datamover_primitives_xpu2.h 42.2 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
// Copyright (c) 2021 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.

#pragma once
#include "xpu/kernel/cluster_header.h"
#include "xpu/kernel/debug.h"
#include "xpu/kernel/math.h"

20
namespace phi {
21
namespace kps {
22 23
namespace details {

24 25 26 27 28 29 30 31 32 33 34
static inline int RoundUpDiv(int n, int k) { return (n + k - 1) / k; }

static inline int GetXpuReadLens(int numel, int block_num, int grid_num) {
  const int buf_size = 256;
  int nthreads = block_num * grid_num;
  if (numel / nthreads == 1) {
    return numel / nthreads * 4;
  }
  int read_lens = std::min(buf_size, RoundUpDiv(numel, 32 * nthreads) * 32);
  return read_lens;
}
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
enum class OptType {    // Optimize type of calc after input shape compressed
  CanNotOptimize = -1,  // can not optimize, broadcast first
  N_1,                  // just like {1} op {100} or {100} op {1}
  MN_N,                 // just like {100} op {3, 100} or {3, 100} op {100}
  MN_M,                 // just like {3} op {3, 100} or {3, 100} op {3}
  MNK_1N1,              // just like {3} op {2, 3, 100} or {2, 3, 100} op {3}
  MNK_M1K,  // just like {2, 1, 100} op {2, 3, 100} or {2, 3, 100} op {2, 1,
            // 100}
};

// Rules to determine whether dimensions can be merged
// rule 0 - xshape[idx] == yshape[idx]
// rule 1 - xshape[idx] == 1 && yshape[idx] != 1
// rule 2 - xshape[idx] != 1 && yshape[idx] == 1
static int judge_case(int a, int b) {
  if (a == b) {
    return 0;
  } else if (a == 1 && b != 1) {
    return 1;
  } else if (a != 1 && b == 1) {
    return 2;
  }
  return -1;
}

static bool case_is_same(int case_front, int case_back) {
  if (case_front == case_back) {
    return true;
  } else {
    return false;
  }
}

69 70 71 72 73 74 75 76 77 78
template <typename T, int VecSize>
struct alignas(sizeof(T) * VecSize) VectorType {
  T val[VecSize];
};

/**
 * Configuration of broadcast. Calculate the input data index according to the
 * index of the output data. if input or output shape is [dim0, dim1] then dims
 * must be [dim1, dim0].
 */
79
#pragma pack(4)
80
struct BroadcastConfig {
81 82 83
  int strides_in[phi::DDim::kMaxRank];
  int strides_out[phi::DDim::kMaxRank];
  int in_dim[phi::DDim::kMaxRank];
84
  int dim_after_cmp[phi::DDim::kMaxRank];
85
  int y_dim_after_cmp[phi::DDim::kMaxRank];
86 87 88 89 90 91 92
  int dim_size_after_cmp = 0;
  int cmp_res = 0;
  OptType cmp_type = OptType::CanNotOptimize;
  int m = 1;
  int n = 1;
  int k = 1;
  int buf_len = 0;
93
  int kDims;
94 95 96 97
  HOSTDEVICE BroadcastConfig() {}

  HOSTDEVICE BroadcastConfig(const std::vector<int64_t>& out_dims,
                             const std::vector<int64_t>& in_dims,
98
                             const std::vector<int64_t>& y_in_dims,
99
                             int dim_size) {
100 101 102 103 104 105 106 107 108
    std::vector<int> strides_in_tmp;
    std::vector<int> strides_out_tmp;
    std::vector<int> dim_tmp;
    strides_in_tmp.resize(dim_size, 1);
    strides_out_tmp.resize(dim_size, 1);
    dim_tmp.resize(dim_size, 1);
    for (int i = 1; i < dim_size; i++) {
      strides_in_tmp[i] = strides_in_tmp[i - 1] * in_dims[i - 1];
      strides_out_tmp[i] = strides_out_tmp[i - 1] * out_dims[i - 1];
109 110
    }

111
    int numel_out = 1;
112 113
    for (int i = 0; i < dim_size; i++) {
      dim_tmp[i] = in_dims[i];
114
      numel_out = out_dims[i] * numel_out;
115
    }
116
    kDims = dim_size;
117 118 119
    memcpy(strides_in, strides_in_tmp.data(), kDims * sizeof(int));
    memcpy(strides_out, strides_out_tmp.data(), kDims * sizeof(int));
    memcpy(in_dim, dim_tmp.data(), kDims * sizeof(int));
120

121 122
    cmp_res = get_mnk_for_broadcast_ops(in_dims, y_in_dims);
    get_opt_type();
123 124 125 126 127 128 129 130 131 132
    buf_len = get_buf_len(numel_out);
    int numel_x = 1;
    int numel_y = 1;
    for (int i = 0; i < dim_size; i++) {
      numel_x = in_dims[i] * numel_x;
      numel_y = y_in_dims[i] * numel_y;
    }
    if (numel_out == numel_x && numel_out == numel_y) {
      buf_len = GetXpuReadLens(numel_out, 8, 64);
    }
133 134
  }

135
  int get_buf_len(int numel) {
136 137 138
    if (cmp_type == OptType::CanNotOptimize) {
      return 256;
    }
139 140 141
    if (cmp_type == OptType::N_1) {
      return kps::details::GetXpuReadLens(numel, 8, 64);
    }
142 143 144 145 146 147
    int max_buf_len = 512;
    int buf_len = m / 16 * 16;
    if (buf_len == 0) {
      buf_len = m;
    }
    return std::min(max_buf_len, buf_len);
148 149 150 151
  }

  __device__ inline int operator()(int index_output) const {
    int index_src = 0;
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

    switch (cmp_type) {
      int div, mod, tmp_index;
      case OptType::MNK_M1K:
        div = index_output / (m * n);
        mod = index_output % (m * n) % m;
        index_src = div * m + mod;
        break;
      case OptType::MNK_1N1:
        // index_src = index_output / m % n;
        index_src = index_output % (m * n) / m;
        break;
      case OptType::N_1:
        index_src = 0;
        break;
      case OptType::MN_N:
        index_src = index_output / m;
        break;
      case OptType::MN_M:
        index_src = index_output % m;
        break;
      case OptType::CanNotOptimize:
        for (int i = kDims - 1; i >= 0; --i) {
          tmp_index = (index_output / strides_out[i]);
          index_output = index_output - tmp_index * strides_out[i];
          index_src += (tmp_index % in_dim[i]) * strides_in[i];
        }
        break;
180 181
    }
    return index_src;
182
  }
183

