elementwise_grad_base.h 68.7 KB
Newer Older
1
/* Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

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. */

#pragma once

17
#include "paddle/phi/backends/all_context.h"
18
#include "paddle/phi/backends/gpu/gpu_info.h"
19
#include "paddle/phi/core/dense_tensor.h"
20
#include "paddle/phi/kernels/funcs/common_shape.h"
21 22 23 24 25 26 27
#include "paddle/phi/kernels/funcs/elementwise_utils.h"
#include "paddle/phi/kernels/funcs/for_range.h"

#if defined(__NVCC__) || defined(__HIPCC__)
// See Note [ Why still include the fluid headers? ]
#include "paddle/fluid/memory/memcpy.h"
#include "paddle/fluid/platform/device/gpu/gpu_device_function.h"
28
#include "paddle/phi/backends/gpu/gpu_launch_config.h"
29
#include "paddle/phi/kernels/primitive/kernel_primitives.h"
30 31

#endif
32

33 34 35 36 37
#ifdef __HIPCC__
constexpr int ELEMWISE_MAX_BLOCK_DIM = 256;
#else
constexpr int ELEMWISE_MAX_BLOCK_DIM = 1024;
#endif
38

39 40 41
#define BLOCK_X 32
#define BLOCK_Y 32

42 43 44 45 46 47 48
#define GetDivMod(dividend, divisor, div, mod) \
  do {                                         \
    const auto dividend_copy = dividend;       \
    *div = dividend_copy / divisor;            \
    *mod = dividend_copy % divisor;            \
  } while (0)

49
namespace phi {
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289
namespace funcs {
using DDim = phi::DDim;

template <typename T, typename DX_OP, typename DY_OP, typename Tout = T>
void CommonGradBroadcastCPU(const DenseTensor &x,
                            const DenseTensor &y,
                            const DenseTensor &out,
                            const DenseTensor &dout,
                            DenseTensor *dx,
                            DenseTensor *dy,
                            int *x_dims_array,
                            int *y_dims_array,
                            int *out_dims_array,
                            int max_dim,
                            const CPUContext &ctx,
                            DX_OP dx_op,
                            DY_OP dy_op) {
  std::vector<int> index_array(max_dim, 0);
  const T *x_data = x.data<T>();
  const T *y_data = y.data<T>();
  const Tout *out_data = out.data<Tout>();
  const Tout *dout_data = dout.data<Tout>();
  T *dx_data = dx == nullptr ? nullptr : ctx.Alloc<T>(dx);
  T *dy_data = dy == nullptr ? nullptr : ctx.Alloc<T>(dy);
  if (dx_data != nullptr) {
    memset(dx_data, 0, dx->numel() * sizeof(T));
  }
  if (dy_data != nullptr) {
    memset(dy_data, 0, dy->numel() * sizeof(T));
  }
  const int out_size = std::accumulate(
      out_dims_array, out_dims_array + max_dim, 1, std::multiplies<int>());
  int x_index, y_index;
  for (int out_index = 0; out_index < out_size; ++out_index) {
    x_index = GetElementwiseIndex(x_dims_array, max_dim, index_array.data());
    y_index = GetElementwiseIndex(y_dims_array, max_dim, index_array.data());
    if (dx_data != nullptr) {
      dx_data[x_index] += dx_op(x_data[x_index],
                                y_data[y_index],
                                out_data[out_index],
                                dout_data[out_index]);
    }
    if (dy_data != nullptr) {
      dy_data[y_index] += dy_op(x_data[x_index],
                                y_data[y_index],
                                out_data[out_index],
                                dout_data[out_index]);
    }

    UpdateElementwiseIndexArray(out_dims_array, max_dim, index_array.data());
  }
}

template <typename T, typename DX_OP, typename DY_OP, typename Tout = T>
static void ElemwiseGradBroadcast1CPU(const T *x,
                                      const T *y,
                                      const Tout *out,
                                      const Tout *dout,
                                      int h,
                                      int w,
                                      bool is_xsize_larger,
                                      DX_OP dx_op,
                                      DY_OP dy_op,
                                      T *dx,
                                      T *dy) {
  if (is_xsize_larger) {
    for (int i = 0; i < h; ++i) {
      for (int j = 0; j < w; ++j) {
        int x_offset = i * w + j;
        if (dx != nullptr) {
          dx[x_offset] =
              dx_op(x[x_offset], y[j], out[x_offset], dout[x_offset]);
        }
        if (dy != nullptr) {
          T tmp = dy_op(x[x_offset], y[j], out[x_offset], dout[x_offset]);
          if (i == 0) {
            dy[j] = tmp;
          } else {
            dy[j] += tmp;
          }
        }
      }
    }
  } else {  // x.dims < y.dims, broadcast for x.
    for (int i = 0; i < h; ++i) {
      for (int j = 0; j < w; ++j) {
        int y_offset = i * w + j;
        if (dy != nullptr) {
          dy[y_offset] =
              dy_op(x[j], y[y_offset], out[y_offset], dout[y_offset]);
        }
        if (dx != nullptr) {
          T tmp = dx_op(x[j], y[y_offset], out[y_offset], dout[y_offset]);
          if (i == 0) {
            dx[j] = tmp;
          } else {
            dx[j] += tmp;
          }
        }
      }
    }
  }
}

template <typename T, typename DX_OP, typename DY_OP, typename Tout = T>
static void ElemwiseGradBroadcast2CPU(const T *x,
                                      const T *y,
                                      const Tout *out,
                                      const Tout *dout,
                                      int pre,
                                      int n,
                                      int post,
                                      bool is_xsize_larger,
                                      DX_OP dx_op,
                                      DY_OP dy_op,
                                      T *dx,
                                      T *dy) {
  if (is_xsize_larger) {
    for (int i = 0; i < pre; ++i) {
      for (int j = 0; j < n; ++j) {
        for (int k = 0; k < post; ++k) {
          int x_offset = i * n * post + j * post + k;
          if (dx != nullptr) {
            dx[x_offset] =
                dx_op(x[x_offset], y[j], out[x_offset], dout[x_offset]);
          }
          if (dy != nullptr) {
            T tmp = dy_op(x[x_offset], y[j], out[x_offset], dout[x_offset]);
            if (i == 0 && k == 0) {
              dy[j] = tmp;
            } else {
              dy[j] += tmp;
            }
          }
        }
      }
    }
  } else {  // x.dims < y.dims, broadcast for x.
    for (int i = 0; i < pre; ++i) {
      for (int j = 0; j < n; ++j) {
        for (int k = 0; k < post; ++k) {
          int y_offset = i * n * post + j * post + k;
          if (dy != nullptr) {
            dy[y_offset] =
                dy_op(x[j], y[y_offset], out[y_offset], dout[y_offset]);
          }
          if (dx != nullptr) {
            T tmp = dx_op(x[j], y[y_offset], out[y_offset], dout[y_offset]);
            if (i == 0 && k == 0) {
              dx[j] = tmp;
            } else {
              dx[j] += tmp;
            }
          }
        }
      }
    }
  }
}

template <typename T, typename DX_OP, typename DY_OP, typename Tout = T>
void CommonElementwiseBroadcastBackward(const CPUContext &ctx,
                                        const DDim &x_dims,
                                        const DDim &y_dims,
                                        const DenseTensor &x,
                                        const DenseTensor &y,
                                        const DenseTensor &out,
                                        const DenseTensor &dout,
                                        int axis,
                                        DenseTensor *dx,
                                        DenseTensor *dy,
                                        DX_OP dx_op,
                                        DY_OP dy_op) {
  int max_dim = std::max(x_dims.size(), y_dims.size());
  axis = (axis == -1 ? std::abs(x_dims.size() - y_dims.size()) : axis);
  std::vector<int> x_dims_array(max_dim);
  std::vector<int> y_dims_array(max_dim);
  std::vector<int> out_dims_array(max_dim);
  GetBroadcastDimsArrays(x_dims,
                         y_dims,
                         x_dims_array.data(),
                         y_dims_array.data(),
                         out_dims_array.data(),
                         max_dim,
                         axis);
  // for inplace strategy. memset will make dx and dout clear and get wrong
  // result.
  if (dx && dx->IsSharedBufferWith(dout)) {
    dx->clear();
    dx->mutable_data<T>(x_dims, ctx.GetPlace());
  }

  VLOG(3) << "CommonElementwiseBroadcastBackward xdims:"
          << phi::make_ddim(x_dims_array)
          << " ydim:" << phi::make_ddim(y_dims_array);

