File size: 5,281 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
// Copyright 2019 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.

$assert CHANNEL_TILE >= 1
$assert KERNEL_TILE >= 2
$assert ACCUMULATORS >= 1
$assert ACTIVATION in ["LINEAR", "MINMAX"]
$assert ACTIVATION != "LINEAR" or not WASM
$ABC = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"
#include <assert.h>

#include <xnnpack/dwconv.h>
#include <xnnpack/math.h>


$MIN_F32 = "__builtin_wasm_min_f32" if WASM else "math_min_f32"
$MAX_F32 = "__builtin_wasm_max_f32" if WASM else "math_max_f32"
$SUFFIX = {"LINEAR": "", "MINMAX": "_minmax"}[ACTIVATION]
$PARAMS = {"LINEAR": "xnn_f32_default_params", "MINMAX": "xnn_f32_minmax_params"}[ACTIVATION]
void xnn_f32_dwconv${SUFFIX}_ukernel_${KERNEL_TILE}p${CHANNEL_TILE}c__${"wasm" if WASM else "scalar"}${"" if ACCUMULATORS == 1 else "_acc%d" % ACCUMULATORS}(
    size_t channels,
    size_t output_width,
    const float** input,
    const float* weights,
    float* output,
    intptr_t input_stride,
    size_t output_increment,
    size_t input_offset,
    const float* zero,
    const union ${PARAMS} params[restrict XNN_MIN_ELEMENTS(1)])
{
  assert(channels != 0);
  assert(output_width != 0);

  $if ACTIVATION == "MINMAX":
    const float vmin = params->scalar.min;
    const float vmax = params->scalar.max;
  do {
    $for K in range(KERNEL_TILE):
      const float* i${K} = input[${K}];
      assert(i${K} != NULL);
      if XNN_UNPREDICTABLE(i${K} != zero) {
        i${K} = (const float*) ((uintptr_t) i${K} + input_offset);
      }
    input = (const float**) ((uintptr_t) input + input_stride);

    size_t c = channels;
    const float* w = weights;
    $if CHANNEL_TILE > 1:
      for (; c >= ${CHANNEL_TILE}; c -= ${CHANNEL_TILE}) {
        $for C in range(CHANNEL_TILE):
          float vacc${C}p0 = w[${C}];

        $for K in range(KERNEL_TILE):

          $for C in range(CHANNEL_TILE):
            const float vi${K}x${C} = i${K}[${C}];
          i${K} += ${CHANNEL_TILE};

          $for C in range(CHANNEL_TILE):
            const float vk${K}x${C} = w[${(K + 1) * CHANNEL_TILE + C}];
            $if 1 <= K < ACCUMULATORS:
              float vacc${C}p${K} = vi${K}x${C} * vk${K}x${C};
            $else:
              vacc${C}p${K % ACCUMULATORS} = math_muladd_f32(vi${K}x${C}, vk${K}x${C}, vacc${C}p${K % ACCUMULATORS});

        w += ${(KERNEL_TILE + 1) * CHANNEL_TILE};

        $if ACCUMULATORS > 1:
          // Add up all accumulators to vacc${ABC[0:CHANNEL_TILE]}p0
          $ACC_SLICE = 1
          $while ACC_SLICE < ACCUMULATORS:
            $for A in range(0, ACCUMULATORS, ACC_SLICE * 2):
              $if A + ACC_SLICE < ACCUMULATORS:
                $for C in range(CHANNEL_TILE):
                  vacc${C}p${A} = vacc${C}p${A} + vacc${C}p${A + ACC_SLICE};
            $ACC_SLICE *= 2

        $if ACTIVATION == "MINMAX":
          $for C in range(CHANNEL_TILE):
            float vacc${C} = ${MAX_F32}(vacc${C}p0, vmin);

          $for C in range(CHANNEL_TILE):
            vacc${C} = ${MIN_F32}(vacc${C}, vmax);

          $for C in range(CHANNEL_TILE):
            output[${C}] = vacc${C};
        $else:
          $for C in range(CHANNEL_TILE):
            output[${C}] = vacc${C}p0;
        output += ${CHANNEL_TILE};
      }
      for (; c >= 1; c -= 1) {
        float vacc0p0 = *w++;

        $for K in range(KERNEL_TILE):
          const float vi${K} = *i${K}++;
          const float vk${K} = w[${(K + 1) * CHANNEL_TILE - 1}];
          $if 1 <= K < ACCUMULATORS:
            float vacc0p${K} = vi${K} * vk${K};
          $else:
            vacc0p${K % ACCUMULATORS} = math_muladd_f32(vi${K}, vk${K}, vacc0p${K % ACCUMULATORS});

        $if ACCUMULATORS > 1:
          // Add up all accumulators to vacc${ABC[0:CHANNEL_TILE]}p0
          $ACC_SLICE = 1
          $while ACC_SLICE < ACCUMULATORS:
            $for A in range(0, ACCUMULATORS, ACC_SLICE * 2):
              $if A + ACC_SLICE < ACCUMULATORS:
                vacc0p${A} = vacc0p${A} + vacc0p${A + ACC_SLICE};
            $ACC_SLICE *= 2

        $if ACTIVATION == "MINMAX":
          float vacc0 = ${MAX_F32}(vacc0p0, vmin);
          vacc0 = ${MIN_F32}(vacc0, vmax);
          *output++ = vacc0;
        $else:
          *output++ = vacc0p0;
      }
    $else:
      do {
        float vacc0p0 = w[0];
        $for K in range(KERNEL_TILE):

          const float vi${K} = *i${K}++;
          const float vk${K} = w[${K+1}];
          $if 1 <= K < ACCUMULATORS:
            float vacc0p${K} = vi${K} * vk${K};
          $else:
            vacc0p${K % ACCUMULATORS} = math_muladd_f32(vi${K}, vk${K}, vacc0p${K % ACCUMULATORS});

        w += ${KERNEL_TILE + 1};

        $ACC_STEP = 1
        $while ACC_STEP < ACCUMULATORS:
          $for A in range(0, ACCUMULATORS, ACC_STEP * 2):
            $if A + ACC_STEP < ACCUMULATORS:
              vacc0p${A} += vacc0p${A + ACC_STEP};
          $ACC_STEP *= 2

        $if ACTIVATION == "MINMAX":
          float vacc0 = ${MAX_F32}(vacc0p0, vmin);
          vacc0 = ${MIN_F32}(vacc0, vmax);
          *output++ = vacc0;
        $else:
          *output++ = vacc0p0;
      } while (--c != 0);

    output = (float*) ((uintptr_t) output + output_increment);
  } while (--output_width != 0);
}