184
  void get_opt_type() {
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
    if (dim_size_after_cmp == 1) {
      if (dim_after_cmp[0] == 1 && y_dim_after_cmp[0] != 1) {  // {1} op {n}
        n = y_dim_after_cmp[0];
        cmp_type = OptType::N_1;
      } else if (dim_after_cmp[0] != 1 &&
                 y_dim_after_cmp[0] == 1) {  // {n} op {1}
        n = dim_after_cmp[0];
        cmp_type = OptType::N_1;
      } else {
        cmp_type = OptType::CanNotOptimize;  // xshape == yshape
      }
    }
    if (dim_size_after_cmp == 2) {
      if (dim_after_cmp[0] == 1 && dim_after_cmp[1] != 1 &&
          y_dim_after_cmp[0] != 1 &&
          y_dim_after_cmp[1] != 1) {  // {n} op {m, n}
        m = y_dim_after_cmp[0];
        n = y_dim_after_cmp[1];
        cmp_type = OptType::MN_N;
      } else if (dim_after_cmp[0] != 1 && dim_after_cmp[1] == 1 &&
                 y_dim_after_cmp[0] != 1 &&
                 y_dim_after_cmp[1] != 1) {  // {m} op {m, n}
        m = y_dim_after_cmp[0];
        n = y_dim_after_cmp[1];
        cmp_type = OptType::MN_M;
      } else if (dim_after_cmp[0] != 1 && dim_after_cmp[1] != 1 &&
                 y_dim_after_cmp[0] == 1 &&
                 y_dim_after_cmp[1] != 1) {  // {m, n} op {n}
        m = dim_after_cmp[0];
        n = dim_after_cmp[1];
        cmp_type = OptType::MN_N;
      } else if (dim_after_cmp[0] != 1 && dim_after_cmp[1] != 1 &&
                 y_dim_after_cmp[0] != 1 &&
                 y_dim_after_cmp[1] == 1) {  // {m, n} op {m}
        m = dim_after_cmp[0];
        n = dim_after_cmp[1];
        cmp_type = OptType::MN_M;
      } else {
        cmp_type = OptType::CanNotOptimize;
      }
    }
    if (dim_size_after_cmp == 3) {
      if (dim_after_cmp[0] == 1 && dim_after_cmp[1] != 1 &&
          dim_after_cmp[2] == 1 && y_dim_after_cmp[0] != 1 &&
          y_dim_after_cmp[1] != 1 &&
          y_dim_after_cmp[2] != 1) {  // {1, n, 1} op {m, n, k}
        m = y_dim_after_cmp[0];
        n = y_dim_after_cmp[1];
        k = y_dim_after_cmp[2];
        cmp_type = OptType::MNK_1N1;
      } else if (dim_after_cmp[0] != 1 && dim_after_cmp[1] != 1 &&
                 dim_after_cmp[2] != 1 && y_dim_after_cmp[0] == 1 &&
                 y_dim_after_cmp[1] != 1 &&
                 y_dim_after_cmp[2] == 1) {  // {m, n, k} op {1, n, 1}
        m = dim_after_cmp[0];
        n = dim_after_cmp[1];
        k = dim_after_cmp[2];
        cmp_type = OptType::MNK_1N1;
      } else if (dim_after_cmp[0] != 1 && dim_after_cmp[1] == 1 &&
                 dim_after_cmp[2] != 1 && y_dim_after_cmp[0] != 1 &&
                 y_dim_after_cmp[1] != 1 &&
                 y_dim_after_cmp[2] != 1) {  // {m, 1, k} op {m, n, k}
        m = y_dim_after_cmp[0];
        n = y_dim_after_cmp[1];
        k = y_dim_after_cmp[2];
        cmp_type = OptType::MNK_M1K;
      } else if (dim_after_cmp[0] != 1 && dim_after_cmp[1] != 1 &&
                 dim_after_cmp[2] != 1 && y_dim_after_cmp[0] != 1 &&
                 y_dim_after_cmp[1] == 1 &&
                 y_dim_after_cmp[2] != 1) {  // {m, n, k} op {m, 1, k}
        m = dim_after_cmp[0];
        n = dim_after_cmp[1];
        k = dim_after_cmp[2];
        cmp_type = OptType::MNK_M1K;
      } else {
        cmp_type = OptType::CanNotOptimize;
      }
    }
  }

  int get_mnk_for_broadcast_ops(const std::vector<int64_t>& xshape,
                                const std::vector<int64_t>& yshape) {
    int idx = 0;
    int cmp_x = 0;
    int cmp_y = 0;
    bool is_same = false;
271

272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305
    std::vector<int64_t> xshape_after_remove_ones = xshape;
    std::vector<int64_t> yshape_after_remove_ones = yshape;
    // first step: remove excess ones
    std::vector<int64_t>::iterator x_iter = xshape_after_remove_ones.begin();
    std::vector<int64_t>::iterator y_iter = yshape_after_remove_ones.begin();
    for (; x_iter != xshape_after_remove_ones.end();) {
      if (*x_iter == 1 && *y_iter == 1) {
        x_iter = xshape_after_remove_ones.erase(x_iter);
        y_iter = yshape_after_remove_ones.erase(y_iter);
      } else {
        x_iter++;
        y_iter++;
      }
    }
    // second step: compress dims
    int after_cmp_idx = 0;
    for (int i = 0; i < 3; i++) {
      cmp_x = xshape_after_remove_ones[idx];
      cmp_y = yshape_after_remove_ones[idx];
      while ((idx + 1) < xshape_after_remove_ones.size()) {
        is_same = case_is_same(judge_case(xshape_after_remove_ones[idx],
                                          yshape_after_remove_ones[idx]),
                               judge_case(xshape_after_remove_ones[idx + 1],
                                          yshape_after_remove_ones[idx + 1]));
        if (is_same) {
          cmp_x = cmp_x * xshape_after_remove_ones[idx + 1];
          cmp_y = cmp_y * yshape_after_remove_ones[idx + 1];
          idx++;
        } else {
          break;
        }
      }
      idx = idx + 1;
      dim_after_cmp[after_cmp_idx] = cmp_x;
306
      y_dim_after_cmp[after_cmp_idx] = cmp_y;
307 308 309 310 311 312 313 314
      after_cmp_idx++;
      if (idx == xshape_after_remove_ones.size()) {
        dim_size_after_cmp = after_cmp_idx;
        return 0;
      }
    }
    return -1;  // can not compress dims
  }
315
};
316
#pragma pack()
317