  CommonGradBroadcastCPU<T, DX_OP, DY_OP, Tout>(x,
                                                y,
                                                out,
                                                dout,
                                                dx,
                                                dy,
                                                x_dims_array.data(),
                                                y_dims_array.data(),
                                                out_dims_array.data(),
                                                max_dim,
                                                ctx,
                                                dx_op,
                                                dy_op);
}

template <typename T, typename DX_OP, typename DY_OP, typename Tout = T>
void ElemwiseGradComputeWithBroadcast(const CPUContext &ctx,
                                      const DDim &x_dims,
                                      const DDim &y_dims,
                                      const DenseTensor &x,
                                      const DenseTensor &y,
                                      const DenseTensor &out,
                                      const DenseTensor &dout,
                                      int axis,
                                      DenseTensor *dx,
                                      DenseTensor *dy,
                                      DX_OP dx_op,
                                      DY_OP dy_op) {
  bool is_xsize_larger = true;

  int max_dim = x_dims.size();
  if (x_dims.size() < y_dims.size()) {
    is_xsize_larger = false;
    max_dim = y_dims.size();
  }

  axis = (axis == -1 ? std::abs(x_dims.size() - y_dims.size()) : axis);
  PADDLE_ENFORCE_GE(
      axis,
      0,
      errors::InvalidArgument(
          "Axis should be great than or equal to 0, but received axis is %d.",
          axis));
290
  PADDLE_ENFORCE_LE(axis,
291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402
                    max_dim,
                    errors::InvalidArgument(
                        "Axis should be less than %d, but received axis is %d.",
                        max_dim,
                        axis));

  int pre, n, post, is_run_common_broadcast, axis_trim = 0;
  if (is_xsize_larger) {
    auto y_dims_trimed = TrimTrailingSingularDims(y_dims);
    axis_trim = (y_dims_trimed.size() == 0) ? x_dims.size() : axis;
    GetMidDims(x_dims,
               y_dims_trimed,
               axis_trim,
               &pre,
               &n,
               &post,
               &is_run_common_broadcast);
  } else {
    auto x_dims_trimed = TrimTrailingSingularDims(x_dims);
    axis_trim = (x_dims_trimed.size() == 0) ? y_dims.size() : axis;
    GetMidDims(y_dims,
               x_dims_trimed,
               axis_trim,
               &pre,
               &n,
               &post,
               &is_run_common_broadcast);
  }
  // special case for common backward implementation.
  if (is_run_common_broadcast) {
    CommonElementwiseBroadcastBackward<T, DX_OP, DY_OP, Tout>(
        ctx, x_dims, y_dims, x, y, out, dout, axis, dx, dy, dx_op, dy_op);
    return;
  }
  if (post == 1) {
    ElemwiseGradBroadcast1CPU(x.data<T>(),
                              y.data<T>(),
                              out.data<Tout>(),
                              dout.data<Tout>(),
                              pre,
                              n,
                              is_xsize_larger,
                              dx_op,
                              dy_op,
                              dx == nullptr ? nullptr : ctx.Alloc<T>(dx),
                              dy == nullptr ? nullptr : ctx.Alloc<T>(dy));
  } else {
    ElemwiseGradBroadcast2CPU(x.data<T>(),
                              y.data<T>(),
                              out.data<Tout>(),
                              dout.data<Tout>(),
                              pre,
                              n,
                              post,
                              is_xsize_larger,
                              dx_op,
                              dy_op,
                              dx == nullptr ? nullptr : ctx.Alloc<T>(dx),
                              dy == nullptr ? nullptr : ctx.Alloc<T>(dy));
  }
}

template <typename T, typename DX_OP, typename DY_OP, typename Tout = T>
struct ElemwiseGradNoBroadcast {
  const T *x_;
  const T *y_;
  const Tout *out_;
  const Tout *dout_;

  HOSTDEVICE void operator()(size_t i) {
    if (dx_ != nullptr) {
      dx_[i] = dx_op_(x_[i], y_[i], out_[i], dout_[i]);
    }
    if (dy_ != nullptr) {
      dy_[i] = dy_op_(x_[i], y_[i], out_[i], dout_[i]);
    }
  }

  DX_OP dx_op_;
  DY_OP dy_op_;
  T *dx_;
  T *dy_;
};

template <typename DeviceContext,
          typename T,
          typename DX_OP,
          typename DY_OP,
          typename Tout = T>
void ElemwiseGradComputeNoBroadcast(const DeviceContext &dev_ctx,
                                    const DDim &x_dim,
                                    const DDim &y_dim,
                                    const DenseTensor &x,
                                    const DenseTensor &y,
                                    const DenseTensor &out,
                                    const DenseTensor &dout,
                                    int axis,
                                    DenseTensor *dx,
                                    DenseTensor *dy,
                                    DX_OP dx_op,
                                    DY_OP dy_op) {
  size_t N = static_cast<size_t>(phi::product(x_dim));
  phi::funcs::ForRange<DeviceContext> for_range(dev_ctx, N);
  for_range(ElemwiseGradNoBroadcast<T, DX_OP, DY_OP, Tout>{
      x.data<T>(),
      y.data<T>(),
      out.data<Tout>(),
      dout.data<Tout>(),
      dx_op,
      dy_op,
      dx == nullptr ? nullptr : dev_ctx.template Alloc<T>(dx),
      dy == nullptr ? nullptr : dev_ctx.template Alloc<T>(dy)});
403 404
}

405
#if defined(__NVCC__) || defined(__HIPCC__)
406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482
// Suppose only has contiguous dims
static inline bool CheckContiguousDims(const std::vector<int> &broadcast_pos) {
  for (int i = 1; i < broadcast_pos.size(); ++i) {
    if (broadcast_pos[i] != broadcast_pos[i - 1] + 1) {
      return false;
    }
  }
  return true;
}

inline void ComputeBroadcastTranspositionArray(const int *x_one_indexs,
                                               int *x_trans_indexs,
                                               const int max_dim,
                                               const int x_one_size) {
  int diff = max_dim - x_one_size;
  std::copy_n(x_one_indexs, x_one_size, x_trans_indexs + diff);
  int p = 0;
  int q = diff;
  for (int i = 0; i < max_dim; ++i) {
    if (q < max_dim && i == x_trans_indexs[q]) {
      ++q;
    } else {
      x_trans_indexs[p++] = i;
    }
  }
}

// Check input can be split into 2 parts
static inline bool SplitDims(const std::vector<int> &y_broadcast_pos,
                             int max_dim) {
  bool can_split_dim2 = true;
  // must at start or end.
  if (y_broadcast_pos[0] != 0 &&
      y_broadcast_pos[y_broadcast_pos.size() - 1] != max_dim - 1) {
    can_split_dim2 = false;
  } else {
    for (int i = 1; i < y_broadcast_pos.size(); ++i) {
      // dim must be continue
      if (y_broadcast_pos[i] != y_broadcast_pos[i - 1] + 1) {
        can_split_dim2 = false;
        break;
      }
    }
  }
  return can_split_dim2;
}

inline void ComputeBroadcastKernelSize(int *x_dims_array,
                                       int *out_dims_array,
                                       int *x_blocks,
                                       int *x_threads,
                                       int max_dim) {
  *x_blocks = 1;
  *x_threads = 1;
  for (int i = 0; i < max_dim; i++) {
    if (x_dims_array[i] == out_dims_array[i]) {
      *x_blocks *= x_dims_array[i];
    } else {
      *x_threads *= out_dims_array[i];
    }
  }
}

template <typename T, typename OP, typename Tout = T>
static __global__ void FastCommonGradBroadcastOneCUDAKernel(const T *x,
                                                            const T *y,
                                                            const Tout *out,
                                                            const Tout *dout,
                                                            int pre,
                                                            int n,
                                                            int post,
                                                            int y_pre,
                                                            int y_n,
                                                            int y_post,
                                                            bool is_xsize,
                                                            OP op,
                                                            T *dd) {
483 484
  int tid = THREAD_ID_X;
  int bid = BLOCK_ID_X;
485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550

  T val(0);
  if (is_xsize) {
    // do reduce for x
    for (int i = tid; i < n; i += ELEMWISE_MAX_BLOCK_DIM) {
      int b_i = bid / post;
      int b_j = bid % post;
      int x_offset = b_i * n * post + b_j;
      int out_offset = b_i * n * post + i * post + b_j;

