diff --git a/atvc/examples/add/add.cpp b/atvc/examples/add/add.cpp
index 4e0ef39cb028c7f3db06f6404c8c76022883daa5..04ea4a22766471183cb00a790918d83d7c443c16 100644
--- a/atvc/examples/add/add.cpp
+++ b/atvc/examples/add/add.cpp
@@ -30,7 +30,8 @@ static constexpr float REL_TOL = 1e-3f;
 static constexpr float ABS_TOL = 1e-5f;
 
 // 判断两个浮点数是否足够接近
-bool IsClose(float a, float b) {
+bool IsClose(float a, float b)
+{
     const float eps = 1e-40f; // 防止分母为零
     float diff = std::abs(a - b);
     return (diff <= ABS_TOL) || (diff <= REL_TOL * std::max(std::abs(a), std::abs(b) + eps));
@@ -49,7 +50,7 @@ struct AddComputeFunc {
         b                   : 参与运算的输入
         c                   : 参与运算的输出
     */
-    template<typename T> 
+    template<typename T>
     // 重载operator，提供给算子模板类调用
     __aicore__ inline void operator()(AscendC::LocalTensor<T> a, AscendC::LocalTensor<T> b, AscendC::LocalTensor<T> c) {
         AscendC::Add(c, a, b, c.GetSize()); // 开发调用AscendC Api自行实现计算逻辑, 通过c.GetSize()获取单次计算的元素数量
@@ -67,7 +68,8 @@ template<class Traits>
 __global__ __aicore__ void AddCustom(GM_ADDR a, GM_ADDR b, GM_ADDR c, GM_ADDR param)
 {
     KERNEL_TASK_TYPE_DEFAULT(KERNEL_TYPE_AIV_ONLY);
-    auto op = ATVC::Kernel::EleWiseOpTemplate<AddComputeFunc<Traits>>();  // 将AddComputeFunc仿函数作为模板参数传入，实例化EleWiseOpTemplate模板类
+    // 将AddComputeFunc仿函数作为模板参数传入，实例化EleWiseOpTemplate模板类
+    auto op = ATVC::Kernel::EleWiseOpTemplate<AddComputeFunc<Traits>>();
     op.Run(a, b, c, param); // 按照输入、输出、param的顺序传入Run函数，实现GM->GM的数据计算
 }
 
@@ -126,7 +128,9 @@ int main()
 
     auto elementParamSize = sizeof(param);
     CHECK_ACL(aclrtMalloc((void**)&paramDevice, elementParamSize, ACL_MEM_MALLOC_HUGE_FIRST));
-    CHECK_ACL(aclrtMemcpy(paramDevice, elementParamSize, reinterpret_cast<uint8_t*>(&param), elementParamSize, ACL_MEMCPY_HOST_TO_DEVICE));
+    CHECK_ACL(aclrtMemcpy(paramDevice, elementParamSize,
+                          reinterpret_cast<uint8_t *>(&param), elementParamSize,
+                          ACL_MEMCPY_HOST_TO_DEVICE));
     uint32_t blockNum = param.tilingData.blockNum;
     // 调用核函数
     AddCustom<ADD_OPTRAITS><<<blockNum, nullptr, stream>>>(xDevice, yDevice, zDevice, paramDevice);
@@ -148,7 +152,9 @@ int main()
 
     for (int32_t i = 0; i < eleNum; i++) {
         if (!IsClose(golden[i], outputZ[i])) {
-            printf("Accuracy verification failed! The expected value of element in index [%d] is %f, but actual value is %f.\n", i, golden[i], outputZ[i]);
+            printf("Accuracy verification failed! The expected value of element "
+                   "in index [%d] is %f, but actual value is %f.\n",
+                   i, golden[i], outputZ[i]);
             return -1;
         }
     }
diff --git a/atvc/examples/add_with_scalar/add_with_scalar.cpp b/atvc/examples/add_with_scalar/add_with_scalar.cpp
index b0eeeedc89b8d7a2e1cdc4432976592857d228a0..c7aa38993b6f87bfae09e4f5928b62034857ec5c 100644
--- a/atvc/examples/add_with_scalar/add_with_scalar.cpp
+++ b/atvc/examples/add_with_scalar/add_with_scalar.cpp
@@ -30,7 +30,8 @@ static constexpr float REL_TOL = 1e-3f;
 static constexpr float ABS_TOL = 1e-5f;
 
 // 判断两个浮点数是否足够接近
-bool IsClose(float a, float b) {
+bool IsClose(float a, float b)
+{
     const float eps = 1e-40f; // 防止分母为零
     float diff = std::abs(a - b);
     return (diff <= ABS_TOL) || (diff <= REL_TOL * std::max(std::abs(a), std::abs(b) + eps));
@@ -42,7 +43,7 @@ using OP_TRAITS =  ATVC::OpTraits<ATVC::OpInputs<float, float>, ATVC::OpOutputs<
 template<typename Traits>
 struct AddComputeFunc {
     template<typename T>
-    __aicore__ inline void operator()(AscendC::LocalTensor<T> a, AscendC::LocalTensor<T> b, 
+    __aicore__ inline void operator()(AscendC::LocalTensor<T> a, AscendC::LocalTensor<T> b,
         AscendC::LocalTensor<T> c, AscendC::LocalTensor<T> temp, bool conditionVal) {
             if (conditionVal) {
                 AscendC::Add(temp, a, a, c.GetSize()); // temp = a + a
@@ -66,8 +67,8 @@ template<class Traits>
 __global__ __aicore__ void AddCustom(GM_ADDR a, GM_ADDR b, GM_ADDR c, GM_ADDR param, bool conditionVal)
 {
     KERNEL_TASK_TYPE_DEFAULT(KERNEL_TYPE_AIV_ONLY);
-    auto op = ATVC::Kernel::EleWiseOpTemplate<AddComputeFunc<Traits>>(); //传入计算仿函数<In, Out, Temp>, 实例化算子
-    op.Run(a, b, c, param, conditionVal);                                //调用Run函数, 执行算子
+    auto op = ATVC::Kernel::EleWiseOpTemplate<AddComputeFunc<Traits>>(); // 传入计算仿函数<In, Out, Temp>, 实例化算子
+    op.Run(a, b, c, param, conditionVal);                                // 调用Run函数, 执行算子
 }
 
 int main()
@@ -127,9 +128,12 @@ int main()
 
     auto elementParamSize = sizeof(param);
     CHECK_ACL(aclrtMalloc((void**)&paramDevice, elementParamSize, ACL_MEM_MALLOC_HUGE_FIRST));
-    CHECK_ACL(aclrtMemcpy(paramDevice, elementParamSize, reinterpret_cast<uint8_t*>(&param), elementParamSize, ACL_MEMCPY_HOST_TO_DEVICE));
+    CHECK_ACL(aclrtMemcpy(paramDevice, elementParamSize,
+                          reinterpret_cast<uint8_t *>(&param), elementParamSize,
+                          ACL_MEMCPY_HOST_TO_DEVICE));
 