318
template <typename T>
319
__device__ __forceinline__ void WriteData(T _global_ptr_* dst,
320 321 322
                                          T* src,
                                          int num) {
  if (num > 0) {
N
niuliling123 已提交
323
    mfence_local();
324 325 326 327 328
    LM2GM(src, dst, num * sizeof(T));
  }
}
#undef INT_BITS

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
}  // namespace details

/**
 * @brief Read 2D data from global memory to register according to Tx type, and
 * store it as Ty type into register.
 *
 * @template paraments
 * Tx: The type of data stored in the global memory.
 * Ty: The type of data that needs to be stored in registers.
 * NX: The number of data columns loaded by each thread.
 * NY: The number of data rows loaded by each thread.
 * BlockSize: Identifies the current device thread index method. For xpu,
 * core_id() is used as the index.
 * IsBoundary: Indicates whether to perform block access storage out-of-bounds
 * judgment. When the number of data processed by the block is less than
 * NX x NY x core_num(), boundary judgment is required to avoid memory access
 * crossing the boundary.
 *
 * @param:
 * dst: The register pointer of the thread, the size is NX * NY.
 * src: The data pointer of the current block.
 * size_nx: The maximum offset of the current block is size_nx elements in the
 * lowest dimension. The parameters are only calculated when isboundary = true.
 * size_ny: The maximum offset of the current block is size_ny elements in the
 * first dimension. The parameters are only calculated when isboundary = true.
 * stride_nx: Each read one element stride stride_nx elements in the last dim.
 * stride_ny: Each read one element stride stride_ny elements in the first dim.
 */
357 358 359 360 361
template <typename Tx,
          typename Ty,
          int NX,
          int NY,
          int BlockSize,
362
          bool IsBoundary = false>
363 364 365 366 367
__device__ __inline__ void ReadData(Ty* dst,
                                    const Tx _global_ptr_* src,
                                    int size_nx,
                                    int size_ny,
                                    int stride_nx,
368
                                    int stride_ny) {
369 370
  int thread_offset = core_id();
  int left_size_nx = size_nx - thread_offset;
371
  __local__ Tx in_temp[1];
372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390
  // Each branch is added for better performance
  if (NX == 1 && NY == 1) {  // for NX == 1 and NY == 1
    if (IsBoundary) {
      if (left_size_nx > 0) {
        GM2LM(src + thread_offset, in_temp, sizeof(Tx));
        dst[0] = static_cast<Ty>(in_temp[0]);
      }
    } else {
      GM2LM(src + thread_offset, in_temp, sizeof(Tx));
      dst[0] = static_cast<Ty>(in_temp[0]);
    }
  } else if (NX == 1) {  // for NX == 1 and NY != 1
#pragma unroll
    for (int idy = 0; idy < NY; ++idy) {
      if (IsBoundary) {
        if (idy * stride_ny >= size_ny) {
          break;
        }
      }
N
niuliling123 已提交
391
      mfence_local();
392 393 394 395 396 397 398 399 400 401 402
      GM2LM(src + thread_offset + idy * stride_ny, in_temp, sizeof(Tx));
      dst[idy] = static_cast<Ty>(in_temp[0]);
    }
  } else if (NY == 1) {  // for NY == 1 and NX != 1
#pragma unroll
    for (int idx = 0; idx < NX; ++idx) {
      if (IsBoundary) {
        if (idx * stride_nx >= left_size_nx) {
          break;
        }
      }
N
niuliling123 已提交
403
      mfence_local();
404 405 406 407 408 409 410 411 412 413 414 415 416 417
      GM2LM(src + thread_offset + idx * stride_nx, in_temp, sizeof(Tx));
      dst[idx] = static_cast<Ty>(in_temp[0]);
    }
  } else {  // for NX != 1 and NY != 1
#pragma unroll
    for (int idx = 0; idx < NX; ++idx) {
#pragma unroll
      for (int idy = 0; idy < NY; ++idy) {
        if (IsBoundary) {
          if (idy * stride_ny >= size_ny || idx * stride_nx >= left_size_nx) {
            break;
          }
        }
        int fix = thread_offset + idx * stride_nx + idy * stride_ny;
N
niuliling123 已提交
418
        mfence_local();
419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437
        GM2LM(src + fix, in_temp, sizeof(Tx));
        dst[idy * NX + idx] = static_cast<Ty>(in_temp[0]);
      }
    }
  }
}

/**
 * @brief Initialize register with init_data.
 *
 * @template paraments
 * T: Data type of register.
 * NX: Number of data to initialize.
 *
 * @param:
 * dst: The register pointer of the thread, the size is NX.
 * init_data: Initial value.
 */
template <typename T, int NX>
438
__device__ __inline__ void Init(T* dst, T init_data) {
439 440 441 442 443 444
#pragma unroll
  for (int i = 0; i < NX; i++) {
    dst[i] = init_data;
  }
}

445 446 447 448 449 450 451 452
template <typename T, int NX>
__device__ __inline__ void Init(T* dst, T init_data, int read_lens) {
#pragma unroll
  for (int i = 0; i < read_lens; i++) {
    dst[i] = init_data;
  }
}

453 454 455 456 457
/**
 * The difference from the above function is that
 * it supports different data types of inputs.
 */
template <typename T, typename ArgsT, int Index, int NX>
458 459
__device__ __forceinline__ void Init(ArgsT* dst, T init_data, int read_lens) {
  mfence();
460
#pragma unroll
461
  for (int i = 0; i < read_lens; i++) {
462 463 464 465
    std::get<Index>(dst[i]) = init_data;
  }
}

