_t vacc0x1 = wasm_f32x4_replace_lane(wasm_f32x4_const_splat(0.0f), 0, ((const float*) w)[1]);
    v128_t vacc0x2 = wasm_f32x4_replace_lane(wasm_f32x4_const_splat(0.0f), 0, ((const float*) w)[2]);
    v128_t vacc0x3 = wasm_f32x4_replace_lane(wasm_f32x4_const_splat(0.0f), 0, ((const float*) w)[3]);
    w = (const void*) ((const int32_t*) w + 4);

    size_t k = 0;
    while (k < kc) {
      const v128_t vxa0 = wasm_i16x8_load8x8(a0);
      a0 += 8;

      const v128_t vb01 = wasm_v128_load(w);
      const v128_t vxb0 = wasm_i16x8_extend_low_i8x16(vb01);
      const v128_t vxb1 = wasm_i16x8_extend_high_i8x16(vb01);

      const v128_t vprod0x0 = wasm_i16x8_mul(vxb0, vxa0);
      vacc0x0 = wasm_i32x4_add(vacc0x0, wasm_i32x4_extend_low_i16x8(vprod0x0));

      const v128_t vprod0x1 = wasm_i16x8_mul(vxb1, vxa0);
      vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_extend_low_i16x8(vprod0x1));
      vacc0x0 = wasm_i32x4_add(vacc0x0, wasm_i32x4_extend_high_i16x8(vprod0x0));

      vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_extend_high_i16x8(vprod0x1));
      const v128_t vb23 = wasm_v128_load((const int8_t*) w + 16);
      const v128_t vxb2 = wasm_i16x8_extend_low_i8x16(vb23);
      const v128_t vxb3 = wasm_i16x8_extend_high_i8x16(vb23);

      const v128_t vprod0x2 = wasm_i16x8_mul(vxb2, vxa0);
      vacc0x2 = wasm_i32x4_add(vacc0x2, wasm_i32x4_extend_low_i16x8(vprod0x2));

      const v128_t vprod0x3 = wasm_i16x8_mul(vxb3, vxa0);
      vacc0x3 = wasm_i32x4_add(vacc0x3, wasm_i32x4_extend_low_i16x8(vprod0x3));
      vacc0x2 = wasm_i32x4_add(vacc0x2, wasm_i32x4_extend_high_i16x8(vprod0x2));

      vacc0x3 = wasm_i32x4_add(vacc0x3, wasm_i32x4_extend_high_i16x8(vprod0x3));

      w = (const void*) ((const int8_t*) w + 32);
      k += 8 * sizeof(int8_t);
    }

    const v128_t vacc0x02 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x0, vacc0x2, 0, 4, 1, 5), wasm_v32x4_shuffle(vacc0x0, vacc0x2, 2, 6, 3, 7));
    const v128_t vacc0x13 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x1, vacc0x3, 0, 4, 1, 5), wasm_v32x4_shuffle(vacc0x1, vacc0x3, 2, 6, 3, 7));

    v128_t vacc0x0123 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x02, vacc0x13, 0, 4, 1, 5), wasm_v32x4_shuffle(vacc0x02, vacc0x13, 2, 6, 3, 7));

    vacc0x0123 = wasm_f32x4_convert_i32x4(vacc0x0123);

    const v128_t vscale = wasm_v128_load64_splat(params->fp32_wasmsimd.scale);
    vacc0x0123 = wasm_f32x4_mul(vacc0x0123, vscale);

    const v128_t vmagic_bias = wasm_v128_load64_splat(params->fp32_wasmsimd.magic_bias);
    vacc0x0123 = wasm_f32x4_add(vacc0x0123, vmagic_bias);

    const v128_t vmagic_min = wasm_v128_load64_splat(params->fp32_wasmsimd.magic_min);
    vacc0x0123 = wasm_i32x4_max(vacc0x0123, vmagic_min);

    const v128_t vmagic_bias_less_output_zero_point = wasm_v128_load64_splat(params->fp32_wasmsimd.magic_bias_less_output_zero_point);
    vacc0x0123 = wasm_i32x4_sub(vacc0x0123, vmagic_bias_less_output_zero_point);

    v128_t vacc00x0123 = wasm_i16x8_narrow_i32x4(vacc0x0123, vacc0x0123);

    v128_t vout = wasm_i8x16_narrow_i16x8(vacc00x0123, vacc00x0123);

    const v128_t voutput_max = wasm_v128_load64_splat(params->fp32_wasmsimd.output_max);
    vout = wasm_i8x16_min(vout, voutput_max);

    if (nc >= 4) {
      *((float*) c0) = (float) wasm_f32x4_extract_lane(vout, 0);

      c0 = (int8_t*) ((uintptr_t) c0 + cn_stride);

      a0 = (const int8_t*) ((uintptr_t) a0 - kc);

      nc -= 4;
    } else {
      uint32_t vout0 = wasm_i32x4_extract_lane(vout, 0);
      if (nc & 2) {
        *((uint16_t*) c0) = (uint16_t) vout0;
        vout0 >>= 16;
        c0 += 2;
      }
      if (nc & 1) {
        *c0 = (int8_t) vout0;
      }

      nc = 0;
    }
  } while (nc != 0);
}