-    AddCustom<OP_TRAITS><<<param.tilingData.blockNum, nullptr, stream>>>(xDevice, yDevice, zDevice, paramDevice, conditionVal);
+    AddCustom<OP_TRAITS><<<param.tilingData.blockNum, nullptr, stream>>>(
+        xDevice, yDevice, zDevice, paramDevice, conditionVal);
 
     CHECK_ACL(aclrtSynchronizeStream(stream));
     CHECK_ACL(aclrtMemcpy(zHost, outputByteSize, zDevice, outputByteSize, ACL_MEMCPY_DEVICE_TO_HOST));
@@ -148,7 +152,9 @@ int main()
 
     for (int32_t i = 0; i < eleNum; i++) {
         if (!IsClose(golden[i], outputZ[i])) {
-            printf("Accuracy verification failed! The expected value of element in index [%d] is %f, but actual value is %f.\n", i, golden[i], outputZ[i]);
+            printf("Accuracy verification failed! The expected value of element "
+                   "in index [%d] is %f, but actual value is %f.\n",
+                   i, golden[i], outputZ[i]);
             return -1;
         }
     }
diff --git a/atvc/examples/reduce_sum/reduce_sum.cpp b/atvc/examples/reduce_sum/reduce_sum.cpp
index 81b84274cf6f133635962637a7eaaac66adf8d33..6f19ab4964668cc9cd63744c46c2541b469fe20c 100644
--- a/atvc/examples/reduce_sum/reduce_sum.cpp
+++ b/atvc/examples/reduce_sum/reduce_sum.cpp
@@ -31,7 +31,8 @@ static constexpr float REL_TOL = 1e-3f;
 static constexpr float ABS_TOL = 1e-5f;
 
 // 判断两个浮点数是否足够接近
-bool IsClose(float a, float b) {
+bool IsClose(float a, float b)
+{
     const float eps = 1e-40f; // 防止分母为零
     float diff = std::abs(a - b);
     return (diff <= ABS_TOL) || (diff <= REL_TOL * std::max(std::abs(a), std::abs(b) + eps));
@@ -67,11 +68,13 @@ void ReduceOpAdapter(uint8_t* x, uint8_t* y, ATVC::ReduceParam &param, ATVC::Red
     param.workspaceAddr = reinterpret_cast<uint64_t>(workspaceDevice);
     auto reduceParamSize = sizeof(param);
     CHECK_ACL(aclrtMalloc((void**)&paramDevice, reduceParamSize, ACL_MEM_MALLOC_HUGE_FIRST));
-    CHECK_ACL(aclrtMemcpy(paramDevice, reduceParamSize, reinterpret_cast<uint8_t*>(&param), reduceParamSize, ACL_MEMCPY_HOST_TO_DEVICE));
+    CHECK_ACL(aclrtMemcpy(paramDevice, reduceParamSize,
+                          reinterpret_cast<uint8_t *>(&param), reduceParamSize,
+                          ACL_MEMCPY_HOST_TO_DEVICE));
     // 将tiling api计算出的ReducePolicy转化为编译态参数并实例化相应的核函数
     if (policy == ATVC::REDUCE_POLICY0) {
         ReduceCustom<OpTraits, ATVC::REDUCE_POLICY0><<<param.tilingData.coreNum, nullptr, stream>>>(x, y, paramDevice);
-     }else if (policy == ATVC::REDUCE_POLICY1) {
+    } else if (policy == ATVC::REDUCE_POLICY1) {
         ReduceCustom<OpTraits, ATVC::REDUCE_POLICY1><<<param.tilingData.coreNum, nullptr, stream>>>(x, y, paramDevice);
     } else if (policy == ATVC::REDUCE_POLICY2) {
         ReduceCustom<OpTraits, ATVC::REDUCE_POLICY2><<<param.tilingData.coreNum, nullptr, stream>>>(x, y, paramDevice);
@@ -177,8 +180,10 @@ int32_t main(int32_t argc, char* argv[])
 
     for (int32_t i = 0; i < outEleNum; i++) {
         if (!IsClose(golden[i], outputY[i])) {
-            printf("Accuracy verification failed! The expected value of element in index [%d] is %f, but actual value is %f.\n", i, golden[i], outputY[i]);
-            return -1;
+            printf("Accuracy verification failed! The expected value of element "
+                   "in index [%d] is %f, but actual value is %f.\n",
+                   i, golden[i], outputY[i]);
+          return -1;
         }
     }
     printf("Accuracy verification passed.\n");
diff --git a/atvc/examples/sinh_custom/sinh_custom.cpp b/atvc/examples/sinh_custom/sinh_custom.cpp
index 5b6a31ed054dfc3f4221bd69185f3c31228d1758..b339ccbbcc8e0d741ea84d195274fc8918521277 100644
--- a/atvc/examples/sinh_custom/sinh_custom.cpp
+++ b/atvc/examples/sinh_custom/sinh_custom.cpp
@@ -31,7 +31,8 @@ static constexpr float REL_TOL = 1e-3f;
 static constexpr float ABS_TOL = 1e-5f;
 
 // 判断两个浮点数是否足够接近
-bool IsClose(float a, float b) {
+bool IsClose(float a, float b)
+{
     const float eps = 1e-40f; // 防止分母为零
     float diff = std::abs(a - b);
     return (diff <= ABS_TOL) || (diff <= REL_TOL * std::max(std::abs(a), std::abs(b) + eps));
@@ -48,14 +49,18 @@ struct SinhComputeFunc {
     // DataType模板参数，根据实际数据类型个数填写
     template<typename T, typename U>
     // 重载operator公有接口，提供给`ATVC::Kernel::EleWiseOpTemplate`调用
-    __aicore__ inline void operator()(AscendC::LocalTensor<T> x, AscendC::LocalTensor<T> y, AscendC::LocalTensor<U> tempBuffer1, AscendC::LocalTensor<U> tempBuffer2) {
+    __aicore__ inline void operator()(AscendC::LocalTensor<T> x,
+                                      AscendC::LocalTensor<T> y,
+                                      AscendC::LocalTensor<U> tempBuffer1,
+                                      AscendC::LocalTensor<U> tempBuffer2)
+    {
         // 开发调用AscendC Api自行实现计算仿函数
         uint32_t tiledCnt = y.GetSize(); // 进行单次基块计算的元素个数
         AscendC::Muls(tempBuffer1, x, static_cast<T>(-1), tiledCnt); // tempBuffer1 = -1 * x
         AscendC::Exp(tempBuffer1, tempBuffer1, tiledCnt); // tempbuffer1 = exp(-x)
         AscendC::Exp(tempBuffer2, x, tiledCnt);  // tempbuffer2 = exp(x)
         AscendC::Sub(y, tempBuffer2, tempBuffer1, tiledCnt); // y = exp(x) - exp(-x)
-        AscendC::Muls(y, y, static_cast<T>(0.5), tiledCnt);  // y = (e^(x) - e^(-x)) / 2 
+        AscendC::Muls(y, y, static_cast<T>(0.5), tiledCnt); // y = (e^(x) - e^(-x)) / 2 
     }
 };
 