466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486
/**
 * @brief Read 1D data from global memory to register. When IsBoundary = true
 * and (NX % 4 == 0 or Nx % 2 == 0), vectorized load data will be used to
 * improve memory access efficiency.
 *
 * @template paraments
 * T: The type of data.
 * NX: Each thread load NX data from global memory continuously.
 * NY: Each thread need to load NY rows, only NY = 1 was supported.
 * BlockSize: Identifies the current device thread index method. For xpu,
 * core_id() is used as the index.
 * IsBoundary: Whether to make an out-of-bounds judgment on access to memory.
 * When the number of data processed by this block is less than
 * NX x NY x core_num(), boundary judgment is required to avoid memory access
 * crossing the boundary.
 *
 * @param:
 * dst: The register pointer of the thread, the size is NX * NY.
 * src: The data pointer of the current block.
 * size: The current block needs to load size data continuously.
 */
487
template <typename T, int NX, int NY, int BlockSize, bool IsBoundary>
488 489
__device__ __inline__ void ReadData(T* dst,
                                    const T _global_ptr_* src,
490
                                    int num) {
N
niuliling123 已提交
491
  mfence_local();
492 493 494 495 496
  int thread_offset = core_id() * NX;
  if (IsBoundary) {  // core_num() * NX > num
#pragma unroll
    for (int idx = 0; idx < NX; ++idx) {
      if (idx + thread_offset < num) {
N
niuliling123 已提交
497
        GM2LM(src + thread_offset + idx, dst + idx, sizeof(T));
498 499 500 501 502 503 504
      }
    }
  } else {  // core_num() * NX < num
    GM2LM(src + thread_offset, dst, NX * sizeof(T));
  }
}

505 506 507 508 509 510
template <typename T, int NX, int NY, int BlockSize, bool IsBoundary>
__device__ __inline__ void ReadData(T* dst,
                                    const T _global_ptr_* src,
                                    int num,
                                    int read_lens) {
  int thread_offset = core_id() * read_lens;
N
niuliling123 已提交
511
  mfence_local();
512 513 514 515
  if (IsBoundary) {  // core_num() * read_lens > num
#pragma unroll
    for (int idx = 0; idx < read_lens; ++idx) {
      if (idx + thread_offset < num) {
N
niuliling123 已提交
516
        GM2LM(src + thread_offset + idx, dst + idx, sizeof(T));
517 518 519 520 521 522 523
      }
    }
  } else {  // core_num() * read_lens < num
    GM2LM(src + thread_offset, dst, read_lens * sizeof(T));
  }
}

524 525 526
/**
 * @brief Read 1D data from global memory to register. The difference
 * from the above function is that it supports different data types of inputs.
527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544
 *
 * @template paraments
 * T: The type of data.
 * NX: Each thread load NX data from global memory continuously.
 * NY: Each thread need to load NY rows, only NY = 1 was supported.
 * ArgsT: The Type if dst, ArgsT can be std::tuple<T> or std::tuple<Args>
 * Index: The index of data stored in dst.
 * BlockSize: Identifies the current device thread index method. For xpu,
 * core_id() is used as the index.
 * IsBoundary: Whether to make an out-of-bounds judgment on access to memory.
 * When the number of data processed by this block is less than
 * NX x NY x blockDim.x, boundary judgment is required to avoid memory access
 * crossing the boundary.
 *
 * @param:
 * dst: The register pointer of the thread, the size is NX * NY.
 * src: The data pointer of the current block.
 * size: The current block needs to load size data continuously.
545 546 547 548 549 550 551
 */
template <typename T,
          int NX,
          int NY,
          int BlockSize,
          typename ArgsT,
          int Index,
552
          bool IsBoundary>
553
__device__ __forceinline__ void ReadData(ArgsT* dst,
554
                                         const T _global_ptr_* src,
555 556 557
                                         int num,
                                         int read_lens) {
  int thread_offset = core_id() * read_lens;
558 559
  __local__ T in_temp[1];
  __local__ T in_vec[NX];
560
  if (IsBoundary) {  // core_num() * read_lens > num
561
#pragma unroll
562
    for (int idx = 0; idx < read_lens; ++idx) {
563 564 565
      if (idx + thread_offset < num) {
        GM2LM(src + thread_offset + idx, in_temp, sizeof(T));
        std::get<Index>(dst[idx]) = in_temp[0];
566
        mfence();
567 568
      }
    }
569 570
  } else {  // core_num() * read_lens < num
    GM2LM(src + thread_offset, in_vec, read_lens * sizeof(T));
571
#pragma unroll
572
    for (int idx = 0; idx < read_lens; ++idx) {
573 574 575 576 577
      std::get<Index>(dst[idx]) = in_vec[idx];
    }
  }
}

578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601
/**
 * @brief Read 2D data from global memory to registers with broadcast form.
 *
 * @template paraments
 * T: The type of data stored in the global memory.
 * NX: The number of data columns loaded by each thread.
 * NY: The number of data rows loaded by each thread.
 * BlockSize: Identifies the current device thread index method. For xpu,
 * core_id() is used as the index.
 * IsBoundary: Indicates whether to perform block access storage out-of-bounds
 * judgment. When the number of data processed by the block is less than
 * NX x NY x core_num(), boundary judgment is required to avoid memory access
 * crossing the boundary.
 *
 * @param:
 * dst: The register pointer of the thread, the size is NX * NY.
 * src: Raw input data pointer of kernel.
 * block_offset: Data offset of this block, core_num() *  cluster_id() * NX;
 * config: Calculation configuration of broadcast. It is used to calculate the
 * coordinate mapping relationship between output data and input data.
 * total_num_output: Total number of original output.
 * stride_nx: Each read one element stride stride_nx elements in the last dim.
 * stride_ny: Each read one element stride stride_ny elements in the first dim.
 */
602
template <typename T, int NX, int NY, int BlockSize, bool IsBoundary = false>
603 604
__device__ __inline__ void ReadDataBc(T* dst,
                                      const T _global_ptr_* src,
605
                                      uint32_t block_offset,
606
                                      const details::BroadcastConfig& config,
607 608
                                      int total_num_output,
                                      int stride_nx,
609
                                      int stride_ny) {
610 611
  uint32_t thread_offset = block_offset + core_id();
  uint32_t index_src = 0;
N
niuliling123 已提交
612
  mfence_local();
613 614 615 616 617 618 619
#pragma unroll
  for (int ny = 0; ny < NY; ++ny) {
#pragma unroll
    for (uint32_t nx = 0; nx < NX; ++nx) {
      uint32_t index_output = thread_offset + ny * stride_ny + nx * stride_nx;
      index_src = 0;
      if (IsBoundary) {
620
        if (index_output >= (uint32_t)total_num_output) {
621 622 623
          break;
        }
      }
624
      index_src = config(index_output);
N
niuliling123 已提交
625
      GM2LM(src + index_src, dst + nx + ny * NX, sizeof(T));
626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660
    }
  }
}

