From c62baec9a4803f1fd0245f17884a8978dc89beff Mon Sep 17 00:00:00 2001
From: zhaoting <zhaoting23@huawei.com>
Date: Fri, 19 Feb 2021 16:21:53 +0800
Subject: [PATCH] add parallel for some CPU ops

---
 .../kernel_compiler/cpu/adam_cpu_kernel.cc    |  48 +-
 .../kernel_compiler/cpu/adam_cpu_kernel.h     |   2 +-
 .../cpu/arithmetic_cpu_kernel.cc              | 504 +++++++++---------
 .../cpu/arithmetic_cpu_kernel.h               |  38 +-
 .../cpu/arithmetic_self_cpu_kernel.cc         | 318 +++++------
 .../cpu/arithmetic_self_cpu_kernel.h          |   1 -
 .../kernel_compiler/cpu/cast_cpu_kernel.cc    | 216 +-------
 .../kernel_compiler/cpu/cast_cpu_kernel.h     | 455 ++++++++++------
 .../cpu/layer_norm_cpu_kernel.cc              |  53 +-
 .../cpu/layer_norm_grad_cpu_kernel.cc         | 100 ++--
 .../cpu/unsorted_segment_sum_cpu_kernel.cc    |  37 +-
 11 files changed, 907 insertions(+), 865 deletions(-)

diff --git a/mindspore/ccsrc/backend/kernel_compiler/cpu/adam_cpu_kernel.cc b/mindspore/ccsrc/backend/kernel_compiler/cpu/adam_cpu_kernel.cc
index c5e28c59a8..591a964782 100644
--- a/mindspore/ccsrc/backend/kernel_compiler/cpu/adam_cpu_kernel.cc
+++ b/mindspore/ccsrc/backend/kernel_compiler/cpu/adam_cpu_kernel.cc
@@ -16,7 +16,6 @@
 #include "backend/kernel_compiler/cpu/adam_cpu_kernel.h"
 
 #include <cmath>
-#include <thread>
 #include "backend/kernel_compiler/cpu/mkldnn/mkl_kernel_engine.h"
 #include "runtime/device/cpu/cpu_device_address.h"
 #include "utils/ms_utils.h"
@@ -25,16 +24,19 @@ namespace mindspore {
 namespace kernel {
 template <typename T>
 void AdamCPUKernel::LaunchAdam(T *var, T *m, T *v, float lr, float beta1, float beta2, float epsilon, const T *gradient,
-                               size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    m[i] += (gradient[i] - m[i]) * (1 - beta1);
-    v[i] += (gradient[i] * gradient[i] - v[i]) * (1 - beta2);
-    if (use_nesterov) {
-      var[i] -= lr * (m[i] * beta1 + (1 - beta1) * gradient[i]) / (std::sqrt(v[i]) + epsilon);
-    } else {
-      var[i] -= lr * m[i] / (std::sqrt(v[i]) + epsilon);
+                               size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      m[i] += (gradient[i] - m[i]) * (1 - beta1);
+      v[i] += (gradient[i] * gradient[i] - v[i]) * (1 - beta2);
+      if (use_nesterov) {
+        var[i] -= lr * (m[i] * beta1 + (1 - beta1) * gradient[i]) / (std::sqrt(v[i]) + epsilon);
+      } else {
+        var[i] -= lr * m[i] / (std::sqrt(v[i]) + epsilon);
+      }
     }
-  }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 void AdamCPUKernel::InitKernel(const CNodePtr &kernel_node) {
@@ -84,31 +86,7 @@ bool AdamCPUKernel::Launch(const std::vector<kernel::AddressPtr> &inputs,
 
   // multithreading
   size_t lens = inputs[0]->size > 0 ? static_cast<size_t>(inputs[0]->size / sizeof(float)) : 1;
-  auto max_thread_num = std::thread::hardware_concurrency();
-  size_t thread_num = lens < 128 * max_thread_num ? std::ceil(lens / 128.0) : max_thread_num;
-  MS_LOG(INFO) << "Lens=" << lens << "; use thread_num=" << thread_num << "; max_thread_num: " << max_thread_num;
-  std::vector<std::thread> threads;
-  if (thread_num < 1) {
-    MS_LOG(ERROR) << "Invalid value: thread_num " << thread_num;
-    return false;
-  }
-  threads.reserve(thread_num);
-  size_t start = 0;
-  size_t once_compute_size = (lens + thread_num - 1) / thread_num;
-  if (once_compute_size < 1) {
-    MS_LOG(ERROR) << "Invalid value: once_compute_size " << once_compute_size;
-    return false;
-  }
-  while (start < lens) {
-    size_t end = (start + once_compute_size) > lens ? lens : (start + once_compute_size);
-    threads.emplace_back(std::thread(&AdamCPUKernel::LaunchAdam<float>, this, var, m, v, new_lr, beta1, beta2, epsilon,
-                                     gradient, start, end));
-    start += once_compute_size;
-  }
-  for (size_t i = 0; i < threads.size(); ++i) {
-    threads[i].join();
-  }
-
+  LaunchAdam<float>(var, m, v, new_lr, beta1, beta2, epsilon, gradient, lens);
   return true;
 }
 }  // namespace kernel
diff --git a/mindspore/ccsrc/backend/kernel_compiler/cpu/adam_cpu_kernel.h b/mindspore/ccsrc/backend/kernel_compiler/cpu/adam_cpu_kernel.h
index 79e42695bf..61a02ee93a 100644
--- a/mindspore/ccsrc/backend/kernel_compiler/cpu/adam_cpu_kernel.h
+++ b/mindspore/ccsrc/backend/kernel_compiler/cpu/adam_cpu_kernel.h
@@ -29,7 +29,7 @@ class AdamCPUKernel : public CPUKernel {
   ~AdamCPUKernel() override = default;
   template <typename T>
   void LaunchAdam(T *var, T *m, T *v, float lr, float beta1, float beta2, float epsilon, const T *gradient,
-                  size_t start, size_t end);
+                  size_t size);
   void InitKernel(const CNodePtr &kernel_node) override;
 