@@ -123,8 +128,10 @@ int main()
     // 将tiling计算的运行时参数EleWiseParam param传到Device侧
     auto elementParamSize = sizeof(param);
     CHECK_ACL(aclrtMalloc((void**)&paramDevice, elementParamSize, ACL_MEM_MALLOC_HUGE_FIRST));
-    CHECK_ACL(aclrtMemcpy(paramDevice, elementParamSize, reinterpret_cast<uint8_t*>(&param), elementParamSize, ACL_MEMCPY_HOST_TO_DEVICE));
-    
+    CHECK_ACL(aclrtMemcpy(paramDevice, elementParamSize,
+                          reinterpret_cast<uint8_t *>(&param), elementParamSize,
+                          ACL_MEMCPY_HOST_TO_DEVICE));
+
     // 调用自定义的Kernel API, <<<>>>的BlockNum参数可通过param的TilingData获取
     SinhCustom<SinhOpTraits><<<param.tilingData.blockNum, nullptr, stream>>>(xDevice, yDevice, paramDevice);
 
@@ -144,8 +151,10 @@ int main()
 
     for (int32_t i = 0; i < eleNum; i++) {
         if (!IsClose(golden[i], outputY[i])) {
-            printf("Accuracy verification failed! The expected value of element in index [%d] is %f, but actual value is %f.\n", i, golden[i], outputY[i]);
-            return -1;
+            printf("Accuracy verification failed! The expected value of element "
+                   "in index [%d] is %f, but actual value is %f.\n",
+                   i, golden[i], outputY[i]);
+          return -1;
         }
     }
     printf("Accuracy verification passed.\n");
diff --git a/atvc/include/broadcast/broadcast_compute.h b/atvc/include/broadcast/broadcast_compute.h
index 6d380b29c7f3449dc779a4289a24e01bd3b2945a..90388f328b0f7be2405d66ad5caf64705d7959df 100644
--- a/atvc/include/broadcast/broadcast_compute.h
+++ b/atvc/include/broadcast/broadcast_compute.h
@@ -23,8 +23,11 @@ public:
     using DataType = typename ATVC::TypeListGet<inputDTypeList, 0>::Type;
 
     template <int32_t patternID>
-    __aicore__ inline void Compute(AscendC::LocalTensor<DataType> &src, uint32_t inputOffset,
-                                   AscendC::LocalTensor<DataType> &dst, uint32_t dimA, uint32_t dimB)
+    __aicore__ inline void Compute(AscendC::LocalTensor<DataType> &src,
+                                   uint32_t inputOffset,
+                                   AscendC::LocalTensor<DataType> &dst,
+                                   uint32_t dimA,
+                                   uint32_t dimB)
     {
         if (patternID == ATVC::AB_PATTERN::ABA) {
             ComputeBA(src, inputOffset, dst, dimA, dimB);
diff --git a/atvc/include/broadcast/broadcast_host.h b/atvc/include/broadcast/broadcast_host.h
index e6035a42950ce803ec0ca6cbfcf2c3a59f5801bd..80fba92814ae79e8be737276c88b7266f6d95ee2 100644
--- a/atvc/include/broadcast/broadcast_host.h
+++ b/atvc/include/broadcast/broadcast_host.h
@@ -49,7 +49,10 @@ void PrintParam(BroadcastPolicy* policy, BroadcastParam* param)
 }
 
 template<class OpTraits>
-bool CalcBroadcastTiling(std::vector<int64_t> shapeIn, std::vector<int64_t> shapeOut, BroadcastPolicy* policy, BroadcastParam* param)
+bool CalcBroadcastTiling(std::vector<int64_t> shapeIn,
+                         std::vector<int64_t> shapeOut,
+                         BroadcastPolicy* policy,
+                         BroadcastParam* param)
 {
     if(policy == nullptr || param == nullptr) {
         printf("[ERROR] Invalid input: policy or param is null pointer!\n");
@@ -64,7 +67,7 @@ bool CalcBroadcastTiling(std::vector<int64_t> shapeIn, std::vector<int64_t> shap
     auto inputDtype = GetOriInputType<DataType>();
     BroadcastTilingInputParam opInput = {shapeIn, shapeOut, inputDtype};
     OpTiling::BroadcastOpTiling tiling(opInput, policy, param);
-    if(!tiling.Run()) {
+    if (!tiling.Run()) {
         printf("[ERROR] Tiling Error\n");
         return false;
     }
diff --git a/atvc/include/broadcast/broadcast_op_template.h b/atvc/include/broadcast/broadcast_op_template.h
index 8dff4f5d10cb14fd28c800911826a451febf30a7..f9d777f5199fd1677c95affa8db46631af470d38 100644
--- a/atvc/include/broadcast/broadcast_op_template.h
+++ b/atvc/include/broadcast/broadcast_op_template.h
@@ -15,8 +15,7 @@
 #include "common/const_def.h"
 #include "broadcast/broadcast_utils/broadcast_buf_pool.h"
 namespace ATVC {
-struct BroadcastDataView
-{
+struct BroadcastDataView {
     uint32_t dimASize;
     uint32_t dimBSize;
     uint32_t inShape[ATVC::MAX_DIM];
@@ -35,7 +34,7 @@ template <class BroadcastCompute, const auto& SelectBroadcastPolicy>
 class BroadcastOpTemplate {
 public:
     using DataType = typename BroadcastCompute::DataType;
-    __aicore__ inline BroadcastOpTemplate(){}
+    __aicore__ inline BroadcastOpTemplate() {}
     /*
     BroadcastOpTemplate对外运行接口，主要完成资源初始化、数据搬入、计算调度、数据搬出操作
     @param src: 输入数据的gm指针
diff --git a/atvc/include/broadcast/broadcast_utils/broadcast_buf_pool.h b/atvc/include/broadcast/broadcast_utils/broadcast_buf_pool.h
index ae96b0a645440885e0e41cea316959fab91499b5..0888c084dc0f58f2f43a9152eaec66b212bd1a38 100644
--- a/atvc/include/broadcast/broadcast_utils/broadcast_buf_pool.h
+++ b/atvc/include/broadcast/broadcast_utils/broadcast_buf_pool.h
@@ -31,17 +31,6 @@ struct BrcPoolManagerUnit {
 class BroadcastBufPool {
 constexpr static int32_t MAX_INPUT_SIZE = 10;
 
-private:
-    BrcPoolManagerUnit inputUnit_;
-    BrcPoolManagerUnit computeUnit_;
-    event_t vecEventId_[MAX_INPUT_SIZE];
-    event_t outEventId_[MAX_INPUT_SIZE];
-    bool isBusyOut_[MAX_INPUT_SIZE] = {false};
-    AscendC::TBuf<> qQue_;
-    AscendC::TPipe* pipe_;
-    int32_t inputNum_;
-    int32_t outputNum_;
-
 public:
     __aicore__ inline BroadcastBufPool(){};
 