/**
 * @brief Read 2D data from global memory to register with reduce form.
 *
 * @template paraments
 * T: The type of data.
 * NX: The number of data columns loaded by each thread.
 * NY: The number of data rows loaded by each thread.
 * BlockSize: Identifies the current device thread index method. For xpu,
 * core_id() is used as the index.
 * Rank: The shape size of out. eg in[1, 35], out[32, 35] then shape size is 2.
 * IsBoundary: Indicates whether to perform block access storage out-of-bounds
 * judgment. When the number of data processed by the block is less than
 * NX x NY x core_num(), boundary judgment is required to avoid memory access
 * crossing the boundary.
 *
 * @param:
 * dst: The register pointer of the thread, the size is NX * NY.
 * src: The input data pointer of this block.
 * block_offset: The data offset of this block, blockDim.x * cluster_id() * NX.
 * index_cal: Calculation configuration of Reduce. It is used to calculate the
 * coordinate mapping relationship between output data and input data.
 * size_nx: The current block needs to load size_nx columns of data, this
 * parameter will participate in the calculation when isboundary = true.
 * size_ny: The current block needs to load size_ny rows of data, this parameter
 * will participate in the calculation when isboundary = true.
 * will be used when IsBoundary = true.
 * stride_nx: Each read one element stride stride_nx columns.
 * stride_ny: Each read one element stride stride_ny raws.
 * reduce_last_dim: Used to indicate whether the dimension of reduce contains
 * the lowest dimension.
 */
661 662
template <typename Tx,
          typename Ty,
663 664 665 666 667
          int NX,
          int NY,
          int BlockSize,
          int Rank,
          typename IndexCal,
668
          typename Functor,
669
          bool IsBoundary = false>
670 671 672 673 674 675 676 677 678 679 680
__device__ __forceinline__ void ReadDataReduce(
    Ty* dst,
    const Tx _global_ptr_* __restrict__ src,
    int block_offset,
    const IndexCal& index_cal,
    int size_nx,
    int size_ny,
    int stride_nx,
    int stride_ny,
    Functor func,
    bool reduce_last_dim) {
681
  __local__ Tx in_temp[1];
682
  int thread_offset = 0;
683
  int left_idx = 0;
684
  if (reduce_last_dim) {
685 686
    thread_offset = core_id();
    left_idx = 0;
687
  } else {
688 689
    thread_offset = 0;
    left_idx = 0;
690 691 692 693 694 695
  }

  if (NX == 1) {
#pragma unroll
    for (int ny = 0; ny < NY; ++ny) {
      if (IsBoundary) {
696
        if (thread_offset >= size_ny) {
697 698 699
          break;
        }
      }
700
      uint32_t index_src = index_cal(thread_offset + block_offset);
N
niuliling123 已提交
701
      mfence_local();
702 703
      GM2LM(src + index_src, in_temp, sizeof(Tx));
      dst[ny] = static_cast<Ty>(func(in_temp[0]));
N
niuliling123 已提交
704

705 706 707 708 709 710 711 712
      thread_offset += stride_ny;
    }
  } else {
#pragma unroll
    for (int nx = 0; nx < NX; ++nx) {
#pragma unroll
      for (int ny = 0; ny < NY; ++ny) {
        if (IsBoundary) {
713 714
          if ((thread_offset >= size_ny) ||
              (left_idx + nx * stride_nx >= size_nx)) {
715 716 717
            break;
          }
        }
718
        uint32_t index_src = index_cal(thread_offset + block_offset);
N
niuliling123 已提交
719
        mfence_local();
720 721
        GM2LM(src + index_src, in_temp, sizeof(Tx));
        dst[nx + ny * NX] = static_cast<Ty>(func(in_temp[0]));
722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748
        thread_offset += stride_ny;
      }
    }
  }
}
/**
 * @brief Write 1D data from registers to global memory. When IsBoundary = true
 * and (NX % 4 == 0 or Nx % 2 == 0), the data will be vectorized to improve the
 * data loading efficiency
 *
 * @template paraments
 * T: The type of data.
 * NX: The number of data continuously writed by each thread.
 * NY: The number of data rows loaded by each thread, only NY = 1 was supported.
 * BlockSize: Identifies the current device thread index method. For xpu,
 * core_id() is used as the index.
 * IsBoundary: Indicates whether to perform block access storage out-of-bounds
 * judgment. When the number of data processed by the block is less than
 * NX x NY x core_num(), boundary judgment is required to avoid memory access
 * crossing the boundary.
 *
 * @param:
 * dst: The data pointer of the current block.
 * src: The register pointer, the size is NX * NY.
 * size: The current block needs to load size elements continuously.
 */

749 750 751 752 753 754
template <typename T, int NX, int NY, int BlockSize, bool IsBoundary>
__device__ void WriteData(T _global_ptr_* dst,
                          const T* src,
                          int num,
                          int read_lens) {
  int thread_offset = core_id() * read_lens;
N
niuliling123 已提交
755
  mfence_local();
756 757 758 759 760

  if (IsBoundary) {  // core_num() * read_lens > num
#pragma unroll
    for (int idx = 0; idx < read_lens; ++idx) {
      if (idx + thread_offset < num) {
N
niuliling123 已提交
761
        LM2GM(src + idx, dst + idx + thread_offset, sizeof(T));
762 763 764 765 766 767 768
      }
    }
  } else {  // core_num() * read_lens < num
    LM2GM(src, dst + thread_offset, read_lens * sizeof(T));
  }
}