      // Get y pre rows id with x post and y_pre.
      int b_yi = bid / (post * y_pre);
      int b_yj = bid % y_post;
      int y_offset = b_yi * y_n + i * y_post + b_yj;

      if (dd) {
        val += op(x[x_offset], y[y_offset], out[out_offset], dout[out_offset]);
      }
    }
    if (dd) {
      int h = n > ELEMWISE_MAX_BLOCK_DIM ? ELEMWISE_MAX_BLOCK_DIM : n;
      val = paddle::platform::reduceSum(val, tid, h);
      if (tid == 0) {
        dd[bid] = val;
      }
    }
  } else {
    // do reduce for y
    for (int i = tid; i < n; i += ELEMWISE_MAX_BLOCK_DIM) {
      int b_i = bid / post;
      int b_j = bid % post;
      int y_offset = b_i * n * post + b_j;
      int out_offset = b_i * n * post + i * post + b_j;

      int b_yi = bid / (post * y_pre);
      int b_yj = bid % y_post;
      int x_offset = b_yi * y_n + i * y_post + b_yj;

      if (dd) {
        val += op(x[x_offset], y[y_offset], out[out_offset], dout[out_offset]);
      }
    }
    if (dd) {
      int h = n > ELEMWISE_MAX_BLOCK_DIM ? ELEMWISE_MAX_BLOCK_DIM : n;
      val = paddle::platform::reduceSum(val, tid, h);
      if (tid == 0) {
        dd[bid] = val;
      }
    }
  }
}

template <typename T, typename DY_OP, typename DX_OP, typename Tout = T>
static __global__ void FastCommonGradBroadcastAllCUDAKernel(
    const T *x,
    const T *y,
    const Tout *out,
    const Tout *dout,
    int pre,
    int n,
    int post,
    bool is_xsize_larger,
    DX_OP dx_op,
    DY_OP dy_op,
    T *dx,
    T *dy) {
551 552
  int tid = THREAD_ID_X;
  int bid = BLOCK_ID_X;
553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614

  T val(0);
  if (is_xsize_larger) {
    for (int i = tid; i < n; i += ELEMWISE_MAX_BLOCK_DIM) {
      int b_i = bid / post;
      int b_j = bid % post;
      int x_offset = b_i * n * post + i * post + b_j;
      int y_offset = b_i * post + b_j;
      if (dx) {
        dx[x_offset] =
            dx_op(x[x_offset], y[y_offset], out[x_offset], dout[x_offset]);
      }
      if (dy) {
        val += dy_op(x[x_offset], y[y_offset], out[x_offset], dout[x_offset]);
      }
    }
    if (dy) {
      int h = n > ELEMWISE_MAX_BLOCK_DIM ? ELEMWISE_MAX_BLOCK_DIM : n;
      val = paddle::platform::reduceSum(val, tid, h);
      if (tid == 0) {
        dy[bid] = val;
      }
    }
  } else {
    for (int i = tid; i < n; i += ELEMWISE_MAX_BLOCK_DIM) {
      int b_i = bid / post;
      int b_j = bid % post;
      int y_offset = b_i * n * post + i * post + b_j;
      int x_offset = b_i * post + b_j;
      if (dy) {
        dy[y_offset] =
            dy_op(x[x_offset], y[y_offset], out[y_offset], dout[y_offset]);
      }
      if (dx) {
        val += dx_op(x[x_offset], y[y_offset], out[y_offset], dout[y_offset]);
      }
    }
    if (dx) {
      int h = n > ELEMWISE_MAX_BLOCK_DIM ? ELEMWISE_MAX_BLOCK_DIM : n;
      val = paddle::platform::reduceSum(val, tid, h);
      if (tid == 0) {
        dx[bid] = val;
      }
    }
  }
}

template <typename T, typename DY_OP, typename Tout = T>
static __global__ void FastCommonGradBroadcastCUDAKernelHeight(const T *x,
                                                               const T *y,
                                                               const Tout *out,
                                                               const Tout *dout,
                                                               int h,
                                                               int w,
                                                               DY_OP dy_op,
                                                               T *dy,
                                                               int x_h,
                                                               int x_w,
                                                               bool is_y) {
  __shared__ T sdata[BLOCK_Y][BLOCK_X + 1];

  T val(0);
615 616
  size_t width_stride = GRID_NUM_X * BLOCK_NUM_X;
  size_t idx = THREAD_ID_X + BLOCK_NUM_X * BLOCK_ID_X;
617 618 619 620 621 622
  size_t full_width =
      (w & (~((uint64_t)(BLOCK_X - 1)))) + ((w & (BLOCK_X - 1)) ? BLOCK_X : 0);
  size_t full_height =
      (h & (~((uint64_t)(BLOCK_Y - 1)))) + ((h & (BLOCK_Y - 1)) ? BLOCK_Y : 0);
  if (is_y) {
    for (int m = idx; m < full_width; m += width_stride) {
623 624
      sdata[THREAD_ID_Y][THREAD_ID_X] = 0;
      for (int n = THREAD_ID_Y; n < full_height; n += BLOCK_Y) {
625 626 627 628 629
        int out_offset = n * w + m;
        int x_offset = (n % x_h) * x_w + m % x_w;
        if (dy) {
          if (m < w && n < h) {
            T val = dy_op(x[x_offset], y[m], out[out_offset], dout[out_offset]);
630
            sdata[THREAD_ID_Y][THREAD_ID_X] += val;
631 632 633 634 635
          }
          __syncthreads();
        }
      }
      if (dy) {
636
        T my_val = sdata[THREAD_ID_X][THREAD_ID_Y];
637 638 639 640
        for (int i = warpSize >> 1; i > 0; i >>= 1) {
          my_val += paddle::platform::CudaShuffleXorSync(0xFFFFFFFF, my_val, i);
        }
        __syncthreads();
641 642
        if ((THREAD_ID_X == 0)) {
          sdata[0][THREAD_ID_Y] = my_val;
643 644
        }
        __syncthreads();
645 646
        if (THREAD_ID_Y == 0 && m < w) {
          dy[m] = sdata[0][THREAD_ID_X];
647 648 649 650 651
        }
      }
    }
  } else {
    for (int m = idx; m < full_width; m += width_stride) {
652 653
      sdata[THREAD_ID_Y][THREAD_ID_X] = 0;
      for (int n = THREAD_ID_Y; n < full_height; n += BLOCK_Y) {
654 655 656 657 658
        int out_offset = n * w + m;
        int y_offset = (n % x_h) * x_w + m % x_w;
        if (dy) {
          if (m < w && n < h) {
            T val = dy_op(x[m], y[y_offset], out[out_offset], dout[out_offset]);
659
            sdata[THREAD_ID_Y][THREAD_ID_X] += val;
660 661 662 663 664
          }
          __syncthreads();
        }
      }
      if (dy) {
665
        T my_val = sdata[THREAD_ID_X][THREAD_ID_Y];
666 667 668 669
        for (int i = warpSize >> 1; i > 0; i >>= 1) {
          my_val += paddle::platform::CudaShuffleXorSync(0xFFFFFFFF, my_val, i);
        }
        __syncthreads();
670 671
        if ((THREAD_ID_X == 0)) {
          sdata[0][THREAD_ID_Y] = my_val;
672 673
        }
        __syncthreads();
674 675
        if (THREAD_ID_Y == 0 && m < w) {
          dy[m] = sdata[0][THREAD_ID_X];
676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693
        }
      }
    }
  }
}

template <typename T, typename DY_OP, typename Tout = T>
static __global__ void CommonGradBroadcast1CUDAKernelHeight(const T *x,
                                                            const T *y,
                                                            const Tout *out,
                                                            const Tout *dout,
                                                            int h,
                                                            int w,
                                                            DY_OP dy_op,
                                                            T *dy,
                                                            int x_h,
                                                            int x_w,
                                                            bool is_y) {
694 695 696
  int j = BLOCK_ID_X;
  int i = THREAD_ID_X;
  int tid = THREAD_ID_X;
697 698 699 700 701 702 703 704 705 706 707 708 709 710 711
  T val(0);

  if (is_y) {
    do {
      int out_offset = i * w + j;
      int x_offset = (i % x_h) * x_w + j % x_w;
      if (dy) {
        val += dy_op(x[x_offset], y[j], out[out_offset], dout[out_offset]);
      }
      i += ELEMWISE_MAX_BLOCK_DIM;
    } while (i < h);

    if (dy) {
      h = h > ELEMWISE_MAX_BLOCK_DIM ? ELEMWISE_MAX_BLOCK_DIM : h;
      val = paddle::platform::reduceSum(val, tid, h);
712
      if (THREAD_ID_X == 0) {
713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728
        dy[j] = val;
      }
    }
  } else {
    do {
      int out_offset = i * w + j;
      int y_offset = (i % x_h) * x_w + j % x_w;
      if (dy) {
        val += dy_op(x[j], y[y_offset], out[out_offset], dout[out_offset]);
      }
      i += ELEMWISE_MAX_BLOCK_DIM;
    } while (i < h);

