diff --git a/src/atb/context/context_base.cpp b/src/atb/context/context_base.cpp
index e87f837c23e18737bf4c84098666c6344fb6a693..5d994c7ca7bb4878bf313d187912d8e3acf11b0b 100644
--- a/src/atb/context/context_base.cpp
+++ b/src/atb/context/context_base.cpp
@@ -349,7 +349,7 @@ Status ContextBase::FreeArgsHostBuffer(void *addr)
 {
     return hostAllocator_->Deallocate(addr);
 }
-bool ContextBase::GetLaunchWithTilingStatus()
+bool ContextBase::GetLaunchWithTilingStatus() const
 {
     return mode_ != GRAPH_LAUNCH_MODE;
 }
diff --git a/src/atb/context/context_base.h b/src/atb/context/context_base.h
index 9ea71adbd6665c3078cd6a9056292fa17a57c531..6cc54754abb3dd5815968896e93c6e5f6fc4e7a6 100644
--- a/src/atb/context/context_base.h
+++ b/src/atb/context/context_base.h
@@ -49,7 +49,7 @@ public:
     void *GetArgsHostBuffer(size_t bufferSize);
     Status FreeArgsDeviceBuffer(void *addr);
     Status FreeArgsHostBuffer(void *addr);
-    bool GetLaunchWithTilingStatus();
+    bool GetLaunchWithTilingStatus() const;
 
 private:
     Status CreateCopyStreamAndEvents();
diff --git a/src/atb/operation/operation_base.cpp b/src/atb/operation/operation_base.cpp
index 8668f9e5121b1d502720fa50f4132b95237374ef..60e8591a63406fa4dd36c2bdf8b14e6ed7a2516c 100644
--- a/src/atb/operation/operation_base.cpp
+++ b/src/atb/operation/operation_base.cpp
@@ -784,7 +784,7 @@ Status OperationBase::ExecuteVariantPackOutTensorCheck(const SVector<TensorType>
     return NO_ERROR;
 }
 
-Status OperationBase::ExecuteVariantPackCheck(const VariantPack &variantPack)
+Status OperationBase::ExecuteVariantPackCheck(const VariantPack &variantPack) const
 {
     Status st = NO_ERROR;
     st = ExecuteVariantPackInTensorCheck(variantPack.inTensors);
@@ -1347,7 +1347,7 @@ aclrtStream OperationBase::GetExecuteStream(Context *context) const
     return streams.at(streamId_);
 }
 
-Status OperationBase::CopyArgsToDevice(Context *context)
+Status OperationBase::CopyArgsToDevice(Context *context) const
 {
     Status st = NO_ERROR;
     if (hostArgsBuffer_ == nullptr) {
diff --git a/src/atb/operation/operation_base.h b/src/atb/operation/operation_base.h
index 309fc0bf804cba403db7188a6b47e3718e52af03..66227d5430363a7f134812dd353cfdd03def8371 100644
--- a/src/atb/operation/operation_base.h
+++ b/src/atb/operation/operation_base.h
@@ -87,7 +87,7 @@ private:
                         Context *context);
     template <typename TensorType> Status ExecuteVariantPackInTensorCheck(const SVector<TensorType> &inTensors) const;
     template <typename TensorType> Status ExecuteVariantPackOutTensorCheck(const SVector<TensorType> &outTensors) const;
-    Status ExecuteVariantPackCheck(const VariantPack &variantPack);
+    Status ExecuteVariantPackCheck(const VariantPack &variantPack) const;
     void InitRunnerVariantPack(const VariantPack &variantPack);
     Status CopyHostTilingToDevice(aclrtStream stream);
     Status CopyTilingToDevice();
@@ -116,7 +116,7 @@ private:
     Status EagerModeLaunch();
     Status GraphModeLaunch();
     void ProfilingPrepare();
-    Status CopyArgsToDevice(Context *context);
+    Status CopyArgsToDevice(Context *context) const;
 
