File size: 9,739 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
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
// Auto-generated file. Do not edit!
//   Template: src/f16-vtanh/neonfp16arith-expm1minus.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/microparams.h>

void xnn_f16_vtanh_ukernel__aarch64_neonfp16arith_expm1minus_rr1_p3h2ts_div_x56(
    size_t n,
    const void* input,
    void* output,
    const union xnn_f16_tanh_params params[restrict XNN_MIN_ELEMENTS(1)]) XNN_OOB_READS
{
  assert(n != 0);
  assert(n % sizeof(uint16_t) == 0);
  assert(input != NULL);
  assert(output != NULL);

  const float16x8_t vsat_cutoff = vreinterpretq_f16_u16(vmovq_n_u16(UINT16_C(0x4482)));
  const float16x8_t vmagic_bias = vreinterpretq_f16_u16(vmovq_n_u16(UINT16_C(0x620F)));
  const float16x8_t vminus_log2e = vreinterpretq_f16_u16(vmovq_n_u16(UINT16_C(0xBDC5)));
  const float16x8_t vln2 = vreinterpretq_f16_u16(vmovq_n_u16(UINT16_C(0x398C)));
  const float16x8_t vc3 = vreinterpretq_f16_u16(vmovq_n_u16(UINT16_C(0xBD5B)));
  const float16x8_t vc2 = vreinterpretq_f16_u16(vmovq_n_u16(UINT16_C(0x4008)));
  const float16x8_t vtwo = vreinterpretq_f16_u16(vmovq_n_u16(UINT16_C(0x4000)));
  const float16x8_t vminus_one = vreinterpretq_f16_u16(vmovq_n_u16(UINT16_C(0xBC00)));
  const uint16x8_t vsign_mask = vmovq_n_u16(UINT16_C(0x8000));

  const uint16_t* i = (const uint16_t*) input;
  uint16_t* o = (uint16_t*) output;
  for (; n >= 7 * sizeof(float16x8_t); n -= 7 * sizeof(float16x8_t)) {
    const float16x8_t vx0 = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;
    const float16x8_t vx1 = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;
    const float16x8_t vx2 = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;
    const float16x8_t vx3 = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;
    const float16x8_t vx4 = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;
    const float16x8_t vx5 = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;
    const float16x8_t vx6 = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;

    float16x8_t vz0 = vabsq_f16(vx0);
    float16x8_t vz1 = vabsq_f16(vx1);
    float16x8_t vz2 = vabsq_f16(vx2);
    float16x8_t vz3 = vabsq_f16(vx3);
    float16x8_t vz4 = vabsq_f16(vx4);
    float16x8_t vz5 = vabsq_f16(vx5);
    float16x8_t vz6 = vabsq_f16(vx6);

    vz0 = vminq_f16(vz0, vsat_cutoff);
    vz1 = vminq_f16(vz1, vsat_cutoff);
    vz2 = vminq_f16(vz2, vsat_cutoff);
    vz3 = vminq_f16(vz3, vsat_cutoff);
    vz4 = vminq_f16(vz4, vsat_cutoff);
    vz5 = vminq_f16(vz5, vsat_cutoff);
    vz6 = vminq_f16(vz6, vsat_cutoff);

    float16x8_t vn0 = vfmaq_f16(vmagic_bias, vz0, vminus_log2e);
    float16x8_t vn1 = vfmaq_f16(vmagic_bias, vz1, vminus_log2e);
    float16x8_t vn2 = vfmaq_f16(vmagic_bias, vz2, vminus_log2e);
    float16x8_t vn3 = vfmaq_f16(vmagic_bias, vz3, vminus_log2e);
    float16x8_t vn4 = vfmaq_f16(vmagic_bias, vz4, vminus_log2e);
    float16x8_t vn5 = vfmaq_f16(vmagic_bias, vz5, vminus_log2e);
    float16x8_t vn6 = vfmaq_f16(vmagic_bias, vz6, vminus_log2e);

