diff --git a/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_dn.h b/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_dn.h
index 150a582743448ef9c1ecd2d6a49a31ae67cfdfc7..2bbdacf8384bbb6a0e0de3a8a907d3df725e055c 100644
--- a/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_dn.h
+++ b/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_dn.h
@@ -50,8 +50,9 @@ public:
                 width);
         });
 
-        static_assert(!IsStaticPaddingEnable(MM_CFG), "Unsupported Static Padding for DN2NZ.");
-        int64_t srcOffset = ((int64_t)row * (int64_t)gCol  + (int64_t)col);
+        ASCENDC_ASSERT(IsStaticPaddingEnable(MM_CFG),
+            { KERNEL_LOG(KERNEL_ERROR, "Unsupported Static Padding for DN2NZ."); });
+        int64_t srcOffset = (static_cast<int64_t>(row) * static_cast<int64_t>(gCol)  + static_cast<int64_t>(col));
         Dn2NzParams dn2nzParams;
         dn2nzParams.dnNum = dnNum;
         dn2nzParams.nValue = width;
diff --git a/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_nd.h b/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_nd.h
index 2df661eaf6bc2dad96232dbaf37e9af26cc7f8ee..1b861f61d8c8f724c4547077a69a553fe34ab955 100644
--- a/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_nd.h
+++ b/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_nd.h
@@ -64,9 +64,10 @@ public:
         }
         int64_t srcOffset;
         if constexpr (IsSupportB4<TransT>()) {
-            srcOffset = ((int64_t)row * (int64_t)gCol * INT4_TWO + (int64_t)col);
+            srcOffset = (static_cast<int64_t>(row) * static_cast<int64_t>(gCol) * INT4_TWO +
+                static_cast<int64_t>(col) * INT4_TWO);
         } else {
-            srcOffset = ((int64_t)row * (int64_t)gCol  + (int64_t)col);
+            srcOffset = (static_cast<int64_t>(row) * static_cast<int64_t>(gCol)  + static_cast<int64_t>(col));
         }
         Nd2NzParams nd2nzParams;
         nd2nzParams.ndNum = ndNum;
@@ -121,7 +122,7 @@ public:
         DataCopyEnhancedParams enhancedParams;
         enhancedParams.blockMode = BlockMode::BLOCK_MODE_VECTOR;
 
-        int32_t srcStride = gCol * sizeof(SrcT) / ONE_BLK_SIZE - 1;
+        int32_t srcStride = gCol * static_cast<int32_t>(sizeof(SrcT)) / ONE_BLK_SIZE - 1;
         if (gCol % c0Size_ || srcStride >= UINT16_MAX) {
             // each block len is only 32B
             for (int32_t i = 0; i < calcWidth; i++) {
diff --git a/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_nz.h b/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_nz.h
index b26d9beda28088c48a6736b4e261983361ef620c..e57149e8a47184053c76b2cba458cd9e7796b46a 100644
--- a/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_nz.h
+++ b/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_nz.h
@@ -72,7 +72,7 @@ public:
     {
         int calcWidth = CeilT(tileWidth, c0Size_) * c0Size_;
         int calcHigh = CeilT(tileHeight, c0Size_) * c0Size_;
-        int64_t size = calcHigh * calcWidth;
+        int64_t size = static_cast<int64_t>(calcHigh * calcWidth);
         LocalTensor<TransT> rightMatrix =
             MATMUL_MODULE(LocalWorkspace)->GetND2NZWorkspace(0).template ReinterpretCast<TransT>();
         rightMatrix.SetSize(size);
diff --git a/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_using_ub_nd.h b/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_using_ub_nd.h
index 30228c84b49f30d77a7e230c97295f09f05ae6ab..79b6a454624270c278eb149e61235a63d345f932 100644
--- a/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_using_ub_nd.h
+++ b/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_using_ub_nd.h
@@ -43,7 +43,7 @@ public:
                                      const int col, const int height, const int width, const int gCol)
     {
         if constexpr (ToMatmulConfig(MM_CFG).enableL1CacheUB) {
-            static_assert(DoMatmulMDL(MM_CFG), "Only MDL version support L1CacheUB.");
+            ASCENDC_ASSERT(DoMatmulMDL(MM_CFG), { KERNEL_LOG(KERNEL_ERROR, "Only MDL version support L1CacheUB."); });
         }
         int calcWidth = CeilT(width, c0Size_);
         bool isBankConflict =
@@ -86,7 +86,7 @@ public:
         }
 