@@ -139,12 +128,14 @@ public:
         return idx;
     }
 
-    __aicore__ inline const void ResetEvent() {
+    __aicore__ inline const void ResetEvent()
+    {
         pipe_->Reset();
     }
 
 private:
-    __aicore__ inline int32_t GetComputeTensorId() {
+    __aicore__ inline int32_t GetComputeTensorId()
+    {
         uint32_t loopCnt = 0;
         do {
             computeUnit_.idx = (computeUnit_.idx + 1) % computeUnit_.bufNum;
@@ -158,10 +149,21 @@ private:
         return computeUnit_.idx;
     }
 
-    __aicore__ inline int32_t GetInputTensorId() {
+    __aicore__ inline int32_t GetInputTensorId()
+    {
         inputUnit_.idx = (inputUnit_.idx + 1) % inputUnit_.bufNum;
         return inputUnit_.idx;
     }
+
+    BrcPoolManagerUnit inputUnit_;
+    BrcPoolManagerUnit computeUnit_;
+    event_t vecEventId_[MAX_INPUT_SIZE];
+    event_t outEventId_[MAX_INPUT_SIZE];
+    bool isBusyOut_[MAX_INPUT_SIZE] = {false};
+    AscendC::TBuf<> qQue_;
+    AscendC::TPipe* pipe_;
+    int32_t inputNum_;
+    int32_t outputNum_;
 };
 } // namespace KernelUtils
 } // namespace ATVC
diff --git a/atvc/include/broadcast/common/broadcast_common.h b/atvc/include/broadcast/common/broadcast_common.h
index 56179401e6aafe16da3e39d6b99a366f94ffad2f..3f51b4e808a3188cedd287fc36df9d6a9b8e3d9e 100644
--- a/atvc/include/broadcast/common/broadcast_common.h
+++ b/atvc/include/broadcast/common/broadcast_common.h
@@ -28,11 +28,12 @@ namespace AB_PATTERN {
 };
 
 struct BroadcastPolicy {
-    public:
+public:
     int32_t patternID = -1;
     int32_t loopABCount = -1;
     int32_t loopInnerABCount = -1;
-    bool operator==(const BroadcastPolicy& rhs) const {
+    bool operator==(const BroadcastPolicy& rhs) const
+    {
         return this->patternID == rhs.patternID && this->loopABCount == rhs.loopABCount &&\
         this->loopInnerABCount == rhs.loopInnerABCount;
     }
diff --git a/atvc/include/broadcast/tiling/broadcast_tiling.h b/atvc/include/broadcast/tiling/broadcast_tiling.h
index 5cc2a3c7e389fc0f36a8ddb13de696532e5f12c0..3e62c61b27006d3c2ae0f2c50951d351a4e5ba6f 100644
--- a/atvc/include/broadcast/tiling/broadcast_tiling.h
+++ b/atvc/include/broadcast/tiling/broadcast_tiling.h
@@ -41,7 +41,7 @@ class BroadcastOpTiling {
 public:
     BroadcastOpTiling(ATVC::BroadcastTilingInputParam& inputParam,
         ATVC::BroadcastPolicy* policy, ATVC::BroadcastParam* param)
-        : opInput_(inputParam), param_(param), policy_(policy) 
+        : opInput_(inputParam), param_(param), policy_(policy)
     {
         compileInfo_ = ATVC::GetOpCompileInfo();
     }
@@ -137,8 +137,10 @@ private:
         return true;
     }
 
-    bool EliminateOne(std::vector<int64_t>& oriShapeIn, std::vector<int64_t>& oriShapeOut, std::vector<uint64_t>& shapeIn,
-                                    std::vector<uint64_t>& shapeOut)
+    bool EliminateOne(std::vector<int64_t> &oriShapeIn,
+                      std::vector<int64_t> &oriShapeOut,
+                      std::vector<uint64_t> &shapeIn,
+                      std::vector<uint64_t> &shapeOut)
     {
         bool isCurB = false;
         bool haveA = false;
diff --git a/atvc/include/common/atvc_opdef.h b/atvc/include/common/atvc_opdef.h
index c7be99f893e7d0b78f5fe0977e2121e6c512da39..bac3dfc7ad99145e44e9f72bc70c6ed8e393b838 100644
--- a/atvc/include/common/atvc_opdef.h
+++ b/atvc/include/common/atvc_opdef.h
@@ -21,7 +21,7 @@ enum class ParamType {
 };
 
 template<ParamType paramType_, typename ... Ts>
-struct ParamTypes{
+struct ParamTypes {
     using types = ATVC::TypeList<Ts...>;
     static constexpr ParamType usage = paramType_;
 };
diff --git a/atvc/include/common/dtype_utils.h b/atvc/include/common/dtype_utils.h
index 349e314d32bd09941ffc550e9854a3ed653204e4..c9e430557d5bf462c83563d7939767ef1aef3b79 100644
--- a/atvc/include/common/dtype_utils.h
+++ b/atvc/include/common/dtype_utils.h
@@ -14,7 +14,8 @@
 #include "graph/types.h"
 namespace ATVC {
 template <typename T>
-inline ge::DataType GetOriInputType() {
+inline ge::DataType GetOriInputType()
+{
     if constexpr (std::is_same<T, float>::value) {
         return ge::DataType::DT_FLOAT;
     } else if constexpr (std::is_same<T, int8_t>::value) {
@@ -32,8 +33,9 @@ inline ge::DataType GetOriInputType() {
     }
 };
 
-inline ge::DataType GetPromoteDataType(ge::DataType dtype) {
-    switch(dtype) {
+inline ge::DataType GetPromoteDataType(ge::DataType dtype)
+{
+    switch (dtype) {
         case ge::DataType::DT_BF16:
         case ge::DataType::DT_FLOAT16:
         case ge::DataType::DT_FLOAT:
diff --git a/atvc/include/common/kernel_utils.h b/atvc/include/common/kernel_utils.h
index c4f31ad45df0950c06474ba4bd556c788acd12c0..810aeff7e30c7d97744ffd4d95cf14b69cbe16f9 100644
--- a/atvc/include/common/kernel_utils.h
+++ b/atvc/include/common/kernel_utils.h
@@ -19,13 +19,15 @@
 namespace ATVC {
 __BLOCK_LOCAL__  static AscendC::TPipe g_pipe;
 template <AscendC::HardEvent EVENT>
-__aicore__ inline void SetEvent(AscendC::HardEvent evt) {
+__aicore__ inline void SetEvent(AscendC::HardEvent evt)
+{
     event_t eventId = static_cast<event_t>(GetTPipePtr()->FetchEventID(evt));
     AscendC::SetFlag<EVENT>(eventId);
     AscendC::WaitFlag<EVENT>(eventId);
 }
 template <AscendC::HardEvent EVENT>
-__aicore__ inline void SyncDataQueue() {
+__aicore__ inline void SyncDataQueue()
+{
     event_t eventId = static_cast<event_t>(GetTPipePtr()->FetchEventID(EVENT));
     AscendC::SetFlag<EVENT>(eventId);
     AscendC::WaitFlag<EVENT>(eventId);
@@ -76,7 +78,8 @@ struct GetPromoteType<bfloat16_t> {
     using T = float;
 };
 