 private:
     std::string logPrefix_;
diff --git a/src/kernels/mixkernels/fused_add_topk_div/tiling/fused_add_topk_div_tiling.cpp b/src/kernels/mixkernels/fused_add_topk_div/tiling/fused_add_topk_div_tiling.cpp
index 782bc8f52517d03d13a56aaff52c9bb9ec248e90..897ae44aff66b9a92649ae09be7b44478a217936 100644
--- a/src/kernels/mixkernels/fused_add_topk_div/tiling/fused_add_topk_div_tiling.cpp
+++ b/src/kernels/mixkernels/fused_add_topk_div/tiling/fused_add_topk_div_tiling.cpp
@@ -74,7 +74,7 @@ namespace AtbOps {
         void GetTilingKey(TilingData *tilingDataPtr);
         void GetUsedCore();
         void SplitUb();
-        void FillTilingData(TilingData *tilingDataPtr);
+        void FillTilingData(TilingData *tilingDataPtr) const;
         template <typename T1, typename T2>
         inline T1 CeilAlign(T1 a, T2 b) const
         {
@@ -139,7 +139,7 @@ namespace AtbOps {
     }
 
     template <typename TilingData>
-    void FusedAddTopkDivTiling<TilingData>::FillTilingData(TilingData *tilingDataPtr)
+    void FusedAddTopkDivTiling<TilingData>::FillTilingData(TilingData *tilingDataPtr) const
     {
         tilingDataPtr->firstDimSize = firstDimSize;
         tilingDataPtr->secondDimSize = secondDimSize;
diff --git a/src/kernels/mixkernels/fusion/fusion_operation.cpp b/src/kernels/mixkernels/fusion/fusion_operation.cpp
index 3f61161d63865f82b2e8ba68f0b93629ac76d1e0..5f81251140ec3e422ff2ae840a843ece407d1971 100644
--- a/src/kernels/mixkernels/fusion/fusion_operation.cpp
+++ b/src/kernels/mixkernels/fusion/fusion_operation.cpp
@@ -210,7 +210,7 @@ public:
         return;
     }
 
-    void ErasedFusion()
+    void ErasedFusion() const
     {
         static uint8_t erasedFusionKernelBinData[DYNAMICNULLSIZE];
         uint32_t counter = 0;
@@ -292,7 +292,7 @@ protected:
         return Status::OkStatus();
     }
     
-    bool IsSoftLink(const char *path)
+    bool IsSoftLink(const char *path) const
     {
         struct stat fileStat;
         if (lstat(path, &fileStat) != 0) {
diff --git a/src/kernels/mixkernels/mla_preprocess/tiling/mla_preprocess_tiling.cpp b/src/kernels/mixkernels/mla_preprocess/tiling/mla_preprocess_tiling.cpp
index 9fe26640a189f3dac3169b9df9a863fbc214789c..4dc0cae28d0231f55384900f5373c70436dd37ce 100644
--- a/src/kernels/mixkernels/mla_preprocess/tiling/mla_preprocess_tiling.cpp
+++ b/src/kernels/mixkernels/mla_preprocess/tiling/mla_preprocess_tiling.cpp
@@ -96,12 +96,12 @@ public:
 
 private:
     void GetTileSize();
-    float GetCost(const uint32_t m0, const uint32_t n0);
+    float GetCost(const uint32_t m0, const uint32_t n0) const;
     void UpdateTileSize(const uint32_t m0, const uint32_t n0);
     void Swizzle();
-    uint32_t ComputeL1AbSize();
-    uint32_t ComputeK0ForABpingpong(uint32_t l1AbSize);
-    bool IsLoadAllAmat(uint32_t l1AbSize);
+    uint32_t ComputeL1AbSize() const;
+    uint32_t ComputeK0ForABpingpong(uint32_t l1AbSize) const;
+    bool IsLoadAllAmat(uint32_t l1AbSize) const;
     uint32_t ComputeK0ForOnlyBpingpong(uint32_t l1AbSize);
 
 private:
@@ -202,13 +202,13 @@ uint32_t PpMatmulTilingApi::ComputeK0ForOnlyBpingpong(uint32_t l1AbSize)
     return k0B0 > CONST_512 ? RoundDown(k0B0, CONST_512) : k0B0;
 }
 