   bool Launch(const std::vector<AddressPtr> &inputs, const std::vector<AddressPtr> &workspace,
diff --git a/mindspore/ccsrc/backend/kernel_compiler/cpu/arithmetic_cpu_kernel.cc b/mindspore/ccsrc/backend/kernel_compiler/cpu/arithmetic_cpu_kernel.cc
index 82920c84ee..262fa0024f 100644
--- a/mindspore/ccsrc/backend/kernel_compiler/cpu/arithmetic_cpu_kernel.cc
+++ b/mindspore/ccsrc/backend/kernel_compiler/cpu/arithmetic_cpu_kernel.cc
@@ -15,7 +15,6 @@
  */
 #include <cmath>
 #include <string>
-#include <thread>
 #include <map>
 #include "backend/kernel_compiler/cpu/arithmetic_cpu_kernel.h"
 #include "runtime/device/cpu/cpu_device_address.h"
@@ -23,227 +22,285 @@
 namespace mindspore {
 namespace kernel {
 template <typename T>
-void ArithmeticCPUKernel::AssignAdd(T *input1, const T *input2, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = input1[i] + input2[i];
-    input1[i] = out[i];
-  }
+void ArithmeticCPUKernel::AssignAdd(T *input1, const T *input2, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = input1[i] + input2[i];
+      input1[i] = out[i];
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::Add(const T *input1, const T *input2, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    out[i] = input1[idx[0]] + input2[idx[1]];
-  }
+void ArithmeticCPUKernel::Add(const T *input1, const T *input2, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      out[i] = input1[idx[0]] + input2[idx[1]];
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::Sub(const T *input1, const T *input2, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    out[i] = input1[idx[0]] - input2[idx[1]];
-  }
+void ArithmeticCPUKernel::Sub(const T *input1, const T *input2, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      out[i] = input1[idx[0]] - input2[idx[1]];
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::Mul(const T *input1, const T *input2, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    out[i] = input1[idx[0]] * input2[idx[1]];
-  }
+void ArithmeticCPUKernel::Mul(const T *input1, const T *input2, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      out[i] = input1[idx[0]] * input2[idx[1]];
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::RealDiv(const T *input1, const T *input2, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    auto dividend = input1[idx[0]];
-    auto divisor = input2[idx[1]];
-    if (divisor == 0) {
-      if (dividend == 0) {
-        out[i] = std::numeric_limits<T>::quiet_NaN();
+void ArithmeticCPUKernel::RealDiv(const T *input1, const T *input2, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      auto dividend = input1[idx[0]];
+      auto divisor = input2[idx[1]];
+      if (divisor == 0) {
+        if (dividend == 0) {
+          out[i] = std::numeric_limits<T>::quiet_NaN();
+          continue;
+        }
+        if (std::numeric_limits<T>::has_infinity) {
+          out[i] = dividend > 0 ? std::numeric_limits<T>::infinity() : -std::numeric_limits<T>::infinity();
+        } else {
+          out[i] = dividend > 0 ? std::numeric_limits<T>::max() : std::numeric_limits<T>::min();
+        }
         continue;
       }
-      if (std::numeric_limits<T>::has_infinity) {
-        out[i] = dividend > 0 ? std::numeric_limits<T>::infinity() : -std::numeric_limits<T>::infinity();
-      } else {
-        out[i] = dividend > 0 ? std::numeric_limits<T>::max() : std::numeric_limits<T>::min();
-      }
-      continue;
+      out[i] = dividend / divisor;
     }
-    out[i] = dividend / divisor;
-  }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::Div(const T *input1, const T *input2, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    auto dividend = input1[idx[0]];
-    auto divisor = input2[idx[1]];
-    if (divisor == 0) {
-      if (dividend == 0) {
-        out[i] = std::numeric_limits<T>::quiet_NaN();
+void ArithmeticCPUKernel::Div(const T *input1, const T *input2, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      auto dividend = input1[idx[0]];
+      auto divisor = input2[idx[1]];
+      if (divisor == 0) {
+        if (dividend == 0) {
+          out[i] = std::numeric_limits<T>::quiet_NaN();
+          continue;
+        }
+        if (std::numeric_limits<T>::has_infinity) {
+          out[i] = dividend > 0 ? std::numeric_limits<T>::infinity() : -std::numeric_limits<T>::infinity();
+        } else {
+          out[i] = dividend > 0 ? std::numeric_limits<T>::max() : std::numeric_limits<T>::min();
+        }
         continue;
       }
-      if (std::numeric_limits<T>::has_infinity) {
-        out[i] = dividend > 0 ? std::numeric_limits<T>::infinity() : -std::numeric_limits<T>::infinity();
-      } else {
-        out[i] = dividend > 0 ? std::numeric_limits<T>::max() : std::numeric_limits<T>::min();
-      }
-      continue;
+      out[i] = dividend / divisor;
     }
-    out[i] = dividend / divisor;
-  }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::FloorDiv(const T *input1, const T *input2, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    auto dividend = input1[idx[0]];
-    auto divisor = input2[idx[1]];
-    if (divisor == 0) {
-      if (dividend == 0) {
-        out[i] = std::numeric_limits<T>::quiet_NaN();
+void ArithmeticCPUKernel::FloorDiv(const T *input1, const T *input2, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      auto dividend = input1[idx[0]];
+      auto divisor = input2[idx[1]];
+      if (divisor == 0) {
+        if (dividend == 0) {
+          out[i] = std::numeric_limits<T>::quiet_NaN();
+          continue;
+        }
+        if (std::numeric_limits<T>::has_infinity) {
+          out[i] = dividend > 0 ? std::numeric_limits<T>::infinity() : -std::numeric_limits<T>::infinity();
+        } else {
+          out[i] = dividend > 0 ? std::numeric_limits<T>::max() : std::numeric_limits<T>::min();
+        }
         continue;
       }
-      if (std::numeric_limits<T>::has_infinity) {
-        out[i] = dividend > 0 ? std::numeric_limits<T>::infinity() : -std::numeric_limits<T>::infinity();
-      } else {
-        out[i] = dividend > 0 ? std::numeric_limits<T>::max() : std::numeric_limits<T>::min();
-      }
-      continue;
+      out[i] = floor(dividend / divisor);
     }
-    out[i] = floor(dividend / divisor);
-  }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::Mod(const T *input1, const T *input2, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    auto x = static_cast<double>(input1[idx[0]]);
-    auto y = static_cast<double>(input2[idx[1]]);
-    auto data_div = x / y;
-    auto data_div_min = data_div < 0.0 ? data_div : 0.0;
-    auto data_div_max = data_div > 0.0 ? data_div : 0.0;
-    auto data_div_max_floor = floor(data_div_max);
-    auto data_div_min_ceil = ceil(data_div_min);
-    auto data_div_res = data_div_max_floor + data_div_min_ceil;
-    out[i] = static_cast<T>(x - data_div_res * y);
-  }
+void ArithmeticCPUKernel::Mod(const T *input1, const T *input2, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      auto x = static_cast<double>(input1[idx[0]]);
+      auto y = static_cast<double>(input2[idx[1]]);
+      auto data_div = x / y;
+      auto data_div_min = data_div < 0.0 ? data_div : 0.0;
+      auto data_div_max = data_div > 0.0 ? data_div : 0.0;
+      auto data_div_max_floor = floor(data_div_max);
+      auto data_div_min_ceil = ceil(data_div_min);
+      auto data_div_res = data_div_max_floor + data_div_min_ceil;
+      out[i] = static_cast<T>(x - data_div_res * y);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::Pow(const T *input1, const T *input2, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    auto x = static_cast<double>(input1[idx[0]]);
-    auto y = static_cast<double>(input2[idx[1]]);
-    out[i] = static_cast<T>(std::pow(x, y));
-  }
+void ArithmeticCPUKernel::Pow(const T *input1, const T *input2, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      auto x = static_cast<double>(input1[idx[0]]);
+      auto y = static_cast<double>(input2[idx[1]]);
+      out[i] = static_cast<T>(std::pow(x, y));
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::Less(const T *input1, const T *input2, bool *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    out[i] = input1[idx[0]] < input2[idx[1]];
-  }
+void ArithmeticCPUKernel::Less(const T *input1, const T *input2, bool *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      out[i] = input1[idx[0]] < input2[idx[1]];
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::Equal(const T *input1, const T *input2, bool *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    out[i] = input1[idx[0]] == input2[idx[1]];
-  }
+void ArithmeticCPUKernel::Equal(const T *input1, const T *input2, bool *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      out[i] = input1[idx[0]] == input2[idx[1]];
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::NotEqual(const T *input1, const T *input2, bool *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    out[i] = input1[idx[0]] != input2[idx[1]];
-  }
+void ArithmeticCPUKernel::NotEqual(const T *input1, const T *input2, bool *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      out[i] = input1[idx[0]] != input2[idx[1]];
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::LogicalAnd(const T *input1, const T *input2, bool *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    out[i] = input1[idx[0]] && input2[idx[1]];
-  }
+void ArithmeticCPUKernel::LogicalAnd(const T *input1, const T *input2, bool *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      out[i] = input1[idx[0]] && input2[idx[1]];
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::LogicalOr(const T *input1, const T *input2, bool *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    out[i] = input1[idx[0]] || input2[idx[1]];
-  }
+void ArithmeticCPUKernel::LogicalOr(const T *input1, const T *input2, bool *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      out[i] = input1[idx[0]] || input2[idx[1]];
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::SquaredDifference(const T *input1, const T *input2, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    T diff = input1[idx[0]] - input2[idx[1]];
-    out[i] = diff * diff;
-  }
+void ArithmeticCPUKernel::SquaredDifference(const T *input1, const T *input2, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      T diff = input1[idx[0]] - input2[idx[1]];
+      out[i] = diff * diff;
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::Greater(const T *input1, const T *input2, bool *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    out[i] = input1[idx[0]] > input2[idx[1]];
-  }
+void ArithmeticCPUKernel::Greater(const T *input1, const T *input2, bool *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      out[i] = input1[idx[0]] > input2[idx[1]];
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::GreaterEqual(const T *input1, const T *input2, bool *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    out[i] = input1[idx[0]] >= input2[idx[1]];
-  }
+void ArithmeticCPUKernel::GreaterEqual(const T *input1, const T *input2, bool *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      out[i] = input1[idx[0]] >= input2[idx[1]];
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::LessEqual(const T *input1, const T *input2, bool *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    out[i] = input1[idx[0]] <= input2[idx[1]];
-  }
+void ArithmeticCPUKernel::LessEqual(const T *input1, const T *input2, bool *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      out[i] = input1[idx[0]] <= input2[idx[1]];
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ArithmeticCPUKernel::Atan2(const T *input1, const T *input2, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    std::vector<size_t> idx;
-    GenIndex(i, &idx);
-    out[i] = atan2(input1[idx[0]], input2[idx[1]]);
-  }
+void ArithmeticCPUKernel::Atan2(const T *input1, const T *input2, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> idx;
+      GenIndex(i, &idx);
+      out[i] = atan2(input1[idx[0]], input2[idx[1]]);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
+
 static const std::map<std::string, OperateType> kArithmeticBinOpTypeMap = {
   {prim::kPrimGreater->name(), GREATER},
   {prim::kPrimAdd->name(), ADD},
@@ -352,49 +409,25 @@ void ArithmeticCPUKernel::LaunchKernelLogic(const std::vector<AddressPtr> &input
   T *input1 = reinterpret_cast<T *>(inputs[0]->addr);
   T *input2 = reinterpret_cast<T *>(inputs[1]->addr);
   bool *output = reinterpret_cast<bool *>(outputs[0]->addr);
-
   size_t lens = outputs[0]->size > 0 ? static_cast<size_t>(outputs[0]->size / sizeof(bool)) : 1;
-  auto max_thread_num = std::thread::hardware_concurrency();
-  size_t thread_num = lens < 128 * max_thread_num ? std::ceil(lens / 128.0) : max_thread_num;
-  MS_LOG(INFO) << "Lens=" << lens << "; use thread_num=" << thread_num << "; max_thread_num: " << max_thread_num;
-  std::vector<std::thread> threads;
-  if (thread_num < 1) {
-    MS_LOG(ERROR) << "Invalid value: thread_num " << thread_num;
-    return;
-  }
-  threads.reserve(thread_num);
-  size_t start = 0;
-  size_t once_compute_size = (lens + thread_num - 1) / thread_num;
-  if (once_compute_size < 1) {
-    MS_LOG(ERROR) << "Invalid value: once_compute_size " << once_compute_size;
-    return;
-  }
-  while (start < lens) {
-    size_t end = (start + once_compute_size) > lens ? lens : (start + once_compute_size);
-    if (operate_type_ == LESS) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::Less<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == EQUAL) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::Equal<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == NOTEQUAL) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::NotEqual<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == GREATER) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::Greater<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == GREATEREQUAL) {
-      threads.emplace_back(
-        std::thread(&ArithmeticCPUKernel::GreaterEqual<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == LESSEQUAL) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::LessEqual<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == LOGICALAND) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::LogicalAnd<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == LOGICALOR) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::LogicalOr<T>, this, input1, input2, output, start, end));
-    } else {
-      MS_LOG(EXCEPTION) << "Not support " << operate_type_;
-    }
-    start += once_compute_size;
-  }
-  for (size_t i = 0; i < threads.size(); ++i) {
-    threads[i].join();
+  if (operate_type_ == LESS) {
+    Less<T>(input1, input2, output, lens);
+  } else if (operate_type_ == EQUAL) {
+    Equal<T>(input1, input2, output, lens);
+  } else if (operate_type_ == NOTEQUAL) {
+    NotEqual<T>(input1, input2, output, lens);
+  } else if (operate_type_ == GREATER) {
+    Greater<T>(input1, input2, output, lens);
+  } else if (operate_type_ == GREATEREQUAL) {
+    GreaterEqual<T>(input1, input2, output, lens);
+  } else if (operate_type_ == LESSEQUAL) {
+    LessEqual<T>(input1, input2, output, lens);
+  } else if (operate_type_ == LOGICALAND) {
+    LogicalAnd<T>(input1, input2, output, lens);
+  } else if (operate_type_ == LOGICALOR) {
+    LogicalOr<T>(input1, input2, output, lens);
+  } else {
+    MS_LOG(EXCEPTION) << "Not support " << operate_type_;
   }
 }
 