    if (dy) {
      h = h > ELEMWISE_MAX_BLOCK_DIM ? ELEMWISE_MAX_BLOCK_DIM : h;
      val = paddle::platform::reduceSum(val, tid, h);
729
      if (THREAD_ID_X == 0) {
730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747
        dy[j] = val;
      }
    }
  }
}

template <typename T, typename DX_OP, typename DY_OP, typename Tout = T>
static __global__ void ElemwiseGradBroadcast1CUDAKernel(const T *x,
                                                        const T *y,
                                                        const Tout *out,
                                                        const Tout *dout,
                                                        int h,
                                                        int w,
                                                        bool is_xsize_larger,
                                                        DX_OP dx_op,
                                                        DY_OP dy_op,
                                                        T *dx,
                                                        T *dy) {
748 749 750
  int j = BLOCK_ID_X;
  int i = THREAD_ID_X;
  int tid = THREAD_ID_X;
751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766
  T val(0);
  if (is_xsize_larger) {
    do {
      int x_offset = i * w + j;
      if (dx) {
        dx[x_offset] = dx_op(x[x_offset], y[j], out[x_offset], dout[x_offset]);
      }
      if (dy) {
        val += dy_op(x[x_offset], y[j], out[x_offset], dout[x_offset]);
      }
      i += ELEMWISE_MAX_BLOCK_DIM;
    } while (i < h);

    if (dy) {
      h = h > ELEMWISE_MAX_BLOCK_DIM ? ELEMWISE_MAX_BLOCK_DIM : h;
      val = paddle::platform::reduceSum(val, tid, h);
767
      if (THREAD_ID_X == 0) {
768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785
        dy[j] = val;
      }
    }
  } else {  // x.dims < y.dims, broadcast for x.
    do {
      int y_offset = i * w + j;
      if (dy) {
        dy[y_offset] = dy_op(x[j], y[y_offset], out[y_offset], dout[y_offset]);
      }
      if (dx) {
        val += dx_op(x[j], y[y_offset], out[y_offset], dout[y_offset]);
      }
      i += ELEMWISE_MAX_BLOCK_DIM;
    } while (i < h);

    if (dx) {
      h = h > ELEMWISE_MAX_BLOCK_DIM ? ELEMWISE_MAX_BLOCK_DIM : h;
      val = paddle::platform::reduceSum(val, tid, h);
786
      if (THREAD_ID_X == 0) {
787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810
        dx[j] = val;
      }
    }
  }
}

// suppose use 2D block is fast because more parallel
// and memory coalesced
template <typename T, typename DX_OP, typename DY_OP, typename Tout = T>
static __global__ void FastElemwiseGradBroadcast1CUDAKernel(
    const T *x,
    const T *y,
    const Tout *out,
    const Tout *dout,
    int h,
    int w,
    bool is_xsize_larger,
    DX_OP dx_op,
    DY_OP dy_op,
    T *dx,
    T *dy) {
  __shared__ T sdata[BLOCK_Y][BLOCK_X + 1];

  T val(0);
811 812
  size_t width_stride = GRID_NUM_X * BLOCK_NUM_X;
  size_t idx = THREAD_ID_X + BLOCK_NUM_X * BLOCK_ID_X;
813 814 815 816 817 818
  size_t full_width =
      (w & (~((uint64_t)(BLOCK_X - 1)))) + ((w & (BLOCK_X - 1)) ? BLOCK_X : 0);
  size_t full_height =
      (h & (~((uint64_t)(BLOCK_Y - 1)))) + ((h & (BLOCK_Y - 1)) ? BLOCK_Y : 0);
  if (is_xsize_larger) {
    for (int m = idx; m < full_width; m += width_stride) {
819 820
      sdata[THREAD_ID_Y][THREAD_ID_X] = 0;
      for (int n = THREAD_ID_Y; n < full_height; n += BLOCK_Y) {
821 822 823 824 825 826 827 828
        int x_offset = n * w + m;
        if (dx && m < w && n < h) {
          dx[x_offset] =
              dx_op(x[x_offset], y[m], out[x_offset], dout[x_offset]);
        }
        if (dy) {
          if (m < w && n < h) {
            T val = dy_op(x[x_offset], y[m], out[x_offset], dout[x_offset]);
829
            sdata[THREAD_ID_Y][THREAD_ID_X] += val;
830 831 832 833 834
          }
          __syncthreads();
        }
      }
      if (dy) {
835
        T my_val = sdata[THREAD_ID_X][THREAD_ID_Y];
836 837 838
        for (int i = warpSize >> 1; i > 0; i >>= 1)
          my_val += paddle::platform::CudaShuffleXorSync(0xFFFFFFFF, my_val, i);
        __syncthreads();
839 840
        if ((THREAD_ID_X == 0)) {
          sdata[0][THREAD_ID_Y] = my_val;
841 842
        }
        __syncthreads();
843 844
        if (THREAD_ID_Y == 0 && m < w) {
          dy[m] = sdata[0][THREAD_ID_X];
845 846 847 848 849
        }
      }
    }
  } else {  // x.dims < y.dims, broadcast for x.
    for (int m = idx; m < full_width; m += width_stride) {
850 851
      sdata[THREAD_ID_Y][THREAD_ID_X] = 0;
      for (int n = THREAD_ID_Y; n < full_height; n += BLOCK_Y) {
852 853 854 855 856 857 858 859
        int y_offset = n * w + m;
        if (dy && m < w && n < h) {
          dy[y_offset] =
              dy_op(x[m], y[y_offset], out[y_offset], dout[y_offset]);
        }
        if (dx) {
          if (m < w && n < h) {
            T val = dx_op(x[m], y[y_offset], out[y_offset], dout[y_offset]);
860
            sdata[THREAD_ID_Y][THREAD_ID_X] += val;
861 862 863 864 865
          }
          __syncthreads();
        }
      }
      if (dx) {
866
        T my_val = sdata[THREAD_ID_X][THREAD_ID_Y];
867 868 869
        for (int i = warpSize >> 1; i > 0; i >>= 1)
          my_val += paddle::platform::CudaShuffleXorSync(0xFFFFFFFF, my_val, i);
        __syncthreads();
870 871
        if ((THREAD_ID_X == 0)) {
          sdata[0][THREAD_ID_Y] = my_val;
872 873
        }
        __syncthreads();
874 875
        if (THREAD_ID_Y == 0 && m < w) {
          dx[m] = sdata[0][THREAD_ID_X];
876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894
        }
      }
    }
  }
}

template <typename T, typename DX_OP, typename DY_OP, typename Tout = T>
static __global__ void ElemwiseGradBroadcast2CUDAKernel(const T *x,
                                                        const T *y,
                                                        const Tout *out,
                                                        const Tout *dout,
                                                        int pre,
                                                        int n,
                                                        int post,
                                                        bool is_xsize_larger,
                                                        DX_OP dx_op,
                                                        DY_OP dy_op,
                                                        T *dx,
                                                        T *dy) {
895 896
  int tid = THREAD_ID_X;
  int j = BLOCK_ID_X;
897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923

  T val(0);
  int ttid = tid;

  if (is_xsize_larger) {
    while (true) {
      int i = ttid / post;
      int k = ttid % post;
      if (i >= pre) break;

      int x_offset = i * n * post + j * post + k;

      if (dx != nullptr) {
        dx[x_offset] = dx_op(x[x_offset], y[j], out[x_offset], dout[x_offset]);
      }

      if (dy != nullptr) {
        val += dy_op(x[x_offset], y[j], out[x_offset], dout[x_offset]);
      }

      ttid += ELEMWISE_MAX_BLOCK_DIM;
    }

    if (dy) {
      int h = pre * post;
      h = h > ELEMWISE_MAX_BLOCK_DIM ? ELEMWISE_MAX_BLOCK_DIM : h;
      val = paddle::platform::reduceSum(val, tid, h);
924
      if (THREAD_ID_X == 0) {
925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950
        dy[j] = val;
      }
    }
  } else {  // x.dims < y.dims, broadcast for x.
    while (true) {
      int i = ttid / post;
      int k = ttid % post;
      if (i >= pre) break;

      int y_offset = i * n * post + j * post + k;

      if (dy != nullptr) {
        dy[y_offset] = dy_op(x[j], y[y_offset], out[y_offset], dout[y_offset]);
      }

      if (dx != nullptr) {
        val += dx_op(x[j], y[y_offset], out[y_offset], dout[y_offset]);
      }