-bool PpMatmulTilingApi::IsLoadAllAmat(uint32_t l1AbSize)
+bool PpMatmulTilingApi::IsLoadAllAmat(uint32_t l1AbSize) const
 {
     return (coreLoop_ > blockDim_) && enDequant_ && (kLoop_ > 1) &&
            (l1AbSize > RoundUp(m_, CONST_16) * RoundUp(k_, CONST_32) * inDataSize_) && (mLoop_ == 1);
 }
 
-uint32_t PpMatmulTilingApi::ComputeK0ForABpingpong(uint32_t l1AbSize)
+uint32_t PpMatmulTilingApi::ComputeK0ForABpingpong(uint32_t l1AbSize) const
 {
     uint32_t k0Max = static_cast<uint32_t>(static_cast<float>(l1AbSize / DIM_2) / ((m0_ + n0_) * inDataSize_));
     uint32_t tmpK0;
@@ -223,7 +223,7 @@ uint32_t PpMatmulTilingApi::ComputeK0ForABpingpong(uint32_t l1AbSize)
     return tmpK0;
 }
 
-uint32_t PpMatmulTilingApi::ComputeL1AbSize()
+uint32_t PpMatmulTilingApi::ComputeL1AbSize() const
 {
     if (enDequant_ && deqOnTheFly_) {
         return L1_BUFFER_SIZE;
@@ -231,7 +231,7 @@ uint32_t PpMatmulTilingApi::ComputeL1AbSize()
     return enDequant_ ? (L1_BUFFER_SIZE - L1_BIAS_SIZE - L1_SCALE_SIZE) : L1_BUFFER_SIZE;
 }
 
-float PpMatmulTilingApi::GetCost(const uint32_t m0, const uint32_t n0)
+float PpMatmulTilingApi::GetCost(const uint32_t m0, const uint32_t n0) const
 {
     float aCoef = 1.0;
     float bCoef = 1.0;
@@ -321,7 +321,7 @@ public:
     void RopeConcatTiling(const OpParam::MlaPreprocess &param, const uint32_t &aicNum);
     void EinSumQuantTiling(const OpParam::MlaPreprocess &param, const uint32_t &aicNum, const TensorDType inDtype);
     void SetTiling(AtbOps::MlaTilingData *tilingParam);
-    void SetTilingKey(const Mki::LaunchParam &launchParam, Mki::KernelInfo &kernelInfo);
+    void SetTilingKey(const Mki::LaunchParam &launchParam, Mki::KernelInfo &kernelInfo) const;
     void SetMlapoWorkSpace(const TensorDType inDtype, const OpParam::MlaPreprocess &param, Mki::KernelInfo &kernelInfo);
 };
 
@@ -477,7 +477,7 @@ void MlaPreprocessTiling::SetTiling(AtbOps::MlaTilingData *tilingParam)
     tilingParam->esqColTail = tilingData.esqColTail;
 }
 
-void MlaPreprocessTiling::SetTilingKey(const Mki::LaunchParam &launchParam, Mki::KernelInfo &kernelInfo)
+void MlaPreprocessTiling::SetTilingKey(const Mki::LaunchParam &launchParam, Mki::KernelInfo &kernelInfo) const
 {
     OpParam::MlaPreprocess param = AnyCast<OpParam::MlaPreprocess>(launchParam.GetParam());
     TensorFormat formatWeight1 = launchParam.GetInTensor(INDEX_WDQKV).desc.format;
diff --git a/src/ops_infer/relay_attention/param.cpp b/src/ops_infer/relay_attention/param.cpp
index 5ca38676319c05053fdd56a54ef5502c8609f072..42e4d7c6d338f8a34fb2f08b63f85097ebcbb32c 100644
--- a/src/ops_infer/relay_attention/param.cpp
+++ b/src/ops_infer/relay_attention/param.cpp
@@ -39,7 +39,7 @@ bool RelayAttentionVariantPackParam::BuildFromTensor(const SVector<Mki::Tensor>
     return true;
 }
 