-__aicore__ inline int64_t FindNearestPower2(const int64_t value) {
+__aicore__ inline int64_t FindNearestPower2(const int64_t value)
+{
     if (value == 0) {
         return 0;
     } else if (value <= CONST2) {
@@ -90,7 +93,8 @@ __aicore__ inline int64_t FindNearestPower2(const int64_t value) {
     }
 }
 
-__aicore__ inline int64_t CalLog2(int64_t value) {
+__aicore__ inline int64_t CalLog2(int64_t value)
+{
     int64_t res = 0;
     while (value > 1) {
         value = value >> 1;
diff --git a/atvc/include/common/ops_utils.h b/atvc/include/common/ops_utils.h
index b5239f7f1d059e4a35a548d87ec36a68a84069d4..efb33063196c3d9d098a42b8abd82e62dc103b9a 100644
--- a/atvc/include/common/ops_utils.h
+++ b/atvc/include/common/ops_utils.h
@@ -63,7 +63,7 @@ __host_aicore__ inline T Aligned(T value, T alignment)
  */
 template <typename T, typename U>
 __host_aicore__ inline typename std::enable_if <std::is_integral<T>::value, T>::type FloorAlign(T x, U align) {
-  return align == 0 ? 0 : x / align * align;
+    return align == 0 ? 0 : x / align * align;
 }
 
 }
diff --git a/atvc/include/common/platform.h b/atvc/include/common/platform.h
index af02863fcdae831085706f53251c2a77c0173e4a..ad27408858cfbde869e122611c75be751a93a874 100644
--- a/atvc/include/common/platform.h
+++ b/atvc/include/common/platform.h
@@ -18,7 +18,7 @@
 
 #include "kernel_operator.h"
 #include "kernel_tiling/kernel_tiling.h"
-namespace ATVC{
+namespace ATVC {
 namespace Platform {
 /**
  * Get the block size of unified buffer in bytes
diff --git a/atvc/include/common/tensor_info.h b/atvc/include/common/tensor_info.h
index 5a7a84058f47b7eeace90aca1ff20f52dbd6fd94..c2338b91d64df776d2353964e7813f14fcf6b0e6 100644
--- a/atvc/include/common/tensor_info.h
+++ b/atvc/include/common/tensor_info.h
@@ -26,7 +26,7 @@ namespace ATVC {
 // 记录一个LocalTensor在In/Out/Local中的偏移位置，以及自己的类型
 template<typename T>
 struct TensorInfo {
-    __aicore__ inline TensorInfo(){}
+    __aicore__ inline TensorInfo() {}
     using Dtype = T;
     AscendC::GlobalTensor<T> gmTensor;
     int32_t local_offset;
diff --git a/atvc/include/elewise/elewise_host.h b/atvc/include/elewise/elewise_host.h
index fbd676d8fef0b5d6c0446e6473f760c1ffeb868f..cffb7f99ea4a7dd4f4fd9b04bea721ed6359c67b 100644
--- a/atvc/include/elewise/elewise_host.h
+++ b/atvc/include/elewise/elewise_host.h
@@ -33,7 +33,8 @@ struct EleWiseTilingHyperParam {
     int32_t nBufferNum = 2;             // 每个Queue中的Tensor数量
 };
 
-int32_t GetEleWiseBasicCnt(const EleWiseTilingHyperParam &hiperParam, int32_t totalCnt, uint32_t blockNum, uint32_t ubufLimitCnt)
+int32_t GetEleWiseBasicCnt(const EleWiseTilingHyperParam &hiperParam,
+                           int32_t totalCnt, uint32_t blockNum, uint32_t ubufLimitCnt)
 {
     uint32_t basicCnt = hiperParam.basicCntBase; // 基本块初始值
     if (blockNum == 0) {
@@ -50,7 +51,7 @@ int32_t GetEleWiseBasicCnt(const EleWiseTilingHyperParam &hiperParam, int32_t to
             basicCnt = basicCnt / hiperParam.nBufferNum;
         }
     }
-    if(basicCnt > ubufLimitCnt) {
+    if (basicCnt > ubufLimitCnt) {
         basicCnt = ubufLimitCnt / basicCntMin * basicCntMin;
     }
     return basicCnt;
@@ -94,7 +95,8 @@ bool CalcEleWiseTiling(int32_t totalCnt, ATVC::EleWiseParam &param)
     if (blockNum > compileInfo.vectorCoreNum) {
         blockNum = compileInfo.vectorCoreNum;
     }
-    uint32_t ubufLimitCnt = ubSize / ((inTensorSumBytes + outTensorSumBytes) * hiperParam.nBufferNum + tempTensorSumBytes) / TILE_CONTROL;
+    uint32_t ubufLimitCnt = ubSize / ((inTensorSumBytes + outTensorSumBytes) * hiperParam.nBufferNum +
+                            tempTensorSumBytes) / TILE_CONTROL;
     if (tempTensorSumBytes == 0) {
         // 未声明tempbuffer时，预留 1 / (bufferNum + 1)的空间给AscendC高阶API内部临时空间使用
         ubufLimitCnt = ubSize / ((inTensorSumBytes + outTensorSumBytes) * (hiperParam.nBufferNum + 1));
diff --git a/atvc/include/elewise/elewise_op_template.h b/atvc/include/elewise/elewise_op_template.h
index 451f121305eb901dc9f11ac3e554150090038d9b..ff1d05604181a42f1c725c740e84fd9f0eafeec9 100644
--- a/atvc/include/elewise/elewise_op_template.h
+++ b/atvc/include/elewise/elewise_op_template.h
@@ -39,7 +39,7 @@ class EleWiseOpTemplate {
     static constexpr size_t tempTensorSumBytes = ATVC::TypeListReduce<Temps, SizeValue<0>, SumSizes>::Type::value;
 
 public:
-    __aicore__ inline EleWiseOpTemplate(){}
+    __aicore__ inline EleWiseOpTemplate() {}
     template<typename ...Args>
     __aicore__ inline void Run(Args&&... args) {
         g_pipe.Reset();
@@ -69,10 +69,12 @@ private:
 
             if (curBlockId < param_->tilingData.tailBlockCnt) {
                 this->curCoreCnt_ = (param_->tilingData.numPerBlock + 1) * param_->tilingData.tiledCnt;
-                this->curCoreStartCnt_ = (param_->tilingData.numPerBlock + 1) * curBlockId * param_->tilingData.tiledCnt;
+                this->curCoreStartCnt_ = (param_->tilingData.numPerBlock + 1) *
+                                          curBlockId * param_->tilingData.tiledCnt;
             } else {
                 this->curCoreCnt_ = param_->tilingData.numPerBlock * param_->tilingData.tiledCnt;
-                this->curCoreStartCnt_ = (curBlockId * param_->tilingData.numPerBlock + param_->tilingData.tailBlockCnt) * param_->tilingData.tiledCnt;
+                this->curCoreStartCnt_ = (curBlockId * param_->tilingData.numPerBlock +
+                                          param_->tilingData.tailBlockCnt) * param_->tilingData.tiledCnt;
             }
             if (curBlockId + 1 == param_->tilingData.blockNum) {
                 this->curCoreCnt_ += param_->tilingData.tailElemCnt;
@@ -88,7 +90,8 @@ private:
 private:
 
