File size: 2,419 Bytes
8b7c501
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
// Auto-generated file. Do not edit!
//   Template: src/f16-rsum/neonfp16arith.c.in
//   Generator: tools/xngen
//
// Copyright 2023 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 <arm_neon.h>

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


void xnn_f16_rsum_ukernel__neonfp16arith_x32_acc2(
    size_t batch,
    const void* input,
    void* output,
    const union xnn_f16_scale_params params[restrict XNN_MIN_ELEMENTS(1)])
{
  assert(batch != 0);
  assert(batch % sizeof(uint16_t) == 0);
  assert(input != NULL);
  assert(output != NULL);

  const uint16_t* i = (const uint16_t*) input;
  uint16_t* o = (uint16_t*) output;
  float16x8_t vacc0 = vreinterpretq_f16_u16(vmovq_n_u16(0));
  float16x8_t vacc1 = vreinterpretq_f16_u16(vmovq_n_u16(0));
  for (; batch >= 32 * sizeof(uint16_t); batch -= 32 * sizeof(uint16_t)) {
    const float16x8_t vt0 = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;
    const float16x8_t vt1 = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;
    const float16x8_t vt2 = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;
    const float16x8_t vt3 = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;

    vacc0 = vaddq_f16(vacc0, vt0);
    vacc1 = vaddq_f16(vacc1, vt1);
    vacc0 = vaddq_f16(vacc0, vt2);
    vacc1 = vaddq_f16(vacc1, vt3);
  }
  vacc0 = vaddq_f16(vacc0, vacc1);
  for (; batch >= 8 * sizeof(uint16_t); batch -= 8 * sizeof(uint16_t)) {
    const float16x8_t vt = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;
    vacc0 = vaddq_f16(vacc0, vt);
  }
  const float16x4_t vscale = vreinterpret_f16_u16(vld1_dup_u16(&params->fp16arith.scale));
  float16x4_t vacc = vadd_f16(vget_low_f16(vacc0), vget_high_f16(vacc0));
  if XNN_UNLIKELY(batch & (4 * sizeof(uint16_t))) {
    const float16x4_t vt = vreinterpret_f16_u16(vld1_u16(i)); i += 4;
    vacc = vadd_f16(vacc, vt);
  }
  vacc = vpadd_f16(vacc, vacc);
  if XNN_UNLIKELY(batch & (2 * sizeof(uint16_t))) {
    const float16x4_t vt = vreinterpret_f16_u32(vld1_dup_u32((const void*) i)); i += 2;
    vacc = vadd_f16(vacc, vt);
  }
  vacc = vpadd_f16(vacc, vacc);
  if XNN_UNLIKELY(batch & (1 * sizeof(uint16_t))) {
    const float16x4_t vt = vreinterpret_f16_u16(vld1_dup_u16(i));
    vacc = vadd_f16(vacc, vt);
  }
  vacc = vmul_f16(vacc, vscale);
  vst1_lane_u16(o, vreinterpret_u16_f16(vacc), 0);
}