      ttid += ELEMWISE_MAX_BLOCK_DIM;
    }

    if (dx) {
      int h = pre * post;
      h = h > ELEMWISE_MAX_BLOCK_DIM ? ELEMWISE_MAX_BLOCK_DIM : h;
      val = paddle::platform::reduceSum(val, tid, h);
951
      if (THREAD_ID_X == 0) {
952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974
        dx[j] = val;
      }
    }
  }
}

template <typename T, typename DX_OP, typename DY_OP, typename Tout = T>
static void ElemwiseGradBroadcast1CUDA(gpuStream_t stream,
                                       const T *x,
                                       const T *y,
                                       const Tout *out,
                                       const Tout *dout,
                                       int h,
                                       int w,
                                       bool is_xsize_larger,
                                       DX_OP dx_op,
                                       DY_OP dy_op,
                                       T *dx,
                                       T *dy) {
  // For small case use 1D block
  constexpr int half_walf = 16;
  if (w < half_walf || h < half_walf) {
    int block_size = std::min(ELEMWISE_MAX_BLOCK_DIM, h);
975 976
    int grid_size = w;
    ElemwiseGradBroadcast1CUDAKernel<<<grid_size, block_size, 0, stream>>>(
977 978 979 980
        x, y, out, dout, h, w, is_xsize_larger, dx_op, dy_op, dx, dy);
  } else {
    // suppose perfoemance improves with h increased.
    dim3 block_size = dim3(BLOCK_X, BLOCK_Y);
981
    dim3 grid_size = dim3((w + BLOCK_X - 1) / BLOCK_X);
982
    auto gplace = phi::GPUPlace(phi::backends::gpu::GetCurrentDeviceId());
983 984
    auto *ctx = static_cast<GPUContext *>(
        paddle::platform::DeviceContextPool::Instance().Get(gplace));
985
    phi::backends::gpu::LimitGridDim(*ctx, &grid_size);
986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005
    FastElemwiseGradBroadcast1CUDAKernel<<<grid_size, block_size, 0, stream>>>(
        x, y, out, dout, h, w, is_xsize_larger, dx_op, dy_op, dx, dy);
  }
}

template <typename T, typename DX_OP, typename DY_OP, typename Tout = T>
static void ElemwiseGradBroadcast2CUDA(gpuStream_t stream,
                                       const T *x,
                                       const T *y,
                                       const Tout *out,
                                       const Tout *dout,
                                       int pre,
                                       int n,
                                       int post,
                                       bool is_xsize_larger,
                                       DX_OP dx_op,
                                       DY_OP dy_op,
                                       T *dx,
                                       T *dy) {
  int block_size = std::min(ELEMWISE_MAX_BLOCK_DIM, pre * post);
1006
  dim3 grid_size = dim3(n);
1007
  auto gplace = phi::GPUPlace(phi::backends::gpu::GetCurrentDeviceId());
1008 1009
  auto *ctx = static_cast<GPUContext *>(
      paddle::platform::DeviceContextPool::Instance().Get(gplace));
1010
  phi::backends::gpu::LimitGridDim(*ctx, &grid_size);
1011
  ElemwiseGradBroadcast2CUDAKernel<<<grid_size, block_size, 0, stream>>>(
1012 1013 1014
      x, y, out, dout, pre, n, post, is_xsize_larger, dx_op, dy_op, dx, dy);
}

1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030
template <typename T, typename DX_OP, typename Tout = T>
__global__ void CommonGradBroadcastCUDAKernel(const int *x_strides_array,
                                              const int *y_strides_array,
                                              const int *out_dims_array,
                                              const int *y_strides_order,
                                              const int *y_dims_order,
                                              const T *x,
                                              const T *y,
                                              const Tout *out,
                                              const Tout *dout,
                                              T *dx,
                                              int out_size,
                                              int max_dim,
                                              int thread_num,
                                              DX_OP dx_op) {
  T val(0);
1031 1032 1033
  int i = BLOCK_ID_X;
  int tid = THREAD_ID_X;
  for (int j = tid; j < thread_num; j += BLOCK_NUM_X) {
1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056
    const int X_index = i * thread_num + j;
    int out_index = X_index;
    int C_index = 0;
    int B_index = i * thread_num + j;
    int remainder = 0;
#pragma unroll
    for (int d = max_dim - 1; d >= 0; --d) {
      GetDivMod(B_index, y_dims_order[d], &B_index, &remainder);
      C_index += remainder * y_strides_order[d];
    }
    int x_index = 0;
    int y_index = 0;
    int C_index_val = C_index;
#pragma unroll
    for (int d = max_dim - 1; d >= 0; --d) {
      GetDivMod(C_index_val, out_dims_array[d], &C_index_val, &remainder);
      x_index += remainder * x_strides_array[d];
      y_index += remainder * y_strides_array[d];
    }
    out_index = C_index;
    val += dx_op(x[x_index], y[y_index], out[out_index], dout[out_index]);
  }
  val = paddle::platform::reduceSum(val, tid, thread_num);
1057
  if (THREAD_ID_X == 0) {
1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075
    dx[i] = val;
  }
}

template <typename T, typename DX_OP, typename DY_OP, typename Tout = T>
void CommonGradBroadcastCUDA(const DenseTensor &x,
                             const DenseTensor &y,
                             const DenseTensor &out,
                             const DenseTensor &dout,
                             DenseTensor *dx,
                             DenseTensor *dy,
                             int *x_dims_array,
                             int *y_dims_array,
                             int *out_dims_array,
                             int max_dim,
                             const GPUContext &ctx,
                             DX_OP dx_op,
                             DY_OP dy_op) {
1076
  const auto gplace = ctx.GetPlace();
1077
  auto cplace = phi::CPUPlace();
1078 1079 1080 1081
  const T *x_data = x.data<T>();
  const T *y_data = y.data<T>();
  const Tout *out_data = out.data<Tout>();
  const Tout *dout_data = dout.data<Tout>();
1082 1083
  T *dx_data = dx == nullptr ? nullptr : ctx.Alloc<T>(dx);
  T *dy_data = dy == nullptr ? nullptr : ctx.Alloc<T>(dy);
1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211

  std::vector<int> x_one_indexs;
  std::vector<int> y_one_indexs;
  for (int i = 0; i < max_dim; i++) {
    if (x_dims_array[i] != y_dims_array[i]) {
      if (x_dims_array[i] == 1) {
        x_one_indexs.push_back(i);
      }
      if (y_dims_array[i] == 1) {
        y_one_indexs.push_back(i);
      }
    }
  }

  std::vector<int> x_trans_indexs(max_dim);
  std::vector<int> y_trans_indexs(max_dim);
  ComputeBroadcastTranspositionArray(
      x_one_indexs.data(), x_trans_indexs.data(), max_dim, x_one_indexs.size());
  ComputeBroadcastTranspositionArray(
      y_one_indexs.data(), y_trans_indexs.data(), max_dim, y_one_indexs.size());

  // compute array stride for cuda kernel;
  // e.g. x.dims=[2,3,4], x_stride=[12,4,1]
  std::vector<int> x_strides_array(max_dim);
  std::vector<int> y_strides_array(max_dim);
  std::vector<int> out_strides_array(max_dim);
  int x_stride = 1;
  int y_stride = 1;
  int z_stride = 1;
  for (int i = max_dim - 1; i >= 0; i--) {
    x_strides_array[i] = x_dims_array[i] == 1 ? 0 : x_stride;
    y_strides_array[i] = y_dims_array[i] == 1 ? 0 : y_stride;
    out_strides_array[i] = z_stride;
    x_stride *= x_dims_array[i];
    y_stride *= y_dims_array[i];
    z_stride *= out_dims_array[i];
  }

  std::vector<int> x_strides_order(max_dim);
  std::vector<int> y_strides_order(max_dim);
  std::vector<int> x_dims_order(max_dim);
  std::vector<int> y_dims_order(max_dim);
  for (int i = 0; i < max_dim; ++i) {
    x_strides_order[i] = out_strides_array[x_trans_indexs[i]];
    y_strides_order[i] = out_strides_array[y_trans_indexs[i]];
    x_dims_order[i] = out_dims_array[x_trans_indexs[i]];
    y_dims_order[i] = out_dims_array[y_trans_indexs[i]];
  }
  std::vector<int> x_broadcast_pos;
  std::vector<int> y_broadcast_pos;

  int bytes = max_dim * sizeof(int);

  for (int i = 0; i < max_dim; ++i) {
    if (x_dims_array[i] != out_dims_array[i] && x_dims_array[i] == 1) {
      x_broadcast_pos.emplace_back(i);
    }
    if (y_dims_array[i] != out_dims_array[i] && y_dims_array[i] == 1) {
      y_broadcast_pos.emplace_back(i);
    }
  }

  auto stream = ctx.stream();
  bool can_split_x = false;
  bool can_split_y = false;

  auto FastCommonCUDAF = [&](const std::vector<int> &broadcast_pos, bool is_y) {
    int h = std::accumulate(out_dims_array,
                            out_dims_array + broadcast_pos.size(),
                            1,
                            std::multiplies<int>());
    int w = std::accumulate(out_dims_array + broadcast_pos.size(),
                            out_dims_array + max_dim,
                            1,
                            std::multiplies<int>());