@@ -409,53 +442,30 @@ void ArithmeticCPUKernel::LaunchKernel(const std::vector<AddressPtr> &inputs, co
   T *output = reinterpret_cast<T *>(outputs[0]->addr);
 
   size_t lens = outputs[0]->size > 0 ? static_cast<size_t>(outputs[0]->size / sizeof(T)) : 1;
-  auto max_thread_num = std::thread::hardware_concurrency();
-  size_t thread_num = lens < 128 * max_thread_num ? std::ceil(lens / 128.0) : max_thread_num;
-  MS_LOG(INFO) << "Lens=" << lens << "; use thread_num=" << thread_num << "; max_thread_num: " << max_thread_num;
-  std::vector<std::thread> threads;
-  if (thread_num < 1) {
-    MS_LOG(ERROR) << "Invalid value: thread_num " << thread_num;
-    return;
-  }
-  threads.reserve(thread_num);
-  size_t start = 0;
-  size_t once_compute_size = (lens + thread_num - 1) / thread_num;
-  if (once_compute_size < 1) {
-    MS_LOG(ERROR) << "Invalid value: once_compute_size " << once_compute_size;
-    return;
-  }
-  while (start < lens) {
-    size_t end = (start + once_compute_size) > lens ? lens : (start + once_compute_size);
-    if (operate_type_ == ADD) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::Add<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == SUB) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::Sub<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == MUL) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::Mul<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == REALDIV) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::RealDiv<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == DIV) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::Div<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == FLOORDIV) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::FloorDiv<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == MOD) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::Mod<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == POW) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::Pow<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == ASSIGNADD) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::AssignAdd<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == ATAN2) {
-      threads.emplace_back(std::thread(&ArithmeticCPUKernel::Atan2<T>, this, input1, input2, output, start, end));
-    } else if (operate_type_ == SQUAREDDIFFERENCE) {
-      threads.emplace_back(
-        std::thread(&ArithmeticCPUKernel::SquaredDifference<T>, this, input1, input2, output, start, end));
-    } else {
-      MS_LOG(EXCEPTION) << "Not support " << operate_type_;
-    }
-    start += once_compute_size;
-  }
-  for (size_t i = 0; i < threads.size(); ++i) {
-    threads[i].join();
+  if (operate_type_ == ADD) {
+    Add<T>(input1, input2, output, lens);
+  } else if (operate_type_ == SUB) {
+    Sub<T>(input1, input2, output, lens);
+  } else if (operate_type_ == MUL) {
+    Mul<T>(input1, input2, output, lens);
+  } else if (operate_type_ == REALDIV) {
+    RealDiv<T>(input1, input2, output, lens);
+  } else if (operate_type_ == DIV) {
+    Div<T>(input1, input2, output, lens);
+  } else if (operate_type_ == FLOORDIV) {
+    FloorDiv<T>(input1, input2, output, lens);
+  } else if (operate_type_ == MOD) {
+    Mod<T>(input1, input2, output, lens);
+  } else if (operate_type_ == POW) {
+    Pow<T>(input1, input2, output, lens);
+  } else if (operate_type_ == ASSIGNADD) {
+    AssignAdd<T>(input1, input2, output, lens);
+  } else if (operate_type_ == ATAN2) {
+    Atan2<T>(input1, input2, output, lens);
+  } else if (operate_type_ == SQUAREDDIFFERENCE) {
+    SquaredDifference<T>(input1, input2, output, lens);
+  } else {
+    MS_LOG(EXCEPTION) << "Not support " << operate_type_;
   }
 }
 }  // namespace kernel
diff --git a/mindspore/ccsrc/backend/kernel_compiler/cpu/arithmetic_cpu_kernel.h b/mindspore/ccsrc/backend/kernel_compiler/cpu/arithmetic_cpu_kernel.h
index cc2ab1a4a3..8a707065c0 100644
--- a/mindspore/ccsrc/backend/kernel_compiler/cpu/arithmetic_cpu_kernel.h
+++ b/mindspore/ccsrc/backend/kernel_compiler/cpu/arithmetic_cpu_kernel.h
@@ -40,43 +40,43 @@ class ArithmeticCPUKernel : public CPUKernel {
  private:
   void GenIndex(size_t num, std::vector<size_t> *tmp);
   template <typename T>
-  void Sub(const T *input1, const T *input2, T *out, size_t start, size_t end);
+  void Sub(const T *input1, const T *input2, T *out, size_t size);
   template <typename T>
-  void Add(const T *input1, const T *input2, T *out, size_t start, size_t end);
+  void Add(const T *input1, const T *input2, T *out, size_t size);
   template <typename T>
-  void Mul(const T *input1, const T *input2, T *out, size_t start, size_t end);
+  void Mul(const T *input1, const T *input2, T *out, size_t size);
   template <typename T>
-  void RealDiv(const T *input1, const T *input2, T *out, size_t start, size_t end);
+  void RealDiv(const T *input1, const T *input2, T *out, size_t size);
   template <typename T>
-  void Div(const T *input1, const T *input2, T *out, size_t start, size_t end);
+  void Div(const T *input1, const T *input2, T *out, size_t size);
   template <typename T>
-  void FloorDiv(const T *input1, const T *input2, T *out, size_t start, size_t end);
+  void FloorDiv(const T *input1, const T *input2, T *out, size_t size);
   template <typename T>
-  void Mod(const T *input1, const T *input2, T *out, size_t start, size_t end);
+  void Mod(const T *input1, const T *input2, T *out, size_t size);
   template <typename T>
-  void Pow(const T *input1, const T *input2, T *out, size_t start, size_t end);
+  void Pow(const T *input1, const T *input2, T *out, size_t size);
   template <typename T>
-  void AssignAdd(T *input1, const T *input2, T *out, size_t start, size_t end);
+  void AssignAdd(T *input1, const T *input2, T *out, size_t size);
   template <typename T>
-  void Atan2(const T *input1, const T *input2, T *out, size_t start, size_t end);
+  void Atan2(const T *input1, const T *input2, T *out, size_t size);
   template <typename T>
-  void Less(const T *input1, const T *input2, bool *out, size_t start, size_t end);
+  void Less(const T *input1, const T *input2, bool *out, size_t size);
   template <typename T>
-  void Equal(const T *input1, const T *input2, bool *out, size_t start, size_t end);
+  void Equal(const T *input1, const T *input2, bool *out, size_t size);
   template <typename T>
-  void NotEqual(const T *input1, const T *input2, bool *out, size_t start, size_t end);
+  void NotEqual(const T *input1, const T *input2, bool *out, size_t size);
   template <typename T>
-  void SquaredDifference(const T *input1, const T *input2, T *out, size_t start, size_t end);
+  void SquaredDifference(const T *input1, const T *input2, T *out, size_t size);
   template <typename T>
-  void Greater(const T *input1, const T *input2, bool *out, size_t start, size_t end);
+  void Greater(const T *input1, const T *input2, bool *out, size_t size);
   template <typename T>
-  void GreaterEqual(const T *input1, const T *input2, bool *out, size_t start, size_t end);
+  void GreaterEqual(const T *input1, const T *input2, bool *out, size_t size);
   template <typename T>
-  void LessEqual(const T *input1, const T *input2, bool *out, size_t start, size_t end);
+  void LessEqual(const T *input1, const T *input2, bool *out, size_t size);
   template <typename T>
-  void LogicalAnd(const T *input1, const T *input2, bool *out, size_t start, size_t end);
+  void LogicalAnd(const T *input1, const T *input2, bool *out, size_t size);
   template <typename T>
-  void LogicalOr(const T *input1, const T *input2, bool *out, size_t start, size_t end);
+  void LogicalOr(const T *input1, const T *input2, bool *out, size_t size);
   std::vector<size_t> input_shape0_;
   std::vector<size_t> input_shape1_;
   std::vector<size_t> input_element_num0_;
diff --git a/mindspore/ccsrc/backend/kernel_compiler/cpu/arithmetic_self_cpu_kernel.cc b/mindspore/ccsrc/backend/kernel_compiler/cpu/arithmetic_self_cpu_kernel.cc
index 04bfd4997e..a73682a888 100644
--- a/mindspore/ccsrc/backend/kernel_compiler/cpu/arithmetic_self_cpu_kernel.cc
+++ b/mindspore/ccsrc/backend/kernel_compiler/cpu/arithmetic_self_cpu_kernel.cc
@@ -24,152 +24,212 @@ namespace mindspore {
 namespace kernel {
 namespace {
 template <typename T>
-void Square(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = in[i] * in[i];
-  }
+void Square(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = in[i] * in[i];
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void Sign(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    if (in[i] < 0) {
-      out[i] = -1;
-    } else if (in[i] > 0) {
-      out[i] = 1;
-    } else {
-      out[i] = 0;
+void Sign(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      if (in[i] < 0) {
+        out[i] = -1;
+      } else if (in[i] > 0) {
+        out[i] = 1;
+      } else {
+        out[i] = 0;
+      }
     }
-  }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void Neg(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = -in[i];
-  }
+void Neg(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = -in[i];
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void LogicalNot(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = !in[i];
-  }
+void LogicalNot(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = !in[i];
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void OnesLike(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = static_cast<T>(1);
-  }
+void OnesLike(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = static_cast<T>(1);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ZerosLike(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = static_cast<T>(0);
-  }
+void ZerosLike(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = static_cast<T>(0);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void Floor(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = static_cast<T>(floor(in[i]));
-  }
+void Floor(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = static_cast<T>(floor(in[i]));
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void Reciprocal(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = static_cast<T>(1.0 / in[i]);
-  }
+void Reciprocal(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = static_cast<T>(1.0 / in[i]);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void Gelu(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    T x = in[i];
-    auto double_x = static_cast<T>(x);
-    T tanh_res = (T)std::tanh(0.7978845608 * (double_x + 0.044715 * double_x * double_x * double_x));
-    out[i] = x * ((T)1.0 + tanh_res) / (T)2.0;
-  }
+void Gelu(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      T x = in[i];
+      auto double_x = static_cast<T>(x);
+      T tanh_res = (T)std::tanh(0.7978845608 * (double_x + 0.044715 * double_x * double_x * double_x));
+      out[i] = x * ((T)1.0 + tanh_res) / (T)2.0;
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void Asin(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = asin(in[i]);
-  }
+void Asin(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = asin(in[i]);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void ACos(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = acos(in[i]);
-  }
+void ACos(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = acos(in[i]);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void Atan(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = atan(in[i]);
-  }
+void Atan(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = atan(in[i]);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void Sin(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = sin(in[i]);
-  }
+void Sin(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = sin(in[i]);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void Cos(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = cos(in[i]);
-  }
+void Cos(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = cos(in[i]);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void Tan(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = tan(in[i]);
-  }
+void Tan(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = tan(in[i]);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void Sinh(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = sinh(in[i]);
-  }
+void Sinh(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = sinh(in[i]);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void Cosh(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = cosh(in[i]);
-  }
+void Cosh(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = cosh(in[i]);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void Asinh(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = asinh(in[i]);
-  }
+void Asinh(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = asinh(in[i]);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void Acosh(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = acosh(in[i]);
-  }
+void Acosh(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = acosh(in[i]);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename T>
-void Atanh(const T *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = atanh(in[i]);
-  }
+void Atanh(const T *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = atanh(in[i]);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 }  // namespace
 