769 770 771
template <typename T, int NX, int NY, int BlockSize, bool IsBoundary>
__device__ void WriteData(T _global_ptr_* dst, const T* src, int num) {
  int thread_offset = core_id() * NX;
N
niuliling123 已提交
772
  mfence_local();
773

774 775 776 777
  if (IsBoundary) {  // core_num() * NX > num
#pragma unroll
    for (int idx = 0; idx < NX; ++idx) {
      if (idx + thread_offset < num) {
N
niuliling123 已提交
778
        LM2GM(src + idx, dst + idx + thread_offset, sizeof(T));
779 780 781
      }
    }
  } else {  // core_num() * NX < num
N
niuliling123 已提交
782
    mfence_local();
783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812
    LM2GM(src, dst + thread_offset, NX * sizeof(T));
  }
}

/**
 * @brief Write 2D data from register to global memory according to Tx type, and
 * store it as Ty type.
 *
 * @template paraments
 * Tx: The type of data that needs to be stored in registers.
 * Ty: The type of data stored in the global memory.
 * NX: The number of data columns loaded by each thread.
 * NY: The number of data rows loaded by each thread.
 * BlockSize: Identifies the current device thread index method. For xpu,
 * core_id() is used as the index.
 * IsBoundary: Indicates whether to perform block access storage out-of-bounds
 * judgment. When the number of data processed by the block is less than
 * NX x NY x core_num(), boundary judgment is required to avoid memory access
 * crossing the boundary.
 *
 * @param:
 * dst: Data pointer of the current block.
 * src: The register pointer of the thread, the size is NX * NY.
 * size_nx: The current block needs to load size_nx columns of data, this
 * parameter will be used when IsBoundary = true.
 * size_ny: The current block needs to load size_ny rows of data. This parameter
 * will be used when IsBoundary = true.
 * stride_nx: Each read one element stride stride_nx elements in the last dim.
 * stride_ny: Each read one element stride stride_ny elements in the first dim.
 */
813 814 815 816 817
template <typename Tx,
          typename Ty,
          int NX,
          int NY,
          int BlockSize,
818
          bool IsBoundary = false>
819 820 821 822 823
__device__ __inline__ void WriteData(Ty _global_ptr_* dst,
                                     const Tx* src,
                                     int size_nx,
                                     int size_ny,
                                     int stride_nx,
824
                                     int stride_ny) {
825 826 827 828 829 830 831 832 833
  int thread_offset = core_id();
  int left_size_nx = size_nx - thread_offset;
  __local__ Ty in_temp[1];

  // Each branch is added for better performance
  if (NX == 1 && NY == 1) {
    if (IsBoundary) {
      if (left_size_nx > 0) {
        in_temp[0] = static_cast<Ty>(src[0]);
N
niuliling123 已提交
834
        mfence_local();
835
        LM2GM(in_temp, dst + thread_offset, sizeof(Ty));
836 837 838
      }
    } else {
      in_temp[0] = static_cast<Ty>(src[0]);
N
niuliling123 已提交
839
      mfence_local();
840
      LM2GM(in_temp, dst + thread_offset, sizeof(Ty));
841 842 843 844 845 846 847 848 849 850 851
    }
  } else if (NX == 1) {
#pragma unroll
    for (int idy = 0; idy < NY; ++idy) {
      if (IsBoundary) {
        if (idy * stride_ny >= size_ny) {
          break;
        }
      }

      in_temp[0] = static_cast<Ty>(src[idy]);
N
niuliling123 已提交
852
      mfence_local();
853
      LM2GM(in_temp, dst + thread_offset + idy * stride_ny, sizeof(Ty));
854 855 856 857 858 859 860 861 862 863 864
    }
  } else if (NY == 1) {  // for NY == 1 and NX != 1
#pragma unroll
    for (int idx = 0; idx < NX; ++idx) {
      if (IsBoundary) {
        if (idx * stride_nx >= left_size_nx) {
          break;
        }
      }

      in_temp[0] = static_cast<Ty>(src[idx]);
N
niuliling123 已提交
865
      mfence_local();
866
      LM2GM(in_temp, dst + thread_offset + idx * stride_nx, sizeof(Ty));
867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883
    }
  } else {  // for NX != 1 and NY != 1
#pragma unroll
    for (int idx = 0; idx < NX; ++idx) {
      if (IsBoundary) {
        if (idx * stride_nx >= left_size_nx) {
          break;
        }
      }
#pragma unroll
      for (int idy = 0; idy < NY; ++idy) {
        if (IsBoundary) {
          if (idy * stride_ny >= size_ny) {
            break;
          }
        }
        in_temp[0] = static_cast<Ty>(src[idx + idy * NX]);
N
niuliling123 已提交
884
        mfence_local();
885 886
        LM2GM(in_temp,
              dst + thread_offset + idx * stride_nx + idy * stride_ny,
887
              sizeof(Ty));
888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904
      }
    }
  }
}

/**
 * @brief Initialize register with init_data.
 *
 * @template paraments
 * T: Data type of register.
 * NX: Number of data to initialize.
 *
 * @param:
 * dst: The register pointer of the thread, the size is NX.
 * init_data: The register pointer of init data, the size is NX.
 */
template <typename T, int NX, bool IsBoundary = false>
905
__device__ __inline__ void Init(T* dst, T* init_data, int num) {
906 907 908 909 910 911 912 913 914 915 916
#pragma unroll
  for (int i = 0; i < NX; i++) {
    if (IsBoundary) {
      if (i >= num) {
        break;
      }
    }
    dst[i] = init_data[i];
  }
}