     // 申请LocalTensor等资源，初始化本核计算的GlobalTensor
-    __aicore__ inline void Init() {
+    __aicore__ inline void Init()
+    {
         // in/out/temp各自使用一个pipe进行管理，每个pipe里面管理的是ub地址连续的多个tensor
         if constexpr (InputCount > 0) {
             g_pipe.InitBuffer(inQueue, param_->nBufferNum, param_->tilingData.tiledCnt * inTensorSumBytes);
@@ -133,10 +136,10 @@ private:
         {
             caclCnt_ = tailCnt;
             CopyIn(inTensors, ATVC::MakeIndexSequence<InputCount>{});
-            Compute(inTensors, outTensors, tempTensors, 
-                ATVC::MakeIndexSequence<InputCount>{}, 
-                ATVC::MakeIndexSequence<OutputCount>{}, 
-                ATVC::MakeIndexSequence<TempCount>{}, 
+            Compute(inTensors, outTensors, tempTensors,
+                ATVC::MakeIndexSequence<InputCount>{},
+                ATVC::MakeIndexSequence<OutputCount>{},
+                ATVC::MakeIndexSequence<TempCount>{},
                 ATVC::Forward<Args>(args)...);
             CopyOut(outTensors, ATVC::MakeIndexSequence<OutputCount>{});
         }
@@ -154,7 +157,7 @@ private:
         if (alignMainCnt > 0) {
             AscendC::DataCopy(inLocal_i, tensorInfo.gmTensor[curCoreStartCnt_ + offsetCnt_], alignMainCnt);
         }
-        if(alignTailCnt > 0) {
+        if (alignTailCnt > 0) {
             struct AscendC::DataCopyExtParams repeatParams = {1, (uint16_t)(alignTailCnt*sizeof(DataType)), 0, 0, 0};
             AscendC::DataCopyPadExtParams<DataType> padParams;
             AscendC::DataCopyPad(inLocal_i[alignMainCnt],
@@ -169,7 +172,7 @@ private:
 
     // 所有 Tensor 的处理入口逻辑：递归完成对每个 Tensor 的处理
     template<typename T, typename... Ts>
-    __aicore__ inline void CopyInAllTensors(AscendC::LocalTensor<uint8_t>& inLocal, int32_t i , T& first, Ts&... rest) {
+    __aicore__ inline void CopyInAllTensors(AscendC::LocalTensor<uint8_t>& inLocal, int32_t i, T& first, Ts&... rest) {
         CopyInAllTensors(inLocal, i, first);
         CopyInAllTensors(inLocal, ++i, rest...);
     }
@@ -200,9 +203,9 @@ private:
         if (alignMainCnt > 0) {
             AscendC::DataCopy(tensorInfo.gmTensor[curCoreStartCnt_ + offsetCnt_], outLocal_i, alignMainCnt);
         }
-        if(alignTailCnt > 0) {
+        if (alignTailCnt > 0) {
             struct AscendC::DataCopyParams repeatParams = {1, (uint16_t)(alignTailCnt*sizeof(DataType)), 0, 0};
-            AscendC::DataCopyPad( tensorInfo.gmTensor[curCoreStartCnt_ + offsetCnt_ + alignMainCnt],
+            AscendC::DataCopyPad(tensorInfo.gmTensor[curCoreStartCnt_ + offsetCnt_ + alignMainCnt],
                 outLocal_i[alignMainCnt], repeatParams);
         }
     }
@@ -212,10 +215,13 @@ private:
     }
 
     // 所有 Tensor 的处理入口逻辑：递归完成对每个 Tensor 的处理
-    template<typename T, typename... Ts>
-    __aicore__ inline void CopyOutAllTensors(AscendC::LocalTensor<uint8_t>& outLocal, int32_t i, T& first, Ts&... rest) {
-        CopyOutAllTensors(outLocal, i, first);
-        CopyOutAllTensors(outLocal, ++i, rest...);
+    template <typename T, typename... Ts>
+    __aicore__ inline void
+    CopyOutAllTensors(AscendC::LocalTensor<uint8_t> &outLocal, int32_t i,
+                      T &first, Ts &...rest)
+    {
+      CopyOutAllTensors(outLocal, i, first);
+      CopyOutAllTensors(outLocal, ++i, rest...);
     }
     // 将所有输出tensor拷贝到gm
     template<typename OutTuple, std::size_t... Is>
@@ -236,9 +242,9 @@ __aicore__ inline void Compute(InTuple& inTensors, OutTuple& outTensors, TmpTupl
     if constexpr(TempCount > 0) {
         tempLocal = tempQueue.Get<uint8_t>();
     }
-    compute_(TupleElemGetLocalTensor<I1>(inLocal, inTensors, this->caclCnt_)..., 
-        TupleElemGetLocalTensor<I2>(outLocal, outTensors, this->caclCnt_)..., 
-        TupleElemGetLocalTensor<I3>(tempLocal, tempTensors, this->caclCnt_)..., 
+    compute_(TupleElemGetLocalTensor<I1>(inLocal, inTensors, this->caclCnt_)...,
+        TupleElemGetLocalTensor<I2>(outLocal, outTensors, this->caclCnt_)...,
+        TupleElemGetLocalTensor<I3>(tempLocal, tempTensors, this->caclCnt_)...,
         ATVC::Forward<Args>(args)...);
     inQueue.FreeTensor(inLocal);
     outQueue.template EnQue<uint8_t>(outLocal);
@@ -286,7 +292,8 @@ private:
 
 private:
     // 填充 addr 到数组
-    __aicore__ inline void FillAddrs(GM_ADDR argsArr[]) {
+    __aicore__ inline void FillAddrs(GM_ADDR argsArr[])
+    {
         for (std::size_t i = 0; i < InputCount; ++i) {
             inGMAddrs_[i].SetGlobalBuffer(argsArr[i]);
         }
@@ -304,7 +311,7 @@ private:
     template <typename List>
     __aicore__ inline constexpr void FillOffsets(std::size_t* offsets) {
         constexpr std::size_t count = ATVC::TypeListSize<List>::value;
-        FillOffsetsImpl<List>(offsets, ATVC::MakeIndexSequence<count>{}); 
+        FillOffsetsImpl<List>(offsets, ATVC::MakeIndexSequence<count>{});
     }
 