@@ -223,79 +283,31 @@ void ArithmeticSelfCPUKernel::LaunchKernelLogic(const std::vector<AddressPtr> &i
   T *input = reinterpret_cast<T *>(inputs[0]->addr);
   T *output = reinterpret_cast<T *>(outputs[0]->addr);
   size_t lens = outputs[0]->size > 0 ? static_cast<size_t>(outputs[0]->size / sizeof(T)) : 1;
-
-  auto max_thread_num = std::thread::hardware_concurrency();
-  size_t thread_num = lens < 128 * max_thread_num ? std::ceil(lens / 128.0) : max_thread_num;
-  MS_LOG(INFO) << "Lens=" << lens << "; use thread_num=" << thread_num << "; max_thread_num: " << max_thread_num;
-  std::vector<std::thread> threads;
-  if (thread_num < 1) {
-    MS_LOG(ERROR) << "Invalid value: thread_num " << thread_num;
-    return;
-  }
-  threads.reserve(thread_num);
-  size_t start = 0;
-  size_t once_compute_size = (lens + thread_num - 1) / thread_num;
-  if (once_compute_size < 1) {
-    MS_LOG(ERROR) << "Invalid value: once_compute_size " << once_compute_size;
-    return;
-  }
-  while (start < lens) {
-    size_t end = (start + once_compute_size) > lens ? lens : (start + once_compute_size);
-    if (operate_type_ == LOGICALNOT) {
-      threads.emplace_back(std::thread(LogicalNot<T>, input, output, start, end));
-    }
-    start += once_compute_size;
-  }
-  for (size_t i = 0; i < threads.size(); ++i) {
-    threads[i].join();
-  }
+  LogicalNot<T>(input, output, lens);
+  return;
 }
 
 template <typename T>
 void ArithmeticSelfCPUKernel::LaunchKernel(const std::vector<AddressPtr> &inputs,
                                            const std::vector<AddressPtr> &outputs) {
-  if (target_dtype_ == kNumberTypeBool) {
-    LaunchKernelLogic<T>(inputs, outputs);
-    return;
-  }
   T *input = reinterpret_cast<T *>(inputs[0]->addr);
   T *output = reinterpret_cast<T *>(outputs[0]->addr);
   size_t lens = outputs[0]->size > 0 ? static_cast<size_t>(outputs[0]->size / sizeof(T)) : 1;
-
-  auto max_thread_num = std::thread::hardware_concurrency();
-  size_t thread_num = lens < 128 * max_thread_num ? std::ceil(lens / 128.0) : max_thread_num;
-  MS_LOG(INFO) << "Lens=" << lens << "; use thread_num=" << thread_num << "; max_thread_num: " << max_thread_num;
-  std::vector<std::thread> threads;
-  if (thread_num < 1) {
-    MS_LOG(ERROR) << "Invalid value: thread_num " << thread_num;
-    return;
-  }
-  threads.reserve(thread_num);
-  size_t start = 0;
-  size_t once_compute_size = (lens + thread_num - 1) / thread_num;
-  if (once_compute_size < 1) {
-    MS_LOG(ERROR) << "Invalid value: once_compute_size " << once_compute_size;
-    return;
-  }
-  static const std::map<OperateType, std::function<void(const T *in, T *out, size_t start, size_t end)>>
-    kArithmeticOpFuncMap = {{SQUARE, Square<T>},     {SIGN, Sign<T>},
-                            {NEG, Neg<T>},           {LOGICALNOT, LogicalNot<T>},
-                            {ONESLIKE, OnesLike<T>}, {ZEROSLIKE, ZerosLike<T>},
-                            {FLOOR, Floor<T>},       {RECIPROCAL, Reciprocal<T>},
-                            {GELU, Gelu<T>},         {SIN, Sin<T>},
-                            {COS, Cos<T>},           {TAN, Tan<T>},
-                            {ASIN, Asin<T>},         {ACOS, ACos<T>},
-                            {ATAN, Atan<T>},         {SINH, Sinh<T>},
-                            {COSH, Cosh<T>},         {ASINH, Asinh<T>},
-                            {ACOSH, Acosh<T>},       {ATANH, Atanh<T>}};
-
-  while (start < lens) {
-    size_t end = (start + once_compute_size) > lens ? lens : (start + once_compute_size);
-    threads.emplace_back(std::thread(kArithmeticOpFuncMap.at(operate_type_), input, output, start, end));
-    start += once_compute_size;
-  }
-  for (size_t i = 0; i < threads.size(); ++i) {
-    threads[i].join();
+  static const std::map<OperateType, std::function<void(const T *in, T *out, size_t size)>> kArithmeticOpFuncMap = {
+    {SQUARE, Square<T>},     {SIGN, Sign<T>},
+    {NEG, Neg<T>},           {LOGICALNOT, LogicalNot<T>},
+    {ONESLIKE, OnesLike<T>}, {ZEROSLIKE, ZerosLike<T>},
+    {FLOOR, Floor<T>},       {RECIPROCAL, Reciprocal<T>},
+    {GELU, Gelu<T>},         {SIN, Sin<T>},
+    {COS, Cos<T>},           {TAN, Tan<T>},
+    {ASIN, Asin<T>},         {ACOS, ACos<T>},
+    {ATAN, Atan<T>},         {SINH, Sinh<T>},
+    {COSH, Cosh<T>},         {ASINH, Asinh<T>},
+    {ACOSH, Acosh<T>},       {ATANH, Atanh<T>}};
+  if (kArithmeticOpFuncMap.find(operate_type_) != kArithmeticOpFuncMap.end()) {
+    kArithmeticOpFuncMap.at(operate_type_)(input, output, lens);
+  } else {
+    MS_LOG(EXCEPTION) << "Not support " << operate_type_;
   }
 }
 }  // namespace kernel
diff --git a/mindspore/ccsrc/backend/kernel_compiler/cpu/arithmetic_self_cpu_kernel.h b/mindspore/ccsrc/backend/kernel_compiler/cpu/arithmetic_self_cpu_kernel.h
index 0a68b3722b..84b6e1d6b0 100644
--- a/mindspore/ccsrc/backend/kernel_compiler/cpu/arithmetic_self_cpu_kernel.h
+++ b/mindspore/ccsrc/backend/kernel_compiler/cpu/arithmetic_self_cpu_kernel.h
@@ -34,7 +34,6 @@ class ArithmeticSelfCPUKernel : public CPUKernel {
 
   template <typename T>
   void LaunchKernel(const std::vector<AddressPtr> &inputs, const std::vector<AddressPtr> &outputs);
-
   template <typename T>
   void LaunchKernelLogic(const std::vector<AddressPtr> &inputs, const std::vector<AddressPtr> &outputs);
 
diff --git a/mindspore/ccsrc/backend/kernel_compiler/cpu/cast_cpu_kernel.cc b/mindspore/ccsrc/backend/kernel_compiler/cpu/cast_cpu_kernel.cc
index b0226fd40b..22574b98c4 100644
--- a/mindspore/ccsrc/backend/kernel_compiler/cpu/cast_cpu_kernel.cc
+++ b/mindspore/ccsrc/backend/kernel_compiler/cpu/cast_cpu_kernel.cc
@@ -16,220 +16,38 @@
 #include <cmath>
 #include <map>
 #include <string>
-#include <thread>
 #include "backend/kernel_compiler/cpu/cast_cpu_kernel.h"
 #include "runtime/device/cpu/cpu_device_address.h"
 
 namespace mindspore {
 namespace kernel {
 template <typename S, typename T>
-void Cast(const S *in, T *out, size_t start, size_t end) {
-  for (size_t i = start; i < end; i++) {
-    out[i] = static_cast<T>(in[i]);
-  }
+void Cast(const S *in, T *out, size_t size) {
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      out[i] = static_cast<T>(in[i]);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, size);
 }
 
 template <typename S, typename T>
-void LaunchCast(const std::vector<kernel::AddressPtr> &inputs, const std::vector<kernel::AddressPtr> &outputs) {
-  S *input = reinterpret_cast<S *>(inputs[0]->addr);
-  T *output = reinterpret_cast<T *>(outputs[0]->addr);
-  MS_LOG(DEBUG) << "Type source: " << typeid(S).name() << "; target: " << typeid(T).name();
-
-  size_t lens = outputs[0]->size > 0 ? static_cast<size_t>(outputs[0]->size / sizeof(T)) : 1;
-  auto max_thread_num = std::thread::hardware_concurrency();
-  size_t thread_num = lens < 128 * max_thread_num ? std::ceil(lens / 128.0) : max_thread_num;
-  MS_LOG(INFO) << "Lens=" << lens << "; use thread_num=" << thread_num << "; max_thread_num: " << max_thread_num;
-  std::vector<std::thread> threads;
-  if (thread_num < 1) {
-    MS_LOG(ERROR) << "Invalid value: thread_num " << thread_num;
-    return;
-  }
-  threads.reserve(thread_num);
-  size_t start = 0;
-  size_t once_compute_size = (lens + thread_num - 1) / thread_num;
-  if (once_compute_size < 1) {
-    MS_LOG(ERROR) << "Invalid value: once_compute_size " << once_compute_size;
-    return;
-  }
-  while (start < lens) {
-    size_t end = (start + once_compute_size) > lens ? lens : (start + once_compute_size);
-    threads.emplace_back(std::thread(Cast<S, T>, input, output, start, end));
-    start += once_compute_size;
-  }
-  for (size_t i = 0; i < threads.size(); ++i) {
-    threads[i].join();
-  }
-}
-
-void CastCPUKernel::InitKernel(const CNodePtr &kernel_node) {
+void CastCPUKernel<S, T>::InitKernel(const CNodePtr &kernel_node) {
   MS_EXCEPTION_IF_NULL(kernel_node);
   source_dtype = AnfAlgo::GetPrevNodeOutputDeviceDataType(kernel_node, 0);
   target_dtype = AnfAlgo::GetOutputInferDataType(kernel_node, 0);
 }
 
