mix.cc 8.5 KB
Newer Older
T
tensor-tang 已提交
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
/* Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserved.
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 * http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License. */

#include "paddle/fluid/operators/jit/more/mix/mix.h"
#include "paddle/fluid/operators/jit/kernels.h"
#include "paddle/fluid/operators/jit/registry.h"
#include "paddle/fluid/platform/cpu_info.h"

namespace paddle {
namespace operators {
namespace jit {
namespace more {
namespace mix {

void VSigmoid(const T* x, T* y, int n) {
  const float min = SIGMOID_THRESHOLD_MIN;
  const float max = SIGMOID_THRESHOLD_MAX;
  for (int i = 0; i < n; ++i) {
    y[i] = (x[i] < min) ? min : ((x[i] > max) ? max : x[i]);
    y[i] = static_cast<T>(0) - y[i];
  }
T
tensor-tang 已提交
33
  auto compute = Get<KernelType::kVExp, XYNTuples<T>, platform::CPUPlace>(n);
T
tensor-tang 已提交
34 35 36 37 38 39 40 41
  compute(y, y, n);
  for (int i = 0; i < n; ++i) {
    y[i] = static_cast<T>(1) / (static_cast<T>(1) + y[i]);
  }
}

void VTanh(const T* x, T* y, int n) {
  const T a = 2, b = -1;
T
tensor-tang 已提交
42 43 44
  auto compute_scal = Get<kVScal, AXYNTuples<T>, platform::CPUPlace>(n);
  auto compute_addbias = Get<kVAddBias, AXYNTuples<T>, platform::CPUPlace>(n);
  auto compute_sigmoid = Get<kVSigmoid, XYNTuples<T>, platform::CPUPlace>(n);
T
tensor-tang 已提交
45 46 47 48 49 50
  compute_scal(&a, x, y, n);
  compute_sigmoid(y, y, n);
  compute_scal(&a, y, y, n);
  compute_addbias(&b, y, y, n);
}

T
tensor-tang 已提交
51
void Softmax(const T* x, T* y, int n, int bs) {
T
tensor-tang 已提交
52 53 54 55 56 57 58 59 60 61
  auto compute_hmax =
      KernelFuncs<kHMax, XRNTuples<T>, platform::CPUPlace>::Cache().At(n);
  auto compute_hsum =
      KernelFuncs<kHSum, XRNTuples<T>, platform::CPUPlace>::Cache().At(n);
  auto compute_vscal =
      KernelFuncs<kVScal, AXYNTuples<T>, platform::CPUPlace>::Cache().At(n);
  auto compute_vaddbias =
      KernelFuncs<kVAddBias, AXYNTuples<T>, platform::CPUPlace>::Cache().At(n);
  auto compute_vexp =
      KernelFuncs<kVExp, XYNTuples<T>, platform::CPUPlace>::Cache().At(n);
T
tensor-tang 已提交
62

T
tensor-tang 已提交
63 64 65 66 67 68 69 70 71 72 73 74 75 76
  for (int i = 0; i < bs; ++i) {
    T scalar;
    compute_hmax(x, &scalar, n);
    scalar = static_cast<T>(0) - scalar;
    compute_vaddbias(&scalar, x, y, n);  // x - max
    compute_vexp(y, y, n);
    compute_hsum(y, &scalar, n);
    scalar = static_cast<T>(1) / scalar;
    compute_vscal(&scalar, y, y, n);
    x += n;
    y += n;
  }
}

77
void (*getActFunc(KernelType type, int d))(const T*, T*, int) {  // NOLINT
T
tensor-tang 已提交
78 79 80 81 82 83 84 85
  if (type == kVSigmoid) {
    return Get<kVSigmoid, XYNTuples<T>, platform::CPUPlace>(d);
  } else if (type == kVRelu) {
    return Get<kVRelu, XYNTuples<T>, platform::CPUPlace>(d);
  } else if (type == kVTanh) {
    return Get<kVTanh, XYNTuples<T>, platform::CPUPlace>(d);
  } else if (type == kVIdentity) {
    return Get<kVIdentity, XYNTuples<T>, platform::CPUPlace>(d);
86 87 88
  }
  PADDLE_THROW("Not support type: %s", type);
  return nullptr;
T
tensor-tang 已提交
89 90
}