    const float16x8_t vs0 = vreinterpretq_f16_s16(vshlq_n_s16(vreinterpretq_s16_f16(vn0), 10));
    vn0 = vsubq_f16(vn0, vmagic_bias);
    const float16x8_t vs1 = vreinterpretq_f16_s16(vshlq_n_s16(vreinterpretq_s16_f16(vn1), 10));
    vn1 = vsubq_f16(vn1, vmagic_bias);
    const float16x8_t vs2 = vreinterpretq_f16_s16(vshlq_n_s16(vreinterpretq_s16_f16(vn2), 10));
    vn2 = vsubq_f16(vn2, vmagic_bias);
    const float16x8_t vs3 = vreinterpretq_f16_s16(vshlq_n_s16(vreinterpretq_s16_f16(vn3), 10));
    vn3 = vsubq_f16(vn3, vmagic_bias);
    const float16x8_t vs4 = vreinterpretq_f16_s16(vshlq_n_s16(vreinterpretq_s16_f16(vn4), 10));
    vn4 = vsubq_f16(vn4, vmagic_bias);
    const float16x8_t vs5 = vreinterpretq_f16_s16(vshlq_n_s16(vreinterpretq_s16_f16(vn5), 10));
    vn5 = vsubq_f16(vn5, vmagic_bias);
    const float16x8_t vs6 = vreinterpretq_f16_s16(vshlq_n_s16(vreinterpretq_s16_f16(vn6), 10));
    vn6 = vsubq_f16(vn6, vmagic_bias);
    const float16x8_t vt0 = vfmaq_f16(vz0, vn0, vln2);
    const float16x8_t vt1 = vfmaq_f16(vz1, vn1, vln2);
    const float16x8_t vt2 = vfmaq_f16(vz2, vn2, vln2);
    const float16x8_t vt3 = vfmaq_f16(vz3, vn3, vln2);
    const float16x8_t vt4 = vfmaq_f16(vz4, vn4, vln2);
    const float16x8_t vt5 = vfmaq_f16(vz5, vn5, vln2);
    const float16x8_t vt6 = vfmaq_f16(vz6, vn6, vln2);
    float16x8_t vp0 = vfmaq_f16(vc2, vc3, vt0);
    float16x8_t vp1 = vfmaq_f16(vc2, vc3, vt1);
    float16x8_t vp2 = vfmaq_f16(vc2, vc3, vt2);
    float16x8_t vp3 = vfmaq_f16(vc2, vc3, vt3);
    float16x8_t vp4 = vfmaq_f16(vc2, vc3, vt4);
    float16x8_t vp5 = vfmaq_f16(vc2, vc3, vt5);
    float16x8_t vp6 = vfmaq_f16(vc2, vc3, vt6);
    vp0 = vfmsq_f16(vtwo, vp0, vt0);
    vp1 = vfmsq_f16(vtwo, vp1, vt1);
    vp2 = vfmsq_f16(vtwo, vp2, vt2);
    vp3 = vfmsq_f16(vtwo, vp3, vt3);
    vp4 = vfmsq_f16(vtwo, vp4, vt4);
    vp5 = vfmsq_f16(vtwo, vp5, vt5);
    vp6 = vfmsq_f16(vtwo, vp6, vt6);

    const float16x8_t vts0 = vmulq_f16(vt0, vs0);
    const float16x8_t vsmo0 = vaddq_f16(vs0, vminus_one);
    const float16x8_t vts1 = vmulq_f16(vt1, vs1);
    const float16x8_t vsmo1 = vaddq_f16(vs1, vminus_one);
    const float16x8_t vts2 = vmulq_f16(vt2, vs2);
    const float16x8_t vsmo2 = vaddq_f16(vs2, vminus_one);
    const float16x8_t vts3 = vmulq_f16(vt3, vs3);
    const float16x8_t vsmo3 = vaddq_f16(vs3, vminus_one);
    const float16x8_t vts4 = vmulq_f16(vt4, vs4);
    const float16x8_t vsmo4 = vaddq_f16(vs4, vminus_one);
    const float16x8_t vts5 = vmulq_f16(vt5, vs5);
    const float16x8_t vsmo5 = vaddq_f16(vs5, vminus_one);
    const float16x8_t vts6 = vmulq_f16(vt6, vs6);
    const float16x8_t vsmo6 = vaddq_f16(vs6, vminus_one);
    const float16x8_t vemo0 = vfmsq_f16(vsmo0, vp0, vts0);
    const float16x8_t vemo1 = vfmsq_f16(vsmo1, vp1, vts1);
    const float16x8_t vemo2 = vfmsq_f16(vsmo2, vp2, vts2);
    const float16x8_t vemo3 = vfmsq_f16(vsmo3, vp3, vts3);
    const float16x8_t vemo4 = vfmsq_f16(vsmo4, vp4, vts4);
    const float16x8_t vemo5 = vfmsq_f16(vsmo5, vp5, vts5);
    const float16x8_t vemo6 = vfmsq_f16(vsmo6, vp6, vts6);
    const float16x8_t vepo0 = vaddq_f16(vemo0, vtwo);
    const float16x8_t vepo1 = vaddq_f16(vemo1, vtwo);
    const float16x8_t vepo2 = vaddq_f16(vemo2, vtwo);
    const float16x8_t vepo3 = vaddq_f16(vemo3, vtwo);
    const float16x8_t vepo4 = vaddq_f16(vemo4, vtwo);
    const float16x8_t vepo5 = vaddq_f16(vemo5, vtwo);
    const float16x8_t vepo6 = vaddq_f16(vemo6, vtwo);