-bool CastCPUKernel::Launch(const std::vector<kernel::AddressPtr> &inputs,
-                           const std::vector<kernel::AddressPtr> & /*workspace*/,
-                           const std::vector<kernel::AddressPtr> &outputs) {
-  using TypePair =
-    std::function<void(const std::vector<kernel::AddressPtr> &, const std::vector<kernel::AddressPtr> &)>;
-  std::map<TypeId, std::map<TypeId, TypePair>> mode_map;
-  mode_map[kNumberTypeBool][kNumberTypeFloat16] = LaunchCast<bool, float16>;
-  mode_map[kNumberTypeBool][kNumberTypeFloat32] = LaunchCast<bool, float>;
-  mode_map[kNumberTypeBool][kNumberTypeFloat64] = LaunchCast<bool, double>;
-  mode_map[kNumberTypeBool][kNumberTypeInt8] = LaunchCast<bool, int8_t>;
-  mode_map[kNumberTypeBool][kNumberTypeInt16] = LaunchCast<bool, int16_t>;
-  mode_map[kNumberTypeBool][kNumberTypeInt32] = LaunchCast<bool, int32_t>;
-  mode_map[kNumberTypeBool][kNumberTypeInt64] = LaunchCast<bool, int64_t>;
-  mode_map[kNumberTypeBool][kNumberTypeUInt8] = LaunchCast<bool, uint8_t>;
-  mode_map[kNumberTypeBool][kNumberTypeUInt16] = LaunchCast<bool, uint16_t>;
-  mode_map[kNumberTypeBool][kNumberTypeUInt32] = LaunchCast<bool, uint32_t>;
-  mode_map[kNumberTypeBool][kNumberTypeUInt64] = LaunchCast<bool, uint64_t>;
-  mode_map[kNumberTypeBool][kNumberTypeBool] = LaunchCast<bool, bool>;
-
-  mode_map[kNumberTypeFloat16][kNumberTypeFloat16] = LaunchCast<float16, float16>;
-  mode_map[kNumberTypeFloat16][kNumberTypeFloat32] = LaunchCast<float16, float>;
-  mode_map[kNumberTypeFloat16][kNumberTypeFloat64] = LaunchCast<float16, double>;
-  mode_map[kNumberTypeFloat16][kNumberTypeInt8] = LaunchCast<float16, int8_t>;
-  mode_map[kNumberTypeFloat16][kNumberTypeInt16] = LaunchCast<float16, int16_t>;
-  mode_map[kNumberTypeFloat16][kNumberTypeInt32] = LaunchCast<float16, int32_t>;
-  mode_map[kNumberTypeFloat16][kNumberTypeInt64] = LaunchCast<float16, int64_t>;
-  mode_map[kNumberTypeFloat16][kNumberTypeUInt8] = LaunchCast<float16, uint8_t>;
-  mode_map[kNumberTypeFloat16][kNumberTypeUInt16] = LaunchCast<float16, uint16_t>;
-  mode_map[kNumberTypeFloat16][kNumberTypeUInt32] = LaunchCast<float16, uint32_t>;
-  mode_map[kNumberTypeFloat16][kNumberTypeUInt64] = LaunchCast<float16, uint64_t>;
-  mode_map[kNumberTypeFloat16][kNumberTypeBool] = LaunchCast<float16, bool>;
-
-  mode_map[kNumberTypeFloat32][kNumberTypeFloat16] = LaunchCast<float, float16>;
-  mode_map[kNumberTypeFloat32][kNumberTypeFloat32] = LaunchCast<float, float>;
-  mode_map[kNumberTypeFloat32][kNumberTypeFloat64] = LaunchCast<float, double>;
-  mode_map[kNumberTypeFloat32][kNumberTypeInt8] = LaunchCast<float, int8_t>;
-  mode_map[kNumberTypeFloat32][kNumberTypeInt16] = LaunchCast<float, int16_t>;
-  mode_map[kNumberTypeFloat32][kNumberTypeInt32] = LaunchCast<float, int32_t>;
-  mode_map[kNumberTypeFloat32][kNumberTypeInt64] = LaunchCast<float, int64_t>;
-  mode_map[kNumberTypeFloat32][kNumberTypeUInt8] = LaunchCast<float, uint8_t>;
-  mode_map[kNumberTypeFloat32][kNumberTypeUInt16] = LaunchCast<float, uint16_t>;
-  mode_map[kNumberTypeFloat32][kNumberTypeUInt32] = LaunchCast<float, uint32_t>;
-  mode_map[kNumberTypeFloat32][kNumberTypeUInt64] = LaunchCast<float, uint64_t>;
-  mode_map[kNumberTypeFloat32][kNumberTypeBool] = LaunchCast<float, bool>;
-
-  mode_map[kNumberTypeFloat64][kNumberTypeFloat16] = LaunchCast<double, float16>;
-  mode_map[kNumberTypeFloat64][kNumberTypeFloat32] = LaunchCast<double, float>;
-  mode_map[kNumberTypeFloat64][kNumberTypeFloat64] = LaunchCast<double, double>;
-  mode_map[kNumberTypeFloat64][kNumberTypeInt8] = LaunchCast<double, int8_t>;
-  mode_map[kNumberTypeFloat64][kNumberTypeInt16] = LaunchCast<double, int16_t>;
-  mode_map[kNumberTypeFloat64][kNumberTypeInt32] = LaunchCast<double, int32_t>;
-  mode_map[kNumberTypeFloat64][kNumberTypeInt64] = LaunchCast<double, int64_t>;
-  mode_map[kNumberTypeFloat64][kNumberTypeUInt8] = LaunchCast<double, uint8_t>;
-  mode_map[kNumberTypeFloat64][kNumberTypeUInt16] = LaunchCast<double, uint16_t>;
-  mode_map[kNumberTypeFloat64][kNumberTypeUInt32] = LaunchCast<double, uint32_t>;
-  mode_map[kNumberTypeFloat64][kNumberTypeUInt64] = LaunchCast<double, uint64_t>;
-  mode_map[kNumberTypeFloat64][kNumberTypeBool] = LaunchCast<double, bool>;
-
-  mode_map[kNumberTypeInt8][kNumberTypeFloat16] = LaunchCast<int8_t, float16>;
-  mode_map[kNumberTypeInt8][kNumberTypeFloat32] = LaunchCast<int8_t, float>;
-  mode_map[kNumberTypeInt8][kNumberTypeFloat64] = LaunchCast<int8_t, double>;
-  mode_map[kNumberTypeInt8][kNumberTypeInt8] = LaunchCast<int8_t, int8_t>;
-  mode_map[kNumberTypeInt8][kNumberTypeInt16] = LaunchCast<int8_t, int16_t>;
-  mode_map[kNumberTypeInt8][kNumberTypeInt32] = LaunchCast<int8_t, int32_t>;
-  mode_map[kNumberTypeInt8][kNumberTypeInt64] = LaunchCast<int8_t, int64_t>;
-  mode_map[kNumberTypeInt8][kNumberTypeUInt8] = LaunchCast<int8_t, uint8_t>;
-  mode_map[kNumberTypeInt8][kNumberTypeUInt16] = LaunchCast<int8_t, uint16_t>;
-  mode_map[kNumberTypeInt8][kNumberTypeUInt32] = LaunchCast<int8_t, uint32_t>;
-  mode_map[kNumberTypeInt8][kNumberTypeUInt64] = LaunchCast<int8_t, uint64_t>;
-  mode_map[kNumberTypeInt8][kNumberTypeBool] = LaunchCast<int8_t, bool>;
-
-  mode_map[kNumberTypeInt16][kNumberTypeFloat16] = LaunchCast<int16_t, float16>;
-  mode_map[kNumberTypeInt16][kNumberTypeFloat32] = LaunchCast<int16_t, float>;
-  mode_map[kNumberTypeInt16][kNumberTypeFloat64] = LaunchCast<int16_t, double>;
-  mode_map[kNumberTypeInt16][kNumberTypeInt8] = LaunchCast<int16_t, int8_t>;
-  mode_map[kNumberTypeInt16][kNumberTypeInt16] = LaunchCast<int16_t, int16_t>;
-  mode_map[kNumberTypeInt16][kNumberTypeInt32] = LaunchCast<int16_t, int32_t>;
-  mode_map[kNumberTypeInt16][kNumberTypeInt64] = LaunchCast<int16_t, int64_t>;
-  mode_map[kNumberTypeInt16][kNumberTypeUInt8] = LaunchCast<int16_t, uint8_t>;
-  mode_map[kNumberTypeInt16][kNumberTypeUInt16] = LaunchCast<int16_t, uint16_t>;
-  mode_map[kNumberTypeInt16][kNumberTypeUInt32] = LaunchCast<int16_t, uint32_t>;
-  mode_map[kNumberTypeInt16][kNumberTypeUInt64] = LaunchCast<int16_t, uint64_t>;
-  mode_map[kNumberTypeInt16][kNumberTypeBool] = LaunchCast<int16_t, bool>;
-
-  mode_map[kNumberTypeInt32][kNumberTypeFloat16] = LaunchCast<int32_t, float16>;