    VLOG(3) << "FastCommonCUDAF elementwise w:" << w << " h:" << h
            << " is_y:" << is_y;

    int split_h;
    int split_w;
    int kh = h;
    int kw = w;

    if (is_y) {
      split_h = std::accumulate(x_dims_array,
                                x_dims_array + broadcast_pos.size(),
                                1,
                                std::multiplies<int>());
      split_w = std::accumulate(x_dims_array + broadcast_pos.size(),
                                x_dims_array + max_dim,
                                1,
                                std::multiplies<int>());

    } else {
      split_h = std::accumulate(y_dims_array,
                                y_dims_array + broadcast_pos.size(),
                                1,
                                std::multiplies<int>());
      split_w = std::accumulate(y_dims_array + broadcast_pos.size(),
                                y_dims_array + max_dim,
                                1,
                                std::multiplies<int>());
    }

    if (h > split_h) kh = split_h;
    if (w > split_w) kw = split_w;

    if (is_y) {
      if (w < 16 || h < 16) {
        int block_size = std::min(ELEMWISE_MAX_BLOCK_DIM, h);
        int grid_size = w;
        CommonGradBroadcast1CUDAKernelHeight<<<grid_size,
                                               block_size,
                                               0,
                                               stream>>>(x_data,
                                                         y_data,
                                                         out_data,
                                                         dout_data,
                                                         h,
                                                         w,
                                                         dy_op,
                                                         dy_data,
                                                         kh,
                                                         kw,
                                                         is_y);
      } else {
        dim3 block_size = dim3(BLOCK_X, BLOCK_Y);
1212
        dim3 grid_size = dim3((w + BLOCK_X - 1) / BLOCK_X);
1213
        phi::backends::gpu::LimitGridDim(ctx, &grid_size);
1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248
        FastCommonGradBroadcastCUDAKernelHeight<<<grid_size,
                                                  block_size,
                                                  0,
                                                  stream>>>(x_data,
                                                            y_data,
                                                            out_data,
                                                            dout_data,
                                                            h,
                                                            w,
                                                            dy_op,
                                                            dy_data,
                                                            kh,
                                                            kw,
                                                            is_y);
      }
    } else {
      if (w < 16 || h < 16) {
        int block_size = std::min(ELEMWISE_MAX_BLOCK_DIM, h);
        int grid_size = w;
        CommonGradBroadcast1CUDAKernelHeight<<<grid_size,
                                               block_size,
                                               0,
                                               stream>>>(x_data,
                                                         y_data,
                                                         out_data,
                                                         dout_data,
                                                         h,
                                                         w,
                                                         dx_op,
                                                         dx_data,
                                                         kh,
                                                         kw,
                                                         is_y);
      } else {
        dim3 block_size = dim3(BLOCK_X, BLOCK_Y);
1249
        dim3 grid_size = dim3((w + BLOCK_X - 1) / BLOCK_X);
1250
        phi::backends::gpu::LimitGridDim(ctx, &grid_size);
1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316
        FastCommonGradBroadcastCUDAKernelHeight<<<grid_size,
                                                  block_size,
                                                  0,
                                                  stream>>>(x_data,
                                                            y_data,
                                                            out_data,
                                                            dout_data,
                                                            h,
                                                            w,
                                                            dx_op,
                                                            dx_data,
                                                            kh,
                                                            kw,
                                                            is_y);
      }
    }
  };

  auto FastBroadCastHeightCUDAF = [&](const std::vector<int> &broadcast_pos,
                                      bool x_large) {
    int h = std::accumulate(out_dims_array,
                            out_dims_array + broadcast_pos.size(),
                            1,
                            std::multiplies<int>());
    int w = std::accumulate(out_dims_array + broadcast_pos.size(),
                            out_dims_array + max_dim,
                            1,
                            std::multiplies<int>());

    VLOG(3) << "FastBroadCastHeightCUDAF w:" << w << " h:" << h;

    if (w < 16 || h < 16) {
      int block_size = std::min(ELEMWISE_MAX_BLOCK_DIM, h);
      int grid_size = w;
      ElemwiseGradBroadcast1CUDAKernel<<<grid_size, block_size, 0, stream>>>(
          x_data,
          y_data,
          out_data,
          dout_data,
          h,
          w,
          x_large,
          dx_op,
          dy_op,
          dx_data,
          dy_data);
    } else {
      dim3 block_size = dim3(BLOCK_X, BLOCK_Y);
      int grid_size = (w + BLOCK_X - 1) / BLOCK_X;
      FastElemwiseGradBroadcast1CUDAKernel<<<grid_size,
                                             block_size,
                                             0,
                                             stream>>>(x_data,
                                                       y_data,
                                                       out_data,
                                                       dout_data,
                                                       h,
                                                       w,
                                                       x_large,
                                                       dx_op,
                                                       dy_op,
                                                       dx_data,
                                                       dy_data);
    }
  };

1317 1318 1319
  auto FastBroadCastAllCUDAF = [&](const std::vector<int> &broadcast_pos,
                                   int max_dim,
                                   bool is_x_large) {
1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346
    int axis = broadcast_pos[0];
    int pre = std::accumulate(
        out_dims_array, out_dims_array + axis, 1, std::multiplies<int>());
    int mid = 1;
    int post = 1;

    if (broadcast_pos.size() == 1) {
      mid = out_dims_array[axis];
      post = std::accumulate(out_dims_array + axis + 1,
                             out_dims_array + max_dim,
                             1,
                             std::multiplies<int>());
    } else {
      mid = std::accumulate(out_dims_array + axis,
                            out_dims_array + broadcast_pos.back() + 1,
                            1,
                            std::multiplies<int>());
      post = std::accumulate(out_dims_array + broadcast_pos.back() + 1,
                             out_dims_array + max_dim,
                             1,
                             std::multiplies<int>());
    }

    VLOG(3) << "FastBroadCastAllCUDAF pre:" << pre << " mid:" << mid
            << " post:" << post;

    int block_size = std::min(ELEMWISE_MAX_BLOCK_DIM, mid);
1347
    dim3 grid_size = dim3(pre * post);
1348
    phi::backends::gpu::LimitGridDim(ctx, &grid_size);
1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364

    FastCommonGradBroadcastAllCUDAKernel<<<grid_size, block_size, 0, stream>>>(
        x_data,
        y_data,
        out_data,
        dout_data,
        pre,
        mid,
        post,
        is_x_large,
        dx_op,
        dy_op,
        dx_data,
        dy_data);
  };

1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389
  auto FastBroadCastOneCUDAF =
      [&](const std::vector<int> &broadcast_pos, int max_dim, bool is_x) {
        int axis = broadcast_pos[0];
        int pre = std::accumulate(
            out_dims_array, out_dims_array + axis, 1, std::multiplies<int>());
        int mid = out_dims_array[axis];
        int post = std::accumulate(out_dims_array + axis + 1,
                                   out_dims_array + max_dim,
                                   1,
                                   std::multiplies<int>());

        int k_pre;
        int k_mid;
        int k_post;

        if (is_x) {
          k_pre = std::accumulate(
              y_dims_array, y_dims_array + axis, 1, std::multiplies<int>());
          k_mid = y_dims_array[axis];
          k_post = std::accumulate(y_dims_array + axis + 1,
                                   y_dims_array + max_dim,
                                   1,
                                   std::multiplies<int>());
          int block_size = std::min(ELEMWISE_MAX_BLOCK_DIM, mid);
          dim3 grid_size = dim3(pre * post);
1390
          phi::backends::gpu::LimitGridDim(ctx, &grid_size);
1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420
          // we need to calc y offset with blockid, so do x_pre/y_pre to get
          // left size.
          if (k_pre != pre) k_pre = pre / k_pre;