917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932
/**
 * @brief Read data from global memory to local memory with broadcast
 * {m, 1, k}-> {m, n, k} form.
 *
 * @template paraments
 * T: Data type of register.
 * Rank: The shape size of out. eg in[1, 35], out[32, 35] then shape size is 2.
 *
 * @param:
 * dst: The register pointer of the thread, the size is NX.
 * src: The original input data pointer of kernel.
 * thread_offset: The data offset of this thread.
 * config: Calculation configuration of broadcast. It is used to calculate the
 * coordinate mapping relationship between output data and input data.
 * read_lens: The number of data continuously loaded by each thread.
 */
933
template <typename T>
934 935 936 937
__device__ __inline__ void ReadDataBcM1kMnk(
    T* dst,
    const T _global_ptr_* src,
    int thread_offset,
938
    const details::BroadcastConfig& config,
939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979
    int read_lens) {
  int index_output = thread_offset;
  int index_base = config(index_output);
  int m = config.m;
  int n = config.n;

  int m_pos = index_base % m;
  if ((m - m_pos) < read_lens) {
    int last_col = m - m_pos;
    GM2LM(src + index_base, dst, last_col * sizeof(T));
    int n_pos = index_output % (m * n) / m;
    int next_part_index = 0;
    if (n_pos != config.n - 1) {
      next_part_index = index_base / m * m;
    } else {
      next_part_index = (index_base / m + 1) * m;
    }
    GM2LM(src + next_part_index,
          dst + last_col,
          (read_lens - last_col) * sizeof(T));
  } else {
    GM2LM(src + index_base, dst, read_lens * sizeof(T));
  }
}

/**
 * @brief Read data from global memory to local memory with broadcast
 * {m, 1}-> {m, n} form.
 *
 * @template paraments
 * T: Data type of register.
 * Rank: The shape size of out. eg in[1, 35], out[32, 35] then shape size is 2.
 *
 * @param:
 * dst: The register pointer of the thread, the size is NX.
 * src: The original input data pointer of kernel.
 * thread_offset: The data offset of this thread.
 * config: Calculation configuration of broadcast. It is used to calculate the
 * coordinate mapping relationship between output data and input data.
 * read_lens: The number of data continuously loaded by each thread.
 */
980
template <typename T>
981 982 983 984
__device__ __inline__ void ReadDataBcM1Mn(
    T* dst,
    const T _global_ptr_* src,
    int thread_offset,
985
    const details::BroadcastConfig& config,
986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016
    int read_lens) {
  int index_output = thread_offset;
  int index_base = config(index_output);
  int m = config.m;
  int n = config.n;

  int m_pos = index_base % m;
  if ((m - m_pos) < read_lens) {
    int last_col = m - m_pos;
    GM2LM(src + index_base, dst, last_col * sizeof(T));
    GM2LM(src, dst + last_col, (read_lens - last_col) * sizeof(T));
  } else {
    GM2LM(src + index_base, dst, read_lens * sizeof(T));
  }
}

/**
 * @brief Read data from global memory to local memory with broadcast
 * {1, n}-> {m, n} form.
 *
 * @template paraments
 * T: Data type of register.
 *
 * @param:
 * dst: The register pointer of the thread, the size is NX.
 * src: The original input data pointer of kernel.
 * thread_offset: The data offset of this thread.
 * config: Calculation configuration of broadcast. It is used to calculate the
 * coordinate mapping relationship between output data and input data.
 * read_lens: The number of data continuously loaded by each thread.
 */
1017
template <typename T>
1018 1019 1020 1021
__device__ __inline__ void ReadDataBc1NMn(
    T* dst,
    const T _global_ptr_* src,
    int thread_offset,
1022
    const details::BroadcastConfig& config,
1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036
    int read_lens) {
  int index_output = thread_offset;
  int index_base = config(index_output);
  int m = config.m;
  int n = config.n;
  T in_temp;

  int m_pos = index_output % m;
  if ((m - m_pos) < read_lens) {
    int last_col = m - m_pos;
    GM2LM(src + index_base, &in_temp, sizeof(T));
    for (int i = 0; i < last_col; i++) {
      dst[i] = in_temp;
    }
N
niuliling123 已提交
1037
    mfence_local();
1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064
    GM2LM(src + index_base + 1, &in_temp, sizeof(T));
    for (int i = 0; i < read_lens - last_col; i++) {
      dst[last_col + i] = in_temp;
    }
  } else {
    GM2LM(src + index_base, &in_temp, sizeof(T));
    for (int i = 0; i < read_lens; i++) {
      dst[i] = in_temp;
    }
  }
}

/**
 * @brief Read data from global memory to local memory with broadcast
 * {1, n, 1}-> {m, n, k} form.
 *
 * @template paraments
 * T: Data type of register.
 *
 * @param:
 * dst: The register pointer of the thread, the size is NX.
 * src: The original input data pointer of kernel.
 * thread_offset: The data offset of this thread.
 * config: Calculation configuration of broadcast. It is used to calculate the
 * coordinate mapping relationship between output data and input data.
 * read_lens: The number of data continuously loaded by each thread.
 */
1065
template <typename T>
1066 1067 1068 1069
__device__ __inline__ void ReadDataBc1N1Mnk(
    T* dst,
    const T _global_ptr_* src,
    int thread_offset,
1070
    const details::BroadcastConfig& config,
1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091
    int read_lens) {
  int index_output = thread_offset;
  int index_base = config(index_output);
  int m = config.m;
  int n = config.n;
  T in_temp;

  int m_pos = index_output % m;
  if ((m - m_pos) < read_lens) {
    int last_col = m - m_pos;
    GM2LM(src + index_base, &in_temp, sizeof(T));
    for (int i = 0; i < last_col; i++) {
      dst[i] = in_temp;
    }
    int n_pos = index_output % (m * n) / m;
    int next_part_index = 0;
    if (n_pos != n - 1) {
      next_part_index = n_pos + 1;
    } else {
      next_part_index = 0;
    }
N
niuliling123 已提交
1092
    mfence_local();
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
    GM2LM(src + next_part_index, &in_temp, sizeof(T));
    for (int i = 0; i < read_lens - last_col; i++) {
      dst[last_col + i] = in_temp;
    }
  } else {
    GM2LM(src + index_base, &in_temp, sizeof(T));
    for (int i = 0; i < read_lens; i++) {
      dst[i] = in_temp;
    }
  }
}