-  mode_map[kNumberTypeInt32][kNumberTypeFloat32] = LaunchCast<int32_t, float>;
-  mode_map[kNumberTypeInt32][kNumberTypeFloat64] = LaunchCast<int32_t, double>;
-  mode_map[kNumberTypeInt32][kNumberTypeInt8] = LaunchCast<int32_t, int8_t>;
-  mode_map[kNumberTypeInt32][kNumberTypeInt16] = LaunchCast<int32_t, int16_t>;
-  mode_map[kNumberTypeInt32][kNumberTypeInt32] = LaunchCast<int32_t, int32_t>;
-  mode_map[kNumberTypeInt32][kNumberTypeInt64] = LaunchCast<int32_t, int64_t>;
-  mode_map[kNumberTypeInt32][kNumberTypeUInt8] = LaunchCast<int32_t, uint8_t>;
-  mode_map[kNumberTypeInt32][kNumberTypeUInt16] = LaunchCast<int32_t, uint16_t>;
-  mode_map[kNumberTypeInt32][kNumberTypeUInt32] = LaunchCast<int32_t, uint32_t>;
-  mode_map[kNumberTypeInt32][kNumberTypeUInt64] = LaunchCast<int32_t, uint64_t>;
-  mode_map[kNumberTypeInt32][kNumberTypeBool] = LaunchCast<int32_t, bool>;
-
-  mode_map[kNumberTypeInt64][kNumberTypeFloat16] = LaunchCast<int64_t, float16>;
-  mode_map[kNumberTypeInt64][kNumberTypeFloat32] = LaunchCast<int64_t, float>;
-  mode_map[kNumberTypeInt64][kNumberTypeFloat64] = LaunchCast<int64_t, double>;
-  mode_map[kNumberTypeInt64][kNumberTypeInt8] = LaunchCast<int64_t, int8_t>;
-  mode_map[kNumberTypeInt64][kNumberTypeInt16] = LaunchCast<int64_t, int16_t>;
-  mode_map[kNumberTypeInt64][kNumberTypeInt32] = LaunchCast<int64_t, int32_t>;
-  mode_map[kNumberTypeInt64][kNumberTypeInt64] = LaunchCast<int64_t, int64_t>;
-  mode_map[kNumberTypeInt64][kNumberTypeUInt8] = LaunchCast<int64_t, uint8_t>;
-  mode_map[kNumberTypeInt64][kNumberTypeUInt16] = LaunchCast<int64_t, uint16_t>;
-  mode_map[kNumberTypeInt64][kNumberTypeUInt32] = LaunchCast<int64_t, uint32_t>;
-  mode_map[kNumberTypeInt64][kNumberTypeUInt64] = LaunchCast<int64_t, uint64_t>;
-  mode_map[kNumberTypeInt64][kNumberTypeBool] = LaunchCast<int64_t, bool>;
-
-  mode_map[kNumberTypeUInt8][kNumberTypeFloat16] = LaunchCast<uint8_t, float16>;
-  mode_map[kNumberTypeUInt8][kNumberTypeFloat32] = LaunchCast<uint8_t, float>;
-  mode_map[kNumberTypeUInt8][kNumberTypeFloat64] = LaunchCast<uint8_t, double>;
-  mode_map[kNumberTypeUInt8][kNumberTypeInt8] = LaunchCast<uint8_t, int8_t>;
-  mode_map[kNumberTypeUInt8][kNumberTypeInt16] = LaunchCast<uint8_t, int16_t>;
-  mode_map[kNumberTypeUInt8][kNumberTypeInt32] = LaunchCast<uint8_t, int32_t>;
-  mode_map[kNumberTypeUInt8][kNumberTypeInt64] = LaunchCast<uint8_t, int64_t>;
-  mode_map[kNumberTypeUInt8][kNumberTypeUInt8] = LaunchCast<uint8_t, uint8_t>;
-  mode_map[kNumberTypeUInt8][kNumberTypeUInt16] = LaunchCast<uint8_t, uint16_t>;
-  mode_map[kNumberTypeUInt8][kNumberTypeUInt32] = LaunchCast<uint8_t, uint32_t>;
-  mode_map[kNumberTypeUInt8][kNumberTypeUInt64] = LaunchCast<uint8_t, uint64_t>;
-  mode_map[kNumberTypeUInt8][kNumberTypeBool] = LaunchCast<uint8_t, bool>;
-
-  mode_map[kNumberTypeUInt16][kNumberTypeFloat16] = LaunchCast<uint16_t, float16>;
-  mode_map[kNumberTypeUInt16][kNumberTypeFloat32] = LaunchCast<uint16_t, float>;
-  mode_map[kNumberTypeUInt16][kNumberTypeFloat64] = LaunchCast<uint16_t, double>;
-  mode_map[kNumberTypeUInt16][kNumberTypeInt8] = LaunchCast<uint16_t, int8_t>;
-  mode_map[kNumberTypeUInt16][kNumberTypeInt16] = LaunchCast<uint16_t, int16_t>;
-  mode_map[kNumberTypeUInt16][kNumberTypeInt32] = LaunchCast<uint16_t, int32_t>;
-  mode_map[kNumberTypeUInt16][kNumberTypeInt64] = LaunchCast<uint16_t, int64_t>;
-  mode_map[kNumberTypeUInt16][kNumberTypeUInt8] = LaunchCast<uint16_t, uint8_t>;
-  mode_map[kNumberTypeUInt16][kNumberTypeUInt16] = LaunchCast<uint16_t, uint16_t>;
-  mode_map[kNumberTypeUInt16][kNumberTypeUInt32] = LaunchCast<uint16_t, uint32_t>;
-  mode_map[kNumberTypeUInt16][kNumberTypeUInt64] = LaunchCast<uint16_t, uint64_t>;
-  mode_map[kNumberTypeUInt16][kNumberTypeBool] = LaunchCast<uint16_t, bool>;
-
-  mode_map[kNumberTypeUInt32][kNumberTypeFloat16] = LaunchCast<uint32_t, float16>;
-  mode_map[kNumberTypeUInt32][kNumberTypeFloat32] = LaunchCast<uint32_t, float>;
-  mode_map[kNumberTypeUInt32][kNumberTypeFloat64] = LaunchCast<uint32_t, double>;
-  mode_map[kNumberTypeUInt32][kNumberTypeInt8] = LaunchCast<uint32_t, int8_t>;
-  mode_map[kNumberTypeUInt32][kNumberTypeInt16] = LaunchCast<uint32_t, int16_t>;
-  mode_map[kNumberTypeUInt32][kNumberTypeInt32] = LaunchCast<uint32_t, int32_t>;
-  mode_map[kNumberTypeUInt32][kNumberTypeInt64] = LaunchCast<uint32_t, int64_t>;
-  mode_map[kNumberTypeUInt32][kNumberTypeUInt8] = LaunchCast<uint32_t, uint8_t>;
-  mode_map[kNumberTypeUInt32][kNumberTypeUInt16] = LaunchCast<uint32_t, uint16_t>;
-  mode_map[kNumberTypeUInt32][kNumberTypeUInt32] = LaunchCast<uint32_t, uint32_t>;
-  mode_map[kNumberTypeUInt32][kNumberTypeUInt64] = LaunchCast<uint32_t, uint64_t>;
-  mode_map[kNumberTypeUInt32][kNumberTypeBool] = LaunchCast<uint32_t, bool>;
-
-  mode_map[kNumberTypeUInt64][kNumberTypeFloat16] = LaunchCast<uint64_t, float16>;
-  mode_map[kNumberTypeUInt64][kNumberTypeFloat32] = LaunchCast<uint64_t, float>;
-  mode_map[kNumberTypeUInt64][kNumberTypeFloat64] = LaunchCast<uint64_t, double>;
-  mode_map[kNumberTypeUInt64][kNumberTypeInt8] = LaunchCast<uint64_t, int8_t>;
-  mode_map[kNumberTypeUInt64][kNumberTypeInt16] = LaunchCast<uint64_t, int16_t>;
-  mode_map[kNumberTypeUInt64][kNumberTypeInt32] = LaunchCast<uint64_t, int32_t>;
-  mode_map[kNumberTypeUInt64][kNumberTypeInt64] = LaunchCast<uint64_t, int64_t>;
-  mode_map[kNumberTypeUInt64][kNumberTypeUInt8] = LaunchCast<uint64_t, uint8_t>;
-  mode_map[kNumberTypeUInt64][kNumberTypeUInt16] = LaunchCast<uint64_t, uint16_t>;
-  mode_map[kNumberTypeUInt64][kNumberTypeUInt32] = LaunchCast<uint64_t, uint32_t>;
-  mode_map[kNumberTypeUInt64][kNumberTypeUInt64] = LaunchCast<uint64_t, uint64_t>;
-  mode_map[kNumberTypeUInt64][kNumberTypeBool] = LaunchCast<uint64_t, bool>;
+template <typename S, typename T>
+bool CastCPUKernel<S, T>::Launch(const std::vector<kernel::AddressPtr> &inputs,
+                                 const std::vector<kernel::AddressPtr> & /*workspace*/,
+                                 const std::vector<kernel::AddressPtr> &outputs) {
+  S *input = reinterpret_cast<S *>(inputs[0]->addr);
+  T *output = reinterpret_cast<T *>(outputs[0]->addr);
+  MS_LOG(DEBUG) << "Type source: " << typeid(S).name() << "; target: " << typeid(T).name();
 