          FastCommonGradBroadcastOneCUDAKernel<<<grid_size,
                                                 block_size,
                                                 0,
                                                 stream>>>(x_data,
                                                           y_data,
                                                           out_data,
                                                           dout_data,
                                                           pre,
                                                           mid,
                                                           post,
                                                           k_pre,
                                                           k_mid,
                                                           k_post,
                                                           true,
                                                           dx_op,
                                                           dx_data);
        } else {
          k_pre = std::accumulate(
              x_dims_array, x_dims_array + axis, 1, std::multiplies<int>());
          k_mid = x_dims_array[axis];
          k_post = std::accumulate(x_dims_array + axis + 1,
                                   x_dims_array + max_dim,
                                   1,
                                   std::multiplies<int>());
          int block_size = std::min(ELEMWISE_MAX_BLOCK_DIM, mid);
          dim3 grid_size = dim3(pre * post);
1421
          phi::backends::gpu::LimitGridDim(ctx, &grid_size);
1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443
          if (k_pre != pre) k_pre = pre / k_pre;

          FastCommonGradBroadcastOneCUDAKernel<<<grid_size,
                                                 block_size,
                                                 0,
                                                 stream>>>(x_data,
                                                           y_data,
                                                           out_data,
                                                           dout_data,
                                                           pre,
                                                           mid,
                                                           post,
                                                           k_pre,
                                                           k_mid,
                                                           k_post,
                                                           false,
                                                           dy_op,
                                                           dy_data);
        }
        VLOG(3) << "FastBroadCastOneCUDAF pre:" << pre << " mid:" << mid
                << " post:" << post;
      };
1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526

  // do fast elementwise if: 1. only one input need to do broadcast, we can
  // fallback
  // to old fast path.
  // 2. if both x and y need broadcast, then do it one by one.
  bool fast_broadcast = false;
  if (x_broadcast_pos.empty() && !y_broadcast_pos.empty()) {
    can_split_y = SplitDims(y_broadcast_pos, max_dim);
    if (can_split_y) {
      // only y need to do broadcast on h
      if (y_broadcast_pos[0] == 0) {
        FastBroadCastHeightCUDAF(y_broadcast_pos, true);
        fast_broadcast = true;
      }
    } else if (y_broadcast_pos.size() == 1 ||
               CheckContiguousDims(y_broadcast_pos)) {  // for only one dim and
                                                        // contiguous broadcast.
      // If cannot split,  which means input has 3 parts
      FastBroadCastAllCUDAF(y_broadcast_pos, max_dim, true);
      fast_broadcast = true;
    }
  } else if (y_broadcast_pos.empty() && !x_broadcast_pos.empty()) {
    // only x need broadcast
    can_split_x = SplitDims(x_broadcast_pos, max_dim);
    if (can_split_x) {
      if (x_broadcast_pos[0] == 0) {
        FastBroadCastHeightCUDAF(x_broadcast_pos, false);
        fast_broadcast = true;
      }
    } else if (x_broadcast_pos.size() == 1 ||
               CheckContiguousDims(x_broadcast_pos)) {
      FastBroadCastAllCUDAF(x_broadcast_pos, max_dim, false);
      fast_broadcast = true;
    }
  } else if (!x_broadcast_pos.empty() && !y_broadcast_pos.empty()) {
    // do x and y broadcast each.
    can_split_y = SplitDims(y_broadcast_pos, max_dim);
    bool fast_broadcast_x = false;
    bool fast_broadcast_y = false;
    if (can_split_y) {
      // begin at start.
      if (y_broadcast_pos[0] == 0) {
        FastCommonCUDAF(y_broadcast_pos, true);
        fast_broadcast_y = true;
      }
    } else if (y_broadcast_pos.size() == 1) {
      FastBroadCastOneCUDAF(y_broadcast_pos, max_dim, false);
      can_split_y = true;
      fast_broadcast_y = true;
    }
    can_split_x = SplitDims(x_broadcast_pos, max_dim);
    if (can_split_x) {
      if (x_broadcast_pos[0] == 0) {
        FastCommonCUDAF(x_broadcast_pos, false);
        fast_broadcast_x = true;
      }
    } else if (x_broadcast_pos.size() == 1) {
      FastBroadCastOneCUDAF(x_broadcast_pos, max_dim, true);
      can_split_x = true;
      fast_broadcast_x = true;
    }
    VLOG(3) << "CommonBroadcast can_split_y:" << can_split_y
            << " can_split_x:" << can_split_x;
    // if both x and y into fast path then return
    if (fast_broadcast_x && fast_broadcast_y) {
      fast_broadcast = true;
    }
    if (can_split_y && can_split_x && fast_broadcast) return;
  }

  // Should remove memory copy, use reg instead.
  if (fast_broadcast) {
    return;
  }
  int x_blocks = 0;
  int x_threads = 0;
  ComputeBroadcastKernelSize(
      x_dims_array, out_dims_array, &x_blocks, &x_threads, max_dim);
  int y_blocks = 0;
  int y_threads = 0;
  ComputeBroadcastKernelSize(
      y_dims_array, out_dims_array, &y_blocks, &y_threads, max_dim);

1527 1528 1529 1530
  auto x_strides_array_tmp = paddle::memory::Alloc(
      ctx.GetPlace(),
      bytes,
      phi::Stream(reinterpret_cast<phi::StreamId>(ctx.stream())));
1531 1532 1533 1534 1535 1536 1537 1538 1539
  int *x_strides_array_gpu =
      reinterpret_cast<int *>(x_strides_array_tmp->ptr());
  paddle::memory::Copy(gplace,
                       x_strides_array_gpu,
                       cplace,
                       x_strides_array.data(),
                       bytes,
                       ctx.stream());

1540 1541 1542 1543
  auto y_strides_array_tmp = paddle::memory::Alloc(
      ctx.GetPlace(),
      bytes,
      phi::Stream(reinterpret_cast<phi::StreamId>(ctx.stream())));
1544 1545 1546 1547 1548 1549 1550 1551 1552
  int *y_strides_array_gpu =
      reinterpret_cast<int *>(y_strides_array_tmp->ptr());
  paddle::memory::Copy(gplace,
                       y_strides_array_gpu,
                       cplace,
                       y_strides_array.data(),
                       bytes,
                       ctx.stream());

1553 1554 1555 1556
  auto out_dims_array_tmp = paddle::memory::Alloc(
      ctx.GetPlace(),
      bytes,
      phi::Stream(reinterpret_cast<phi::StreamId>(ctx.stream())));
1557 1558 1559 1560 1561 1562 1563 1564 1565
  int *out_dims_array_gpu = reinterpret_cast<int *>(out_dims_array_tmp->ptr());
  paddle::memory::Copy(
      gplace, out_dims_array_gpu, cplace, out_dims_array, bytes, ctx.stream());

  const int out_size = std::accumulate(
      out_dims_array, out_dims_array + max_dim, 1, std::multiplies<int>());
  int x_block_size = std::min(ELEMWISE_MAX_BLOCK_DIM, x_threads);
  int y_block_size = std::min(ELEMWISE_MAX_BLOCK_DIM, y_threads);
  if (dx) {
1566 1567 1568 1569
    auto x_strides_order_tmp = paddle::memory::Alloc(
        ctx.GetPlace(),
        bytes,
        phi::Stream(reinterpret_cast<phi::StreamId>(ctx.stream())));
1570 1571 1572 1573 1574 1575 1576 1577 1578
    int *x_strides_order_gpu =
        reinterpret_cast<int *>(x_strides_order_tmp->ptr());
    paddle::memory::Copy(gplace,
                         x_strides_order_gpu,
                         cplace,
                         x_strides_order.data(),
                         bytes,
                         ctx.stream());

1579 1580 1581 1582
    auto x_dims_order_tmp = paddle::memory::Alloc(
        ctx.GetPlace(),
        bytes,
        phi::Stream(reinterpret_cast<phi::StreamId>(ctx.stream())));
1583 1584 1585 1586 1587 1588 1589
    int *x_dims_order_gpu = reinterpret_cast<int *>(x_dims_order_tmp->ptr());
    paddle::memory::Copy(gplace,
                         x_dims_order_gpu,
                         cplace,
                         x_dims_order.data(),
                         bytes,
                         ctx.stream());
1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604
    CommonGradBroadcastCUDAKernel<T, DX_OP, Tout>
        <<<x_blocks, x_block_size, 0, ctx.stream()>>>(x_strides_array_gpu,
                                                      y_strides_array_gpu,
                                                      out_dims_array_gpu,
                                                      x_strides_order_gpu,
                                                      x_dims_order_gpu,
                                                      x_data,
                                                      y_data,
                                                      out_data,
                                                      dout_data,
                                                      dx_data,
                                                      out_size,
                                                      max_dim,
                                                      x_threads,
                                                      dx_op);
1605 1606
  }
  if (dy) {
1607 1608 1609 1610
    auto y_strides_order_tmp = paddle::memory::Alloc(
        ctx.GetPlace(),
        bytes,
        phi::Stream(reinterpret_cast<phi::StreamId>(ctx.stream())));
1611 1612 1613 1614 1615 1616 1617 1618 1619
    int *y_strides_order_gpu =
        reinterpret_cast<int *>(y_strides_order_tmp->ptr());
    paddle::memory::Copy(gplace,
                         y_strides_order_gpu,
                         cplace,
                         y_strides_order.data(),
                         bytes,
                         ctx.stream());