-void RelayAttentionVariantPackParam::ReintCastShapeFix(const Mki::Tensor tensor, std::vector<atb::Tensor> &tensorList)
+void RelayAttentionVariantPackParam::ReintCastShapeFix(const Mki::Tensor tensor, std::vector<atb::Tensor> &tensorList) const
 {
     if (tensor.desc.dims.size() - 1 != tensorList[0].desc.shape.dimNum) {
         size_t diffDimNum = static_cast<size_t>(tensorList[0].desc.shape.dimNum);
diff --git a/src/ops_infer/relay_attention/param.h b/src/ops_infer/relay_attention/param.h
index b8b6c7b9dea54d7a5a7d8eb1c7be01a5e83b8a9b..68bc70392ba4db60fafb6c1fbcbb43391838c842 100644
--- a/src/ops_infer/relay_attention/param.h
+++ b/src/ops_infer/relay_attention/param.h
@@ -30,7 +30,7 @@ struct RelayAttentionVariantPackParam {
     std::vector<Tensor> valueShare;
     bool BuildFromTensor(const SVector<Mki::Tensor> &inTensors);
     bool HostDataCheck(const SVector<Mki::Tensor> &inTensors);
-    void ReintCastShapeFix(const Mki::Tensor tensor, std::vector<atb::Tensor> &tensorList);
+    void ReintCastShapeFix(const Mki::Tensor tensor, std::vector<atb::Tensor> &tensorList) const;
 };
 } // namespace atb
 #endif
\ No newline at end of file
diff --git a/src/torch_atb/graph_node.cpp b/src/torch_atb/graph_node.cpp
index fa207305d2e34e24215a63d37e54f2667a2e1424..d356587301a4a5c72f4b7610f02c00487b85fbd7 100644
--- a/src/torch_atb/graph_node.cpp
+++ b/src/torch_atb/graph_node.cpp
@@ -32,7 +32,7 @@ bool GraphNode::FindOutput(const std::string &id) const
     return it != outTensorIds.end();
 }
 
-void GraphNode::SetStreamId(uint32_t streamId)
+void GraphNode::SetStreamId(uint32_t streamId) const
 {
     if (!operation) {
         throw std::runtime_error("Set execute stream id fail, operation is nullptr");
@@ -40,7 +40,7 @@ void GraphNode::SetStreamId(uint32_t streamId)
     SetExecuteStreamId(operation, streamId);
 }
 
-uint32_t GraphNode::GetStreamId()
+uint32_t GraphNode::GetStreamId() const
 {
     if (!operation) {
         throw std::runtime_error("Get execute stream id fail, operation is nullptr");
diff --git a/src/torch_atb/graph_node.h b/src/torch_atb/graph_node.h
index fd5a73dcabcee8e2054383cb6f21c6d25c2f035c..35a1f1f208235fbf169df20c1ad49f6b66edf20e 100644
--- a/src/torch_atb/graph_node.h
+++ b/src/torch_atb/graph_node.h
@@ -22,8 +22,8 @@ public:
     void SetOperation(atb::Operation *op);
     std::string GetOutput(size_t index) const;
     bool FindOutput(const std::string &id) const;
-    void SetStreamId(uint32_t streamId);
-    uint32_t GetStreamId();
+    void SetStreamId(uint32_t streamId) const;
+    uint32_t GetStreamId() const;
 
     std::vector<std::string> inTensorIds;
     std::vector<std::string> outTensorIds;