-  mode_map[source_dtype][target_dtype](inputs, outputs);
+  size_t lens = outputs[0]->size > 0 ? static_cast<size_t>(outputs[0]->size / sizeof(T)) : 1;
+  Cast<S, T>(input, output, lens);
   return true;
 }
 }  // namespace kernel
diff --git a/mindspore/ccsrc/backend/kernel_compiler/cpu/cast_cpu_kernel.h b/mindspore/ccsrc/backend/kernel_compiler/cpu/cast_cpu_kernel.h
index 7c68e8f612..d4ea77f095 100644
--- a/mindspore/ccsrc/backend/kernel_compiler/cpu/cast_cpu_kernel.h
+++ b/mindspore/ccsrc/backend/kernel_compiler/cpu/cast_cpu_kernel.h
@@ -23,6 +23,7 @@
 
 namespace mindspore {
 namespace kernel {
+template <typename S, typename T>
 class CastCPUKernel : public CPUKernel {
  public:
   CastCPUKernel() = default;
@@ -38,161 +39,305 @@ class CastCPUKernel : public CPUKernel {
   TypeId target_dtype{kTypeUnknown};
 };
 
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeFloat16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeFloat32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeFloat64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeBool), CastCPUKernel);
-
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeFloat16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeFloat32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeFloat64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeBool), CastCPUKernel);
-
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeFloat16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeFloat32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeFloat64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeBool), CastCPUKernel);
-
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeFloat16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeFloat32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeFloat64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeBool), CastCPUKernel);
-
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeFloat16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeFloat32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeFloat64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeBool), CastCPUKernel);
-
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeFloat16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeFloat32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeFloat64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeBool), CastCPUKernel);
-
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeFloat16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeFloat32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeFloat64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeBool), CastCPUKernel);
-
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeFloat16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeFloat32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeFloat64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeBool), CastCPUKernel);
-
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeFloat16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeFloat32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeFloat64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeBool), CastCPUKernel);
-
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeFloat16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeFloat32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeFloat64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeBool), CastCPUKernel);
-
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeFloat16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeFloat32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeFloat64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeBool), CastCPUKernel);
-
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeFloat16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeFloat32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeFloat64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel);
-MS_REG_CPU_KERNEL(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeBool), CastCPUKernel);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeFloat16), CastCPUKernel,
+                      bool, float16);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeFloat32), CastCPUKernel,
+                      bool, float);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeFloat64), CastCPUKernel,
+                      bool, double);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeInt8), CastCPUKernel,
+                      bool, int8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeInt16), CastCPUKernel,
+                      bool, int16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeInt32), CastCPUKernel,
+                      bool, int32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeInt64), CastCPUKernel,
+                      bool, int64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel,
+                      bool, uint8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel,
+                      bool, uint16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel,
+                      bool, uint32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel,
+                      bool, uint64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeBool).AddOutputAttr(kNumberTypeBool), CastCPUKernel,
+                      bool, bool);
+
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeFloat16),
+                      CastCPUKernel, float16, float16);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeFloat32),
+                      CastCPUKernel, float16, float);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeFloat64),
+                      CastCPUKernel, float16, double);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeInt8), CastCPUKernel,
+                      float16, int8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeInt16),
+                      CastCPUKernel, float16, int16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeInt32),
+                      CastCPUKernel, float16, int32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeInt64),
+                      CastCPUKernel, float16, int64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeUInt8),
+                      CastCPUKernel, float16, uint8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeUInt16),
+                      CastCPUKernel, float16, uint16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeUInt32),
+                      CastCPUKernel, float16, uint32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeUInt64),
+                      CastCPUKernel, float16, uint64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeBool), CastCPUKernel,
+                      float16, bool);
+
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeFloat16),
+                      CastCPUKernel, float, float16);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeFloat32),
+                      CastCPUKernel, float, float);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeFloat64),
+                      CastCPUKernel, float, double);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeInt8), CastCPUKernel,
+                      float, int8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeInt16),
+                      CastCPUKernel, float, int16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeInt32),
+                      CastCPUKernel, float, int32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeInt64),
+                      CastCPUKernel, float, int64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeUInt8),
+                      CastCPUKernel, float, uint8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeUInt16),
+                      CastCPUKernel, float, uint16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeUInt32),
+                      CastCPUKernel, float, uint32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeUInt64),
+                      CastCPUKernel, float, uint64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeBool), CastCPUKernel,
+                      float, bool);
+
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeFloat16),
+                      CastCPUKernel, double, float16);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeFloat32),
+                      CastCPUKernel, double, float);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeFloat64),
+                      CastCPUKernel, double, double);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeInt8), CastCPUKernel,
+                      double, int8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeInt16),
+                      CastCPUKernel, double, int16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeInt32),
+                      CastCPUKernel, double, int32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeInt64),
+                      CastCPUKernel, double, int64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeUInt8),
+                      CastCPUKernel, double, uint8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeUInt16),
+                      CastCPUKernel, double, uint16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeUInt32),
+                      CastCPUKernel, double, uint32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeUInt64),
+                      CastCPUKernel, double, uint64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeBool), CastCPUKernel,
+                      double, bool);
+
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeFloat16), CastCPUKernel,
+                      int8_t, float16);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeFloat32), CastCPUKernel,
+                      int8_t, float);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeFloat64), CastCPUKernel,
+                      int8_t, double);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeInt8), CastCPUKernel,
+                      int8_t, int8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeInt16), CastCPUKernel,
+                      int8_t, int16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeInt32), CastCPUKernel,
+                      int8_t, int32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeInt64), CastCPUKernel,
+                      int8_t, int64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel,
+                      int8_t, uint8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel,
+                      int8_t, uint16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel,
+                      int8_t, uint32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel,
+                      int8_t, uint64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt8).AddOutputAttr(kNumberTypeBool), CastCPUKernel,
+                      int8_t, bool);
+
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeFloat16),
+                      CastCPUKernel, int16_t, float16);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeFloat32),
+                      CastCPUKernel, int16_t, float);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeFloat64),
+                      CastCPUKernel, int16_t, double);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeInt8), CastCPUKernel,
+                      int16_t, int8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeInt16), CastCPUKernel,
+                      int16_t, int16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeInt32), CastCPUKernel,
+                      int16_t, int32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeInt64), CastCPUKernel,
+                      int16_t, int64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel,
+                      int16_t, uint8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel,
+                      int16_t, uint16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel,
+                      int16_t, uint32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel,
+                      int16_t, uint64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt16).AddOutputAttr(kNumberTypeBool), CastCPUKernel,
+                      int16_t, bool);
+
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeFloat16),
+                      CastCPUKernel, int32_t, float16);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeFloat32),
+                      CastCPUKernel, int32_t, float);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeFloat64),
+                      CastCPUKernel, int32_t, double);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt8), CastCPUKernel,
+                      int32_t, int8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt16), CastCPUKernel,
+                      int32_t, int16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt32), CastCPUKernel,
+                      int32_t, int32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt64), CastCPUKernel,
+                      int32_t, int64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel,
+                      int32_t, uint8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel,
+                      int32_t, uint16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel,
+                      int32_t, uint32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel,
+                      int32_t, uint64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeBool), CastCPUKernel,
+                      int32_t, bool);
+
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeFloat16),
+                      CastCPUKernel, int64_t, float16);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeFloat32),
+                      CastCPUKernel, int64_t, float);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeFloat64),
+                      CastCPUKernel, int64_t, double);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeInt8), CastCPUKernel,
+                      int64_t, int8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeInt16), CastCPUKernel,
+                      int64_t, int16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeInt32), CastCPUKernel,
+                      int64_t, int32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeInt64), CastCPUKernel,
+                      int64_t, int64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel,
+                      int64_t, uint8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel,
+                      int64_t, uint16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel,
+                      int64_t, uint32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel,
+                      int64_t, uint64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeBool), CastCPUKernel,
+                      int64_t, bool);
+
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeFloat16),
+                      CastCPUKernel, uint8_t, float16);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeFloat32),
+                      CastCPUKernel, uint8_t, float);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeFloat64),
+                      CastCPUKernel, uint8_t, double);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeInt8), CastCPUKernel,
+                      uint8_t, int8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeInt16), CastCPUKernel,
+                      uint8_t, int16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeInt32), CastCPUKernel,
+                      uint8_t, int32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeInt64), CastCPUKernel,
+                      uint8_t, int64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel,
+                      uint8_t, uint8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeUInt16), CastCPUKernel,
+                      uint8_t, uint16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeUInt32), CastCPUKernel,
+                      uint8_t, uint32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeUInt64), CastCPUKernel,
+                      uint8_t, uint64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt8).AddOutputAttr(kNumberTypeBool), CastCPUKernel,
+                      uint8_t, bool);
+
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeFloat16),
+                      CastCPUKernel, uint16_t, float16);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeFloat32),
+                      CastCPUKernel, uint16_t, float);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeFloat64),
+                      CastCPUKernel, uint16_t, double);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeInt8), CastCPUKernel,
+                      uint16_t, int8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeInt16), CastCPUKernel,
+                      uint16_t, int16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeInt32), CastCPUKernel,
+                      uint16_t, int32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeInt64), CastCPUKernel,
+                      uint16_t, int64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel,
+                      uint16_t, uint8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeUInt16),
+                      CastCPUKernel, uint16_t, uint16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeUInt32),
+                      CastCPUKernel, uint16_t, uint32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeUInt64),
+                      CastCPUKernel, uint16_t, uint64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt16).AddOutputAttr(kNumberTypeBool), CastCPUKernel,
+                      uint16_t, bool);
+
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeFloat16),
+                      CastCPUKernel, uint32_t, float16);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeFloat32),
+                      CastCPUKernel, uint32_t, float);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeFloat64),
+                      CastCPUKernel, uint32_t, double);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeInt8), CastCPUKernel,
+                      uint32_t, int8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeInt16), CastCPUKernel,
+                      uint32_t, int16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeInt32), CastCPUKernel,
+                      uint32_t, int32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeInt64), CastCPUKernel,
+                      uint32_t, int64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel,
+                      uint32_t, uint8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeUInt16),
+                      CastCPUKernel, uint32_t, uint16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeUInt32),
+                      CastCPUKernel, uint32_t, uint32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeUInt64),
+                      CastCPUKernel, uint32_t, uint64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeBool), CastCPUKernel,
+                      uint32_t, bool);
+
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeFloat16),
+                      CastCPUKernel, uint64_t, float16);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeFloat32),
+                      CastCPUKernel, uint64_t, float);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeFloat64),
+                      CastCPUKernel, uint64_t, double);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeInt8), CastCPUKernel,
+                      uint64_t, int8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeInt16), CastCPUKernel,
+                      uint64_t, int16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeInt32), CastCPUKernel,
+                      uint64_t, int32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeInt64), CastCPUKernel,
+                      uint64_t, int64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeUInt8), CastCPUKernel,
+                      uint64_t, uint8_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeUInt16),
+                      CastCPUKernel, uint64_t, uint16_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeUInt32),
+                      CastCPUKernel, uint64_t, uint32_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeUInt64),
+                      CastCPUKernel, uint64_t, uint64_t);
+MS_REG_CPU_KERNEL_T_S(Cast, KernelAttr().AddInputAttr(kNumberTypeUInt64).AddOutputAttr(kNumberTypeBool), CastCPUKernel,
+                      uint64_t, bool);
 }  // namespace kernel
 }  // namespace mindspore
 
diff --git a/mindspore/ccsrc/backend/kernel_compiler/cpu/layer_norm_cpu_kernel.cc b/mindspore/ccsrc/backend/kernel_compiler/cpu/layer_norm_cpu_kernel.cc
index 353ee5d4bd..548b23a3a0 100644
--- a/mindspore/ccsrc/backend/kernel_compiler/cpu/layer_norm_cpu_kernel.cc
+++ b/mindspore/ccsrc/backend/kernel_compiler/cpu/layer_norm_cpu_kernel.cc
@@ -14,8 +14,11 @@
  * limitations under the License.
  */
 
+#include <cmath>
 #include "backend/kernel_compiler/cpu/layer_norm_cpu_kernel.h"
+#include "backend/kernel_compiler/common_utils.h"
 #include "runtime/device/cpu/cpu_device_address.h"
+#include "common/thread_pool.h"
 
 namespace mindspore {
 namespace kernel {
@@ -72,23 +75,43 @@ void LayerNormCPUKernel::LaunchKernel(const std::vector<AddressPtr> &inputs, con
   auto y = reinterpret_cast<T *>(outputs[0]->addr);
   auto mean = reinterpret_cast<T *>(outputs[1]->addr);
   auto var = reinterpret_cast<T *>(outputs[2]->addr);
-  for (size_t i = 0; i < block_num_; ++i) {
-    T sum = (T)0.0;
-    T square_sum = (T)0.0;
-    for (size_t j = i * block_size_; j < (i + 1) * block_size_; ++j) {
-      sum += x[j];
-      square_sum += x[j] * x[j];
-    }
-    T block_mean = sum / block_size_;
-    T block_var = square_sum / block_size_ - block_mean * block_mean;
-    for (size_t j = i * block_size_; j < (i + 1) * block_size_; ++j) {
-      auto param_shift = j % param_num_;
-      y[j] = (x[j] - block_mean) / (T)std::sqrt(static_cast<double>(block_var) + eps_) * gamma[param_shift] +
-             beta[param_shift];
+  size_t thread_num = common::ThreadPool::GetInstance().GetSyncRunThreadNum();
+  if (block_num_ < thread_num) {
+    thread_num = block_num_;
+  }
+  std::vector<common::Task> tasks;
+  tasks.reserve(thread_num);
+  auto task = [&](size_t start, size_t end) {
+    for (size_t c = 0; c < ceil(static_cast<double>(block_num_) / thread_num); ++c) {
+      if (c * thread_num + start >= block_num_) {
+        continue;
+      }
+      size_t i = c * thread_num + start;
+      T sum = (T)0.0;
+      T square_sum = (T)0.0;
+      for (size_t j = i * block_size_; j < (i + 1) * block_size_; ++j) {
+        sum += x[j];
+        square_sum += x[j] * x[j];
+      }
+      T block_mean = sum / block_size_;
+      T block_var = square_sum / block_size_ - block_mean * block_mean;
+      for (size_t j = i * block_size_; j < (i + 1) * block_size_; ++j) {
+        auto param_shift = j % param_num_;
+        y[j] = (x[j] - block_mean) / (T)std::sqrt(static_cast<double>(block_var) + eps_) * gamma[param_shift] +
+               beta[param_shift];
+      }
+      mean[i] = block_mean;
+      var[i] = block_var;
     }
-    mean[i] = block_mean;
-    var[i] = block_var;
+  };
+  for (size_t i = 0; i < thread_num; ++i) {
+    auto block = [&, i]() {
+      task(i, i + 1);
+      return common::SUCCESS;
+    };
+    tasks.emplace_back(block);
   }
+  common::ThreadPool::GetInstance().SyncRun(tasks);
 }
 
 void LayerNormCPUKernel::CheckParam(const CNodePtr &kernel_node) {
diff --git a/mindspore/ccsrc/backend/kernel_compiler/cpu/layer_norm_grad_cpu_kernel.cc b/mindspore/ccsrc/backend/kernel_compiler/cpu/layer_norm_grad_cpu_kernel.cc
index 63cefe0ab9..1b9f4c89c2 100644
--- a/mindspore/ccsrc/backend/kernel_compiler/cpu/layer_norm_grad_cpu_kernel.cc
+++ b/mindspore/ccsrc/backend/kernel_compiler/cpu/layer_norm_grad_cpu_kernel.cc
@@ -15,7 +15,9 @@
  */
 