 private:
diff --git a/atvc/include/reduce/common/patterns.h b/atvc/include/reduce/common/patterns.h
index b88b115a172366e11098e84d55907415398260e7..364100b7aec41b35ced061e9754d407dd75d11eb 100644
--- a/atvc/include/reduce/common/patterns.h
+++ b/atvc/include/reduce/common/patterns.h
@@ -14,8 +14,7 @@
 #include "common/const_def.h"
 
 namespace ATVC {
-namespace ReducePattern
-{
+namespace ReducePattern {
 constexpr int32_t MAX_LOOP_DIM = (MAX_DIM + 1) / 2;
 constexpr int32_t PATTERN_A = 10;
 constexpr int32_t PATTERN_RA = 0;
diff --git a/atvc/include/reduce/common/reduce_common.h b/atvc/include/reduce/common/reduce_common.h
index 38de9f526767db0e1d99505ca34d6335516c9d2e..fca80f3fe2d28dc5a2dbdf9fc8877f3610c19758 100644
--- a/atvc/include/reduce/common/reduce_common.h
+++ b/atvc/include/reduce/common/reduce_common.h
@@ -46,7 +46,7 @@ namespace BASIC_CNT {
     static constexpr uint32_t BASIC_4096 = 4096;
 };
 
-namespace AR_COUNT{
+namespace AR_COUNT {
     static constexpr uint32_t A0R1 = 1;
     static constexpr uint32_t A0R2 = 2;
     static constexpr uint32_t A1R0 = 10;
@@ -77,11 +77,12 @@ namespace AR_COUNT{
 };
 
 struct ReducePolicy {
-    public:
+public:
     int32_t patternID = -1;
     int32_t loopARCount = -1;
     int32_t loopInnerARCount = -1;
-    bool operator==(const ReducePolicy& rhs) const {
+    bool operator==(const ReducePolicy& rhs) const
+    {
         return this->patternID == rhs.patternID && this->loopARCount == rhs.loopARCount &&\
         this->loopInnerARCount == rhs.loopInnerARCount;
     }
diff --git a/atvc/include/reduce/reduce_host.h b/atvc/include/reduce/reduce_host.h
index fe99aca7dfe5349036a650edcf9370cfa36c9a9c..479a98d9aaea6e8f461b323acbf8cc4edd59754e 100644
--- a/atvc/include/reduce/reduce_host.h
+++ b/atvc/include/reduce/reduce_host.h
@@ -45,9 +45,11 @@ void PrintParam(ReducePolicy* policy, ReduceParam* param)
  * @return bool 返回true表示计算成功，false表示失败。
  */
 template<class OpTraits>
-bool CalcReduceTiling(std::vector<int64_t> inputShape, std::vector<int64_t> reduceDim, ReducePolicy* policy, ReduceParam* param)
+bool CalcReduceTiling(std::vector<int64_t> inputShape,
+                      std::vector<int64_t> reduceDim, ReducePolicy *policy,
+                      ReduceParam *param) 
 {
-    if(policy == nullptr || param == nullptr) {
+    if (policy == nullptr || param == nullptr) {
         printf("[ERROR] Invalid input: policy or param is null pointer!\n");
         return false;
     }
@@ -56,7 +58,8 @@ bool CalcReduceTiling(std::vector<int64_t> inputShape, std::vector<int64_t> redu
         int32_t nBufferNum = 2;          // 每个Queue中的Tensor数量
     };
     using inputDTypeList = typename OpTraits::In::types;
-    static constexpr size_t inTensorSumBytes = ATVC::TypeListReduce<inputDTypeList, SizeValue<0>, SumSizes>::Type::value;
+    static constexpr size_t inTensorSumBytes =
+        ATVC::TypeListReduce<inputDTypeList, SizeValue<0>, SumSizes>::Type::value;
     if (inTensorSumBytes == 0) {
         printf("[ERROR] Tiling Error: OpTraits Input cannot be null!\n");
         return false;
@@ -76,7 +79,7 @@ bool CalcReduceTiling(std::vector<int64_t> inputShape, std::vector<int64_t> redu
     }
     ReduceTilingInputParam opInput = {reduceDim, inputShape, inputDtype, GetPromoteDataType(inputDtype)};
     OpTiling::ReduceOpTiling tiling(opInput, policy, param);
-    if(tiling.Run() != 0) {
+    if (tiling.Run() != 0) {
         printf("[ERROR] Tiling Error\n");
         return false;
     }
diff --git a/atvc/include/reduce/reduce_utils/reduce_block_aux_util.h b/atvc/include/reduce/reduce_utils/reduce_block_aux_util.h
index 9e723d1a33878e781ea568579aba94940f5c82ca..612b1d0e3a5eccb8a935f6232715eec6f843154f 100644
--- a/atvc/include/reduce/reduce_utils/reduce_block_aux_util.h
+++ b/atvc/include/reduce/reduce_utils/reduce_block_aux_util.h
@@ -27,7 +27,8 @@ namespace ATVC {
 namespace KernelUtils {
 namespace Reduce {
 constexpr uint16_t MAX_OFFSET = 16;
-__aicore__ inline int16_t MainR(int64_t dimR, bool isAR) {
+__aicore__ inline int16_t MainR(int64_t dimR, bool isAR)
+{
     if (isAR && dimR < UB_ALIGN_32) {
         return 0;
     }
@@ -41,7 +42,8 @@ __aicore__ inline int16_t MainR(int64_t dimR, bool isAR) {
     return static_cast<int16_t>(mainR);
 }
 
-__aicore__ inline int64_t GetCacheID(const int64_t idx) {
+__aicore__ inline int64_t GetCacheID(const int64_t idx)
+{
     return bcnt1(idx ^ (idx + CONST1)) - CONST1;
 }
 
@@ -273,7 +275,8 @@ __aicore__ inline constexpr ReduceSchLoopInfo GetSchLoopInfo5() {
     return schInfo;
 }
 
-__aicore__ inline constexpr ReduceSchLoopInfo GetGroupSchLoopInfo() {
+__aicore__ inline constexpr ReduceSchLoopInfo GetGroupSchLoopInfo()
+{
     constexpr ReduceSchLoopInfo schInfo = {.patternID = ReducePattern::PATTERN_RA,
                                            .reduceDichotomy = CONST1,
                                            .loopACount = CONST1,
diff --git a/atvc/include/reduce/reduce_utils/reduce_buf_pool.h b/atvc/include/reduce/reduce_utils/reduce_buf_pool.h
index f84f8741c029a228bc551fdd4ba48a8b1f4a5cb7..01c63adaaa0391efa278c47d1dfb3d10790cdb5f 100644
--- a/atvc/include/reduce/reduce_utils/reduce_buf_pool.h
+++ b/atvc/include/reduce/reduce_utils/reduce_buf_pool.h
@@ -33,15 +33,6 @@ struct PoolManagerUnit {
 class ReduceBufPool {
     constexpr static int32_t MAX_INPUT_SIZE = 10;
 