/**
 * @brief Read data from global memory to local memory with broadcast
 * {1}-> {n} form.
 *
 * @template paraments
 * T: Data type of register.
 *
 * @param:
 * dst: The register pointer of the thread, the size is NX.
 * src: The original input data pointer of kernel.
 * thread_offset: The data offset of this thread.
 * config: Calculation configuration of broadcast. It is used to calculate the
 * coordinate mapping relationship between output data and input data.
 * read_lens: The number of data continuously loaded by each thread.
 */
1120 1121 1122 1123 1124 1125
template <typename T>
__device__ __inline__ void ReadDataBc1N(T* dst,
                                        const T _global_ptr_* src,
                                        int thread_offset,
                                        const details::BroadcastConfig& config,
                                        int read_lens) {
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
  int index_output = thread_offset;
  int index_base = config(index_output);
  T in_temp;

  GM2LM(src + index_base, &in_temp, sizeof(T));
  for (int i = 0; i < read_lens; i++) {
    dst[i] = in_temp;
  }
}

/**
 * @brief Read data from global memory to local memory with broadcast
 * form which can not compress.
 *
 * @template paraments
 * T: Data type of register.
 * Rank: The shape size of out. eg in[1, 35], out[32, 35] then shape size is 2.
 *
 * @param:
 * dst: The register pointer of the thread, the size is NX.
 * src: The original input data pointer of kernel.
 * thread_offset: The data offset of this thread.
 * config: Calculation configuration of broadcast. It is used to calculate the
 * coordinate mapping relationship between output data and input data.
 * total_num_output: Total number of original output.
 * read_lens: The number of data continuously loaded by each thread.
 */
1153
template <typename T, bool IsBoundary = false>
1154 1155 1156 1157
__device__ __inline__ void ReadDataBcCanNotCmp(
    T* dst,
    const T _global_ptr_* src,
    int thread_offset,
1158
    const details::BroadcastConfig& config,
1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178
    int total_num_output,
    int read_lens) {
  int index_output = thread_offset;
  int index_base = config(index_output);
  T in_temp;
  int cache_size = 256;
  __local__ T src_temp[cache_size];
  GM2LM(src + index_base, src_temp, cache_size * sizeof(T));

  for (int nx = 0; nx < read_lens; ++nx) {
    index_output = thread_offset + nx;
    if (IsBoundary) {
      if (index_output >= total_num_output) {
        break;
      }
    }
    int index_src = config(index_output);
    if (index_src >= index_base && index_src < index_base + cache_size) {
      in_temp = src_temp[index_src - index_base];
    } else {
N
niuliling123 已提交
1179
      mfence_local();
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
      GM2LM(src + index_src, &in_temp, sizeof(T));
    }
    dst[nx] = in_temp;
  }
}

/**
 * @brief Read 1D data from global memory to register with broadcast form.
 *
 * @template paraments
 * T: The type of data stored in the global memory.
 * NX: The number of data continuously loaded by each thread.
 * NY: The number of data rows loaded by each thread, only NY = 1 was supported.
 * BlockSize: Identifies the current device thread index method. For xpu,
 * core_id() is used as the index.
 * IsBoundary: Indicates whether to perform block access storage out-of-bounds
 * judgment. When the number of data processed by the block is less than
 * NX x NY x core_num(), boundary judgment is required to avoid memory access
 * crossing the boundary.
 *
 * @param:
 * dst: The register pointer of the thread, the size is NX * NY.
 * src: The original input data pointer of kernel.
 * block_offset: The data offset of this block, core_num() * blockIdx.x * NX;
 * config: Calculation configuration of broadcast. It is used to calculate the
 * coordinate mapping relationship between output data and input data.
 * read_lens: The number of data continuously loaded by each thread.
 * total_num_output: Total number of original output.
 */
1209 1210 1211 1212 1213 1214 1215
template <typename T, int NX, int NY, int BlockSize, bool IsBoundary = false>
__device__ __inline__ void ReadDataBc(T* dst,
                                      const T _global_ptr_* src,
                                      uint32_t block_offset,
                                      const details::BroadcastConfig& config,
                                      int total_num_output,
                                      int read_lens) {
1216 1217 1218
  int thread_offset = block_offset + core_id() * read_lens;

  if (config.cmp_type == details::OptType::MNK_M1K) {
1219
    ReadDataBcM1kMnk<T>(dst, src, thread_offset, config, read_lens);
1220
  } else if (config.cmp_type == details::OptType::N_1) {
1221
    ReadDataBc1N<T>(dst, src, thread_offset, config, read_lens);
1222
  } else if (config.cmp_type == details::OptType::MN_M) {
1223
    ReadDataBcM1Mn<T>(dst, src, thread_offset, config, read_lens);
1224
  } else if (config.cmp_type == details::OptType::MN_N) {
1225
    ReadDataBc1NMn<T>(dst, src, thread_offset, config, read_lens);
1226
  } else if (config.cmp_type == details::OptType::MNK_1N1) {
1227
    ReadDataBc1N1Mnk<T>(dst, src, thread_offset, config, read_lens);
1228
  } else {
1229
    ReadDataBcCanNotCmp<T, IsBoundary>(
1230 1231 1232 1233
        dst, src, thread_offset, config, total_num_output, read_lens);
  }
}

1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256
/**
 * @brief Initialize register with data index.
 *
 * @template paraments
 * T: Data type of register.
 * NX: Number of data to initialize.
 * NY: Number of data to initialize, NY only can be 1.
 * BlockSize: Identifies the current device thread index method. For xpu,
 * core_id() is used as the index.
 *
 * @param:
 * dst: The register pointer of the thread, the size is NX.
 * init_data: The register pointer of init data, the size is NX.
 */
template <typename T, int NX, int NY, int BlockSize>
__device__ __forceinline__ void InitWithDataIndex(T* dst, int block_offset) {
  int thread_offset = block_offset + core_id() * NX;
#pragma unroll
  for (int nx = 0; nx < NX; ++nx) {
    dst[nx] = static_cast<T>(thread_offset + nx);
  }
}

1257
}  // namespace kps
1258
}  // namespace phi