 #include "backend/kernel_compiler/cpu/layer_norm_grad_cpu_kernel.h"
+#include "backend/kernel_compiler/common_utils.h"
 #include "runtime/device/cpu/cpu_device_address.h"
+#include "common/thread_pool.h"
 
 namespace mindspore {
 namespace kernel {
@@ -73,41 +75,75 @@ void LayerNormGradCPUKernel::LaunchKernel(const std::vector<AddressPtr> &inputs,
   auto dx = reinterpret_cast<T *>(outputs[0]->addr);
   auto dg = reinterpret_cast<T *>(outputs[1]->addr);
   auto db = reinterpret_cast<T *>(outputs[2]->addr);
-
-  for (size_t i = 0; i < param_num_; ++i) {
-    T dgamma = (T)0.0;
-    T dbeta = (T)0.0;
-    for (size_t j = i; j < param_size_ * param_num_; j += param_num_) {
-      auto norm_shift = static_cast<int>(j / block_size_);
-      dgamma += dy[j] * (T)std::pow(static_cast<double>(var[norm_shift]) + eps_, -0.5) * (x[j] - mean[norm_shift]);
-      dbeta += dy[j];
-    }
-    dg[i] = dgamma;
-    db[i] = dbeta;
-  }
-  for (size_t i = 0; i < block_num_; ++i) {
-    T sum1 = (T)0.0;
-    T sum2 = (T)0.0;
-    T sum3 = (T)0.0;
-    for (size_t j = i * block_size_; j < (i + 1) * block_size_; ++j) {
-      auto param_shift = j % param_num_;
-      auto norm_shift = static_cast<int>(j / block_size_);
-      auto dxm = x[j] - mean[norm_shift];
-      auto dyg = dy[j] * gamma[param_shift];
-      sum1 += (T)(-0.5) * dyg * dxm * (T)std::pow(static_cast<double>(var[norm_shift]) + eps_, -1.5);
-      sum2 += dyg;
-      sum3 += (T)(-2.0) * dxm;
+  size_t thread_num = common::ThreadPool::GetInstance().GetSyncRunThreadNum();
+  auto thread_num1 = param_num_ < thread_num ? param_num_ : thread_num;
+  std::vector<common::Task> tasks1;
+  tasks1.reserve(thread_num1);
+  auto thread_num2 = block_num_ < thread_num ? block_num_ : thread_num;
+  std::vector<common::Task> tasks2;
+  tasks2.reserve(thread_num2);
+  auto task1 = [&](size_t start, size_t end) {
+    for (size_t c = 0; c < ceil(static_cast<double>(param_num_) / thread_num1); ++c) {
+      if (c * thread_num1 + start >= param_num_) {
+        continue;
+      }
+      size_t param_index = c * thread_num1 + start;
+      T dgamma = (T)0.0;
+      T dbeta = (T)0.0;
+      for (size_t j = param_index; j < param_size_ * param_num_; j += param_num_) {
+        auto norm_shift = static_cast<int>(j / block_size_);
+        dgamma += dy[j] * (T)std::pow(static_cast<double>(var[norm_shift]) + eps_, -0.5) * (x[j] - mean[norm_shift]);
+        dbeta += dy[j];
+      }
+      dg[param_index] = dgamma;
+      db[param_index] = dbeta;
     }
-    for (size_t j = i * block_size_; j < (i + 1) * block_size_; ++j) {
-      auto param_shift = j % param_num_;
-      auto norm_shift = static_cast<int>(j / block_size_);
-      auto var_sqrt = (T)std::pow(static_cast<double>(var[norm_shift]) + eps_, -0.5);
-      auto dx1 = dy[j] * gamma[param_shift] * var_sqrt;
-      auto dx2 = sum1 * (T)2.0 / block_size_ * (x[j] - mean[norm_shift]);
-      auto dx3 = ((T)(-1.0) * var_sqrt * sum2 + ((T)1.0 / block_size_) * sum1 * sum3) * ((T)1.0 / block_size_);
-      dx[j] = dx1 + dx2 + dx3;
+  };
+  auto task2 = [&](size_t start, size_t end) {
+    for (size_t c = 0; c < ceil(static_cast<double>(block_num_) / thread_num2); ++c) {
+      if (c * thread_num2 + start >= block_num_) {
+        continue;
+      }
+      size_t block_index = c * thread_num2 + start;
+      T sum1 = (T)0.0;
+      T sum2 = (T)0.0;
+      T sum3 = (T)0.0;
+      for (size_t j = block_index * block_size_; j < (block_index + 1) * block_size_; ++j) {
+        auto param_shift = j % param_num_;
+        auto norm_shift = static_cast<int>(j / block_size_);
+        auto dxm = x[j] - mean[norm_shift];
+        auto dyg = dy[j] * gamma[param_shift];
+        sum1 += (T)(-0.5) * dyg * dxm * (T)std::pow(static_cast<double>(var[norm_shift]) + eps_, -1.5);
+        sum2 += dyg;
+        sum3 += (T)(-2.0) * dxm;
+      }
+      for (size_t j = block_index * block_size_; j < (block_index + 1) * block_size_; ++j) {
+        auto param_shift = j % param_num_;
+        auto norm_shift = static_cast<int>(j / block_size_);
+        auto var_sqrt = (T)std::pow(static_cast<double>(var[norm_shift]) + eps_, -0.5);
+        auto dx1 = dy[j] * gamma[param_shift] * var_sqrt;
+        auto dx2 = sum1 * (T)2.0 / block_size_ * (x[j] - mean[norm_shift]);
+        auto dx3 = ((T)(-1.0) * var_sqrt * sum2 + ((T)1.0 / block_size_) * sum1 * sum3) * ((T)1.0 / block_size_);
+        dx[j] = dx1 + dx2 + dx3;
+      }
     }
+  };
+  for (size_t i = 0; i < thread_num1; ++i) {
+    auto block = [&, i]() {
+      task1(i, i + 1);
+      return common::SUCCESS;
+    };
+    tasks1.emplace_back(block);
+  }
+  common::ThreadPool::GetInstance().SyncRun(tasks1);
+  for (size_t i = 0; i < thread_num2; ++i) {
+    auto block = [&, i]() {
+      task2(i, i + 1);
+      return common::SUCCESS;
+    };
+    tasks2.emplace_back(block);
   }
+  common::ThreadPool::GetInstance().SyncRun(tasks2);
 }
 
 void LayerNormGradCPUKernel::CheckParam(const CNodePtr &kernel_node) {
diff --git a/mindspore/ccsrc/backend/kernel_compiler/cpu/unsorted_segment_sum_cpu_kernel.cc b/mindspore/ccsrc/backend/kernel_compiler/cpu/unsorted_segment_sum_cpu_kernel.cc
index a6505b0927..d9afeb211c 100644
--- a/mindspore/ccsrc/backend/kernel_compiler/cpu/unsorted_segment_sum_cpu_kernel.cc
+++ b/mindspore/ccsrc/backend/kernel_compiler/cpu/unsorted_segment_sum_cpu_kernel.cc
@@ -16,6 +16,7 @@
 
 #include "backend/kernel_compiler/cpu/unsorted_segment_sum_cpu_kernel.h"
 #include <string>
+#include <cmath>
 #include "runtime/device/cpu/cpu_device_address.h"
 #include "common/thread_pool.h"
 
@@ -78,17 +79,37 @@ bool UnsortedSegmentSumCPUKernel::LaunchKernel(const std::vector<AddressPtr> &in
     MS_LOG(ERROR) << "Output buff memset fail. ret:" << ret;
     return false;
   }
-  for (size_t i = 0; i < unit_num_; ++i) {
-    size_t j = i / input_dim1_;
-    size_t k = i % input_dim1_;
+  size_t thread_num = common::ThreadPool::GetInstance().GetSyncRunThreadNum();
+  if (unit_num_ < thread_num) {
+    thread_num = unit_num_;
+  }
+  std::vector<common::Task> tasks;
+  tasks.reserve(thread_num);
+  auto task = [&](size_t start, size_t end) {
+    for (size_t c = 0; c < ceil(static_cast<double>(unit_num_) / thread_num); ++c) {
+      if (c * thread_num + start >= unit_num_) {
+        continue;
+      }
+      size_t i = c * thread_num + start;
+      size_t j = i / input_dim1_;
+      size_t k = i % input_dim1_;
 
-    T index = indices_addr[j];
-    if (index < 0 || index >= SizeToInt(output_dim0_)) {
-      continue;
+      T index = indices_addr[j];
+      if (index < 0 || index >= SizeToInt(output_dim0_)) {
+        continue;
+      }
+      size_t output_index = index * output_dim1_ + k;
+      output_addr[output_index] += input_addr[i];
     }
-    size_t output_index = index * output_dim1_ + k;
-    output_addr[output_index] += input_addr[i];
+  };
+  for (size_t t = 0; t < thread_num; ++t) {
+    auto block = [&, t]() {
+      task(t, t + 1);
+      return common::SUCCESS;
+    };
+    tasks.emplace_back(block);
   }
+  common::ThreadPool::GetInstance().SyncRun(tasks);
   return true;
 }
 }  // namespace kernel