-private:
-    bool memo_[MAX_INPUT_SIZE] = {0};
-    PoolManagerUnit inputUnit_;
-    PoolManagerUnit computeUnit_;
-    event_t eventIdV2Mte2_[MAX_INPUT_SIZE];
-    AscendC::TBuf<> qQue_;
-    AscendC::TPipe* pipe_;
-    int32_t basicNum_;
-
 public:
     __aicore__ inline ReduceBufPool(){};
 
@@ -64,7 +55,8 @@ public:
         AscendC::Duplicate<DataType>(inputUb, 0, basicNum_ * inputNum);
     }
 
-    __aicore__ inline void ResetEvent() {
+    __aicore__ inline void ResetEvent()
+    {
         inputUnit_.idx = -1;
         computeUnit_.idx = -1;
     }
@@ -75,7 +67,8 @@ public:
         computeUnit_.offset = basicNum_ * sizeof(T) * inputNum;
     }
 
-    __aicore__ inline void ResetComputeSize(int32_t computeNum) {
+    __aicore__ inline void ResetComputeSize(int32_t computeNum)
+    {
         computeUnit_.bufNum = computeNum;
     }
 
@@ -113,15 +106,25 @@ public:
     }
 
 private:
-    __aicore__ inline int32_t GetComputeTensorId() {
+    __aicore__ inline int32_t GetComputeTensorId()
+    {
         computeUnit_.idx = (computeUnit_.idx + 1) % computeUnit_.bufNum;
         return computeUnit_.idx;
     }
 
-    __aicore__ inline int32_t GetInputTensorId() {
+    __aicore__ inline int32_t GetInputTensorId()
+    {
         inputUnit_.idx = (inputUnit_.idx + 1) % inputUnit_.bufNum;
         return inputUnit_.idx;
     }
+
+    bool memo_[MAX_INPUT_SIZE] = {0};
+    PoolManagerUnit inputUnit_;
+    PoolManagerUnit computeUnit_;
+    event_t eventIdV2Mte2_[MAX_INPUT_SIZE];
+    AscendC::TBuf<> qQue_;
+    AscendC::TPipe* pipe_;
+    int32_t basicNum_;
 };  // class ReduceBufPool
 } // namespace KernelUtils
 } // namespace ATVC
diff --git a/atvc/include/reduce/tiling/reduce_tiling.h b/atvc/include/reduce/tiling/reduce_tiling.h
index d35bf19855c01f020a09ce79bad76fc175fb90e7..f649485c8e7db518c5383d53222893b16a653bcf 100644
--- a/atvc/include/reduce/tiling/reduce_tiling.h
+++ b/atvc/include/reduce/tiling/reduce_tiling.h
@@ -69,12 +69,12 @@ static void MakeWrapDim(const std::vector<int64_t>& shape, std::vector<int64_t>&
     }
 }
 
-class ReduceOpTiling
-{
+class ReduceOpTiling {
 public:
 ReduceOpTiling(ReduceTilingInputParam& inputParam,
     ATVC::ReducePolicy* policy, ATVC::ReduceParam* param)
-    : param_(param), policy_(policy), opInput_(inputParam) {
+    : param_(param), policy_(policy), opInput_(inputParam)
+    {
         compileInfo_ = ATVC::GetOpCompileInfo();
     };
 
@@ -438,7 +438,8 @@ void ComputeSplit(std::vector<uint64_t>& shape)
     param_->tilingData.basicBlock = basicBlock_;
     param_->tilingData.coreNum = static_cast<int32_t>(compileInfo_.vectorCoreNum);
     ComputeStride<Pattern>(shape);
-    uint32_t realCore = OpsUtils::CeilDiv(unitA_.outer, factorACntPerCore) * OpsUtils::CeilDiv(unitR_.outer, factorRCntPerCore);
+    uint32_t realCore = OpsUtils::CeilDiv(unitA_.outer, factorACntPerCore) *
+                        OpsUtils::CeilDiv(unitR_.outer, factorRCntPerCore);
     param_->tilingData.coreNum = realCore;
 }
 template <class Pattern>
@@ -481,23 +482,16 @@ bool CalcCacheLineStep(const std::vector<uint64_t>& shape)
             cBlock_.axis = i;
         }
     }
-    for (int32_t i = Pattern::TailA ? Pattern::Dim - 1 : Pattern::Dim - ATVC::CONST2; i > cBlock_.axis; i -= ATVC::CONST2) {
-        if (i == Pattern::Dim - 1) {
-            aInCacheLine *= OpsUtils::CeilAlign(shape[i], ubBlockSize);
-        } else {
-            aInCacheLine *= shape[i];
-        }
+    const int32_t startIndex = Pattern::TailA ? Pattern::Dim - 1 : Pattern::Dim - ATVC::CONST2;
+    for (int32_t i = startIndex; i > cBlock_.axis; i -= ATVC::CONST2) {
+        aInCacheLine *= (i == Pattern::Dim - 1) ? OpsUtils::CeilAlign(shape[i], ubBlockSize) : shape[i];
     }
     bool basicSplitA = (cBlock_.axis + (Pattern::FirstA ? 1 : 0)) % ATVC::CONST2;
     if (basicSplitA) {
         aInCacheLine *= cacheLineStep;
     }
     for (int32_t i = Pattern::Dim - 1; i > cBlock_.axis; --i) {
-        if (i == Pattern::Dim - 1) {
-            cBlock_.size = cBlock_.size * OpsUtils::CeilAlign(shape[i], ubBlockSize);
-        } else {
-            cBlock_.size *= shape[i];
-        }
+        cBlock_.size *= (i == Pattern::Dim - 1) ? OpsUtils::CeilAlign(shape[i], ubBlockSize) : shape[i];
     }
     cBlock_.size *= cacheLineStep;
     cBlock_.cacheLineStep = cacheLineStep;
@@ -513,8 +507,9 @@ void ComputeUnitA(const std::vector<uint64_t>& shape)
     uint64_t outerA = unitA_.outer;
     uint64_t innerA = unitA_.inner;
     uint64_t maxCacheA = MAX_INNER_A * sizeof(float) / ge::GetSizeByDataType(opInput_.promoteDtpye);
-    uint64_t maxInnerA =
-        Pattern::ID == ATVC::ReducePattern::PATTERN_A ? basicBlock_ / ge::GetSizeByDataType(opInput_.inputDtype) : maxCacheA;
+    const bool isPatternA = (Pattern::ID == ATVC::ReducePattern::PATTERN_A);
+    const uint64_t patternABlockSize = basicBlock_ / ge::GetSizeByDataType(opInput_.inputDtype);
+    uint64_t maxInnerA = isPatternA ? patternABlockSize : maxCacheA;
     uint64_t stepLen = Pattern::ID == ATVC::ReducePattern::PATTERN_A ? A_STEP_LEN : 1;  // 纯A的步长为4, 减少循环次数
     bool basicSplitA = (axisInCacheLine + (Pattern::FirstA ? 1 : 0)) % ATVC::CONST2;
     uint64_t bBlockNum = basicBlock_ / ge::GetSizeByDataType(opInput_.inputDtype);