         if constexpr (IsSameTypeV<TransT, half> && IsSameTypeV<SrcT, int8_t>) {
-            static_assert(DoMatmulMDL(MM_CFG), "Only MDL version support AntiQuant.");
+            ASCENDC_ASSERT(DoMatmulMDL(MM_CFG), { KERNEL_LOG(KERNEL_ERROR, "Only MDL version support AntiQuant."); });
             if (!MATMUL_MODULE(CopyCubeInParams)->IsTranspose()) {
                 enQueEvtID = GetTPipePtr()->FetchEventID(HardEvent::MTE2_S);
                 SetFlag<HardEvent::MTE2_S>(enQueEvtID);
@@ -225,10 +225,10 @@ public:
                     tail = CeilT<int32_t>(tail, DIV_TWO);
                 }
             }
-            uint16_t mask_tail_16bit = ~((1 << tail) - 1);
-            uint64_t mask_tail_64bit = mask_tail_16bit;
-            mask[0] = mask_tail_64bit + (mask_tail_64bit << FIRST_16BIT_OFFSET_MM_API) +
-                      (mask_tail_64bit << SECOND_16BIT_OFFSET_MM_API) + (mask_tail_64bit << THIRD_16BIT_OFFSET_MM_API);
+            uint16_t maskTail16bit = ~((1 << tail) - 1);
+            uint64_t maskTail64bit = static_cast<uint64_t>(maskTail16bit);
+            mask[0] = maskTail64bit + (maskTail64bit << FIRST_16BIT_OFFSET_MM_API) +
+                      (maskTail64bit << SECOND_16BIT_OFFSET_MM_API) + (maskTail64bit << THIRD_16BIT_OFFSET_MM_API);
             mask[1] = mask[0];
             constexpr int32_t DUP_CEIL_NUM = 8;
             if constexpr (IsSameTypeV<TransT, int8_t>) {
diff --git a/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_utils.h b/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_utils.h
index d894cd24eb4b47349aedb308130157deeae84375..4b021c96b7b747e65c3d98c99cc17bfbc9846ecb 100644
--- a/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_utils.h
+++ b/impl/matmul/stage/copy_cube_in/copy_tile_to_cube/data_copy_wrapper_utils.h
@@ -37,14 +37,14 @@ __aicore__ inline void NDPadZeroForWidth(LocalTensor<TransT>& dst, const int hei
 {
     constexpr int32_t c0Size_ = AuxGetC0Size<TransT>();
 
-    uint16_t mask_tail_16bit = ~((1 << tail) - 1);
-    uint64_t mask_tail_64bit = mask_tail_16bit;
-    if (mask_tail_64bit == 0) {
+    uint16_t maskTail16bit = ~((1 << tail) - 1);
+    uint64_t maskTail64bit = static_cast<uint64_t>(maskTail16bit);
+    if (maskTail64bit == 0) {
         return;
     }
     uint64_t mask[2];
-    mask[0] = mask_tail_64bit + (mask_tail_64bit << FIRST_16BIT_OFFSET_MM_API) +
-              (mask_tail_64bit << SECOND_16BIT_OFFSET_MM_API) + (mask_tail_64bit << THIRD_16BIT_OFFSET_MM_API);
+    mask[0] = maskTail64bit + (maskTail64bit << FIRST_16BIT_OFFSET_MM_API) +
+              (maskTail64bit << SECOND_16BIT_OFFSET_MM_API) + (maskTail64bit << THIRD_16BIT_OFFSET_MM_API);
     mask[1] = mask[0];
     int stride = calcWidth * (c0Size_ * sizeof(TransT) / DEFAULT_C0_SIZE);
     int32_t totalRep = CeilT<int32_t>(height, REPEAT_BLOCK_NUM_MM_API);