91 92 93 94 95 96 97 98 99 100
void LSTMCtHt(lstm_t* step, const lstm_attr_t* attr) {
  T* gates = reinterpret_cast<T*>(step->gates);
  const T* ct_1 = reinterpret_cast<const T*>(step->ct_1);
  T* ct = reinterpret_cast<T*>(step->ct);
  T* ht = reinterpret_cast<T*>(step->ht);
  const T* wp = reinterpret_cast<const T*>(step->wp);
  T* checked = reinterpret_cast<T*>(step->checked);
  const int d = attr->d;
  const int d2 = d * 2;
  const int d3 = d * 3;
T
tensor-tang 已提交
101 102 103
  auto vmul_d = Get<kVMul, XYZNTuples<T>, platform::CPUPlace>(d);
  auto vadd_d = Get<kVAdd, XYZNTuples<T>, platform::CPUPlace>(d);
  auto vadd_d2 = Get<kVAdd, XYZNTuples<T>, platform::CPUPlace>(d2);
104 105
  auto act_gate_d = getActFunc(attr->act_gate, d);
  auto act_gate_d2 = getActFunc(attr->act_gate, d2);
T
tensor-tang 已提交
106
  auto act_gate_d3 = getActFunc(attr->act_gate, d3);
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
  auto act_cand_d = getActFunc(attr->act_cand, d);
  auto act_cell_d = getActFunc(attr->act_cell, d);

  if (attr->use_peephole) {
    vmul_d(wp, ct_1, checked, d);
    vmul_d(wp + d, ct_1, checked + d, d);
    vadd_d2(checked, gates + d, gates + d, d2);
    act_gate_d2(gates + d, gates + d, d2);
  } else {
    act_gate_d3(gates + d, gates + d, d3);
  }

  // C_t = C_t-1 * fgated + cand_gated * igated
  act_cand_d(gates, gates, d);
  vmul_d(gates, gates + d, gates + d, d);
  vmul_d(ct_1, gates + d2, gates + d2, d);
  vadd_d(gates + d, gates + d2, ct, d);

  if (attr->use_peephole) {
    // get ogated
    vmul_d(wp + d2, ct, gates + d, d);
    vadd_d(gates + d, gates + d3, gates + d3, d);
    act_gate_d(gates + d3, gates + d3, d);
  }
  // H_t = act_cell(C_t) * ogated
  act_cell_d(ct, gates + d2, d);
  vmul_d(gates + d2, gates + d3, ht, d);
T
tensor-tang 已提交
134 135
}

136 137 138 139 140 141 142
void LSTMC1H1(lstm_t* step, const lstm_attr_t* attr) {
  T* gates = reinterpret_cast<T*>(step->gates);
  T* ct = reinterpret_cast<T*>(step->ct);
  T* ht = reinterpret_cast<T*>(step->ht);
  int d = attr->d;
  int d2 = d * 2;
  int d3 = d * 3;
T
tensor-tang 已提交
143 144
  auto vmul_d = Get<kVMul, XYZNTuples<T>, platform::CPUPlace>(d);
  auto vadd_d = Get<kVAdd, XYZNTuples<T>, platform::CPUPlace>(d);
145 146 147 148 149 150 151 152 153 154 155 156
  auto act_gate_d = getActFunc(attr->act_gate, d);
  auto act_cand_d = getActFunc(attr->act_cand, d);
  auto act_cell_d = getActFunc(attr->act_cell, d);
  /* C_t = igated * cgated*/
  act_gate_d(gates + d, gates + d, d);
  act_cand_d(gates, gates, d);
  vmul_d(gates, gates + d, ct, d);
  if (attr->use_peephole) {
    // get outgated, put W_oc * C_t on igated
    const T* wp = reinterpret_cast<const T*>(step->wp);
    vmul_d(wp + d2, ct, gates + d, d);
    vadd_d(gates + d, gates + d3, gates + d3, d);
T
tensor-tang 已提交
157
  }
158 159 160 161 162 163 164 165 166 167 168 169 170 171
  /* H_t = act_cell(C_t) * ogated */
  act_gate_d(gates + d3, gates + d3, d);
  act_cell_d(ct, gates + d2, d);
  vmul_d(gates + d2, gates + d3, ht, d);
}

// compute h1 without h0
void GRUH1(gru_t* step, const gru_attr_t* attr) {
  T* gates = reinterpret_cast<T*>(step->gates);
  T* ht = reinterpret_cast<T*>(step->ht);
  int d = attr->d;
  int d2 = d * 2;
  auto act_gate = getActFunc(attr->act_gate, d);
  auto act_cand = getActFunc(attr->act_cand, d);
T
tensor-tang 已提交
172
  auto vmul_d = Get<kVMul, XYZNTuples<T>, platform::CPUPlace>(d);
173 174 175 176 177 178 179 180 181 182 183 184
  act_gate(gates, gates, d);
  act_cand(gates + d2, gates + d2, d);
  vmul_d(gates, gates + d2, ht, d);
}