    float16x8_t vy0 = vdivq_f16(vemo0, vepo0);
    float16x8_t vy1 = vdivq_f16(vemo1, vepo1);
    float16x8_t vy2 = vdivq_f16(vemo2, vepo2);
    float16x8_t vy3 = vdivq_f16(vemo3, vepo3);
    float16x8_t vy4 = vdivq_f16(vemo4, vepo4);
    float16x8_t vy5 = vdivq_f16(vemo5, vepo5);
    float16x8_t vy6 = vdivq_f16(vemo6, vepo6);

    vy0 = vbslq_f16(vsign_mask, vx0, vy0);
    vy1 = vbslq_f16(vsign_mask, vx1, vy1);
    vy2 = vbslq_f16(vsign_mask, vx2, vy2);
    vy3 = vbslq_f16(vsign_mask, vx3, vy3);
    vy4 = vbslq_f16(vsign_mask, vx4, vy4);
    vy5 = vbslq_f16(vsign_mask, vx5, vy5);
    vy6 = vbslq_f16(vsign_mask, vx6, vy6);

    vst1q_u16(o, vreinterpretq_u16_f16(vy0)); o += 8;
    vst1q_u16(o, vreinterpretq_u16_f16(vy1)); o += 8;
    vst1q_u16(o, vreinterpretq_u16_f16(vy2)); o += 8;
    vst1q_u16(o, vreinterpretq_u16_f16(vy3)); o += 8;
    vst1q_u16(o, vreinterpretq_u16_f16(vy4)); o += 8;
    vst1q_u16(o, vreinterpretq_u16_f16(vy5)); o += 8;
    vst1q_u16(o, vreinterpretq_u16_f16(vy6)); o += 8;
  }
  for (; n >= 1 * sizeof(float16x8_t); n -= 1 * sizeof(float16x8_t)) {
    const float16x8_t vx = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;

    float16x8_t vz = vabsq_f16(vx);

    vz = vminq_f16(vz, vsat_cutoff);

    float16x8_t vn = vfmaq_f16(vmagic_bias, vz, vminus_log2e);

    const float16x8_t vs = vreinterpretq_f16_s16(vshlq_n_s16(vreinterpretq_s16_f16(vn), 10));
    vn = vsubq_f16(vn, vmagic_bias);
    const float16x8_t vt = vfmaq_f16(vz, vn, vln2);
    float16x8_t vp = vfmaq_f16(vc2, vc3, vt);
    vp = vfmsq_f16(vtwo, vp, vt);

    const float16x8_t vts = vmulq_f16(vt, vs);
    const float16x8_t vsmo = vaddq_f16(vs, vminus_one);
    const float16x8_t vemo = vfmsq_f16(vsmo, vp, vts);
    const float16x8_t vepo = vaddq_f16(vemo, vtwo);

    float16x8_t vy = vdivq_f16(vemo, vepo);

    vy = vbslq_f16(vsign_mask, vx, vy);

    vst1q_u16(o, vreinterpretq_u16_f16(vy)); o += 8;
  }

  if (n != 0) {
    const float16x8_t vx = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;
    float16x8_t vz = vabsq_f16(vx);
    vz = vminq_f16(vz, vsat_cutoff);

    float16x8_t vn = vfmaq_f16(vmagic_bias, vz, vminus_log2e);
    const float16x8_t vs = vreinterpretq_f16_s16(vshlq_n_s16(vreinterpretq_s16_f16(vn), 10));
    vn = vsubq_f16(vn, vmagic_bias);
    const float16x8_t vt = vfmaq_f16(vz, vn, vln2);
    float16x8_t vp = vfmaq_f16(vc2, vc3, vt);
    vp = vfmsq_f16(vtwo, vp, vt);

    const float16x8_t vts = vmulq_f16(vt, vs);
    const float16x8_t vsmo = vaddq_f16(vs, vminus_one);
    const float16x8_t vemo = vfmsq_f16(vsmo, vp, vts);
    const float16x8_t vepo = vaddq_f16(vemo, vtwo);

    float16x8_t vy = vdivq_f16(vemo, vepo);

    vy = vbslq_f16(vsign_mask, vx, vy);

    float16x4_t vy_lo = vget_low_f16(vy);
    if (n & 4 * sizeof(uint16_t)) {
      vst1_u16(o, vreinterpret_u16_f16(vy_lo)); o += 4;
      vy_lo = vget_high_f16(vy);
    }
    if (n & 2 * sizeof(uint16_t)) {
      vst1_lane_u32((void*) o, vreinterpret_u32_f16(vy_lo), 0); o+= 2;
      vy_lo = vext_f16(vy_lo, vy_lo, 2);
    }
    if (n & 1 * sizeof(uint16_t)) {
      vst1_lane_u16(o, vreinterpret_u16_f16(vy_lo), 0);
    }
  }
}