1620 1621 1622 1623
    auto y_dims_order_tmp = paddle::memory::Alloc(
        ctx.GetPlace(),
        bytes,
        phi::Stream(reinterpret_cast<phi::StreamId>(ctx.stream())));
1624 1625 1626 1627 1628 1629 1630
    int *y_dims_order_gpu = reinterpret_cast<int *>(y_dims_order_tmp->ptr());
    paddle::memory::Copy(gplace,
                         y_dims_order_gpu,
                         cplace,
                         y_dims_order.data(),
                         bytes,
                         ctx.stream());
1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645
    CommonGradBroadcastCUDAKernel<T, DY_OP, Tout>
        <<<y_blocks, y_block_size, 0, ctx.stream()>>>(x_strides_array_gpu,
                                                      y_strides_array_gpu,
                                                      out_dims_array_gpu,
                                                      y_strides_order_gpu,
                                                      y_dims_order_gpu,
                                                      x_data,
                                                      y_data,
                                                      out_data,
                                                      dout_data,
                                                      dy_data,
                                                      out_size,
                                                      max_dim,
                                                      y_threads,
                                                      dy_op);
1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666
  }
}

template <typename T, typename DX_OP, typename DY_OP, typename Tout = T>
void CommonElementwiseBroadcastBackward(const GPUContext &ctx,
                                        const DDim &x_dims,
                                        const DDim &y_dims,
                                        const DenseTensor &x,
                                        const DenseTensor &y,
                                        const DenseTensor &out,
                                        const DenseTensor &dout,
                                        int axis,
                                        DenseTensor *dx,
                                        DenseTensor *dy,
                                        DX_OP dx_op,
                                        DY_OP dy_op) {
  int max_dim = std::max(x_dims.size(), y_dims.size());
  axis = (axis == -1 ? std::abs(x_dims.size() - y_dims.size()) : axis);
  std::vector<int> x_dims_array(max_dim);
  std::vector<int> y_dims_array(max_dim);
  std::vector<int> out_dims_array(max_dim);
1667 1668 1669 1670 1671 1672 1673
  GetBroadcastDimsArrays(x_dims,
                         y_dims,
                         x_dims_array.data(),
                         y_dims_array.data(),
                         out_dims_array.data(),
                         max_dim,
                         axis);
1674 1675 1676 1677 1678 1679 1680 1681
  // for inplace strategy. memset will make dx and dout clear and get wrong
  // result.
  if (dx && dx->IsSharedBufferWith(dout)) {
    dx->clear();
    dx->mutable_data<T>(x_dims, ctx.GetPlace());
  }

  VLOG(3) << "CommonElementwiseBroadcastBackward xdims:"
1682 1683
          << phi::make_ddim(x_dims_array)
          << " ydim:" << phi::make_ddim(y_dims_array);
1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724

  CommonGradBroadcastCUDA<T, DX_OP, DY_OP, Tout>(x,
                                                 y,
                                                 out,
                                                 dout,
                                                 dx,
                                                 dy,
                                                 x_dims_array.data(),
                                                 y_dims_array.data(),
                                                 out_dims_array.data(),
                                                 max_dim,
                                                 ctx,
                                                 dx_op,
                                                 dy_op);
}

template <typename T, typename DX_OP, typename DY_OP, typename Tout = T>
void ElemwiseGradComputeWithBroadcast(const GPUContext &ctx,
                                      const DDim &x_dims,
                                      const DDim &y_dims,
                                      const DenseTensor &x,
                                      const DenseTensor &y,
                                      const DenseTensor &out,
                                      const DenseTensor &dout,
                                      int axis,
                                      DenseTensor *dx,
                                      DenseTensor *dy,
                                      DX_OP dx_op,
                                      DY_OP dy_op) {
  bool is_xsize_larger = true;

  int max_dim = x_dims.size();
  if (x_dims.size() < y_dims.size()) {
    is_xsize_larger = false;
    max_dim = y_dims.size();
  }

  axis = (axis == -1 ? std::abs(x_dims.size() - y_dims.size()) : axis);
  PADDLE_ENFORCE_GE(
      axis,
      0,
1725
      errors::InvalidArgument(
1726 1727
          "Axis should be great than or equal to 0, but received axis is %d.",
          axis));
1728
  PADDLE_ENFORCE_LE(axis,
1729
                    max_dim,
1730
                    errors::InvalidArgument(
1731 1732 1733 1734 1735 1736
                        "Axis should be less than %d, but received axis is %d.",
                        max_dim,
                        axis));

  int pre, n, post, is_run_common_broadcast, axis_trim = 0;
  if (is_xsize_larger) {
1737
    auto y_dims_trimed = TrimTrailingSingularDims(y_dims);
1738
    axis_trim = (y_dims_trimed.size() == 0) ? x_dims.size() : axis;
1739 1740 1741 1742 1743 1744 1745
    GetMidDims(x_dims,
               y_dims_trimed,
               axis_trim,
               &pre,
               &n,
               &post,
               &is_run_common_broadcast);
1746
  } else {
1747
    auto x_dims_trimed = TrimTrailingSingularDims(x_dims);
1748
    axis_trim = (x_dims_trimed.size() == 0) ? y_dims.size() : axis;
1749 1750 1751 1752 1753 1754 1755
    GetMidDims(y_dims,
               x_dims_trimed,
               axis_trim,
               &pre,
               &n,
               &post,
               &is_run_common_broadcast);
1756 1757 1758 1759 1760 1761 1762 1763
  }
  // special case for common backward implementation.
  if (is_run_common_broadcast) {
    CommonElementwiseBroadcastBackward<T, DX_OP, DY_OP, Tout>(
        ctx, x_dims, y_dims, x, y, out, dout, axis, dx, dy, dx_op, dy_op);
    return;
  }
  if (post == 1) {
1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775
    ElemwiseGradBroadcast1CUDA(ctx.stream(),
                               x.data<T>(),
                               y.data<T>(),
                               out.data<Tout>(),
                               dout.data<Tout>(),
                               pre,
                               n,
                               is_xsize_larger,
                               dx_op,
                               dy_op,
                               dx == nullptr ? nullptr : ctx.Alloc<T>(dx),
                               dy == nullptr ? nullptr : ctx.Alloc<T>(dy));
1776
  } else {
1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789
    ElemwiseGradBroadcast2CUDA(ctx.stream(),
                               x.data<T>(),
                               y.data<T>(),
                               out.data<Tout>(),
                               dout.data<Tout>(),
                               pre,
                               n,
                               post,
                               is_xsize_larger,
                               dx_op,
                               dy_op,
                               dx == nullptr ? nullptr : ctx.Alloc<T>(dx),
                               dy == nullptr ? nullptr : ctx.Alloc<T>(dy));
1790 1791 1792
  }
}

1793 1794
#endif

1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820
template <typename DeviceContext,
          typename T,
          typename DX_OP,
          typename DY_OP,
          typename Tout = T>
void ElemwiseGradCompute(const DeviceContext &dev_ctx,
                         const DenseTensor &x,
                         const DenseTensor &y,
                         const DenseTensor &out,
                         const DenseTensor &dout,
                         int axis,
                         DenseTensor *dx,
                         DenseTensor *dy,
                         DX_OP dx_op,
                         DY_OP dy_op) {
  const DDim &x_dim = x.dims();
  const DDim &y_dim = y.dims();
  if (x.dims() == y.dims()) {
    ElemwiseGradComputeNoBroadcast<DeviceContext, T, DX_OP, DY_OP, Tout>(
        dev_ctx, x_dim, y_dim, x, y, out, dout, axis, dx, dy, dx_op, dy_op);
  } else {
    ElemwiseGradComputeWithBroadcast<T, DX_OP, DY_OP, Tout>(
        dev_ctx, x_dim, y_dim, x, y, out, dout, axis, dx, dy, dx_op, dy_op);
  }
}

1821
}  // namespace funcs
1822
}  // namespace phi