// compute the first part of GRU: ht = act_gate(r) * ht_1
void GRUHtPart1(gru_t* step, const gru_attr_t* attr) {
  // W: {W_update, W_reset; W_state}
  T* gates = reinterpret_cast<T*>(step->gates);
  T* ht = reinterpret_cast<T*>(step->ht);
  const T* ht_1 = reinterpret_cast<const T*>(step->ht_1);
  auto act_gate = getActFunc(attr->act_gate, attr->d);
T
tensor-tang 已提交
185
  auto vmul_d = Get<kVMul, XYZNTuples<T>, platform::CPUPlace>(attr->d);
186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208
  act_gate(gates + attr->d, gates + attr->d, attr->d);
  vmul_d(ht_1, gates + attr->d, ht, attr->d);
}

// compute the second part of GRU:
// ht = act_gate(u) * act_cand(s) + (1-act_gate(u)) * ht_1
void GRUHtPart2(gru_t* step, const gru_attr_t* attr) {
  T* gates = reinterpret_cast<T*>(step->gates);
  T* ht = reinterpret_cast<T*>(step->ht);
  const T* ht_1 = reinterpret_cast<const T*>(step->ht_1);
  int d = attr->d;
  auto act_gate = getActFunc(attr->act_gate, d);
  auto act_cand = getActFunc(attr->act_cand, d);
  T* y = gates + d * 2;
  act_gate(gates, gates, d);
  act_cand(y, y, d);
  // out = zt*ht~ + (1-zt)*ht_1
  for (int i = 0; i < d; ++i) {
    ht[i] = gates[i] * y[i] + (static_cast<T>(1) - gates[i]) * ht_1[i];
  }
}

// TODO(TJ): tuning me
T
tensor-tang 已提交
209
bool VSigmoidKernel::UseMe(const int& d) const { return true; }
210

T
tensor-tang 已提交
211
bool VTanhKernel::UseMe(const int& d) const { return true; }
212

T
tensor-tang 已提交
213 214
bool SoftmaxKernel::UseMe(const int& d) const { return true; }

T
tensor-tang 已提交
215
bool LSTMCtHtKernel::UseMe(const lstm_attr_t& attr) const { return true; }
216

T
tensor-tang 已提交
217
bool LSTMC1H1Kernel::UseMe(const lstm_attr_t& attr) const { return true; }
218

T
tensor-tang 已提交
219
bool GRUH1Kernel::UseMe(const gru_attr_t& attr) const { return true; }
T
tensor-tang 已提交
220

T
tensor-tang 已提交
221
bool GRUHtPart1Kernel::UseMe(const gru_attr_t& attr) const { return true; }
T
tensor-tang 已提交
222

T
tensor-tang 已提交
223
bool GRUHtPart2Kernel::UseMe(const gru_attr_t& attr) const { return true; }
T
tensor-tang 已提交
224 225 226 227 228 229 230 231 232

}  // namespace mix
}  // namespace more
}  // namespace jit
}  // namespace operators
}  // namespace paddle

namespace mix = paddle::operators::jit::more::mix;

233 234
#define REGISTER_MORE_KERNEL(key, func) \
  REGISTER_JITKERNEL_MORE(key, mix, mix::func##Kernel)
T
tensor-tang 已提交
235

T
tensor-tang 已提交
236 237
REGISTER_MORE_KERNEL(kVSigmoid, VSigmoid);
REGISTER_MORE_KERNEL(kVTanh, VTanh);
T
tensor-tang 已提交
238
REGISTER_MORE_KERNEL(kSoftmax, Softmax);
T
tensor-tang 已提交
239 240 241 242 243
REGISTER_MORE_KERNEL(kLSTMCtHt, LSTMCtHt);
REGISTER_MORE_KERNEL(kLSTMC1H1, LSTMC1H1);
REGISTER_MORE_KERNEL(kGRUH1, GRUH1);
REGISTER_MORE_KERNEL(kGRUHtPart1, GRUHtPart1);
REGISTER_MORE_KERNEL(kGRUHtPart2, GRUHtPart2);
T
tensor-tang 已提交
244 245

#undef REGISTER_MORE_KERNEL