Add memory index guard in wmma device ops (#667)

include/ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_d_wmma_cshuffle.hpp

@@ -505,6 +505,15 @@ struct GridwiseGemmMultipleD_k0mk1_k0nk1_mn_wmma_cshuffle
         }
 
         // TODO: also check validity of all components (blockwise-copy, threadwise-copy, etc)
+        constexpr long_index_t TwoGB = (long_index_t{1} << 31);
+
+        if(!(a_grid_desc_k0_m_k1.GetElementSpaceSize() * sizeof(ADataType) <= TwoGB &&
+             b_grid_desc_k0_n_k1.GetElementSpaceSize() * sizeof(BDataType) <= TwoGB &&
+             e_grid_desc_m_n.GetElementSpaceSize() * sizeof(EDataType) <= TwoGB))
+        {
+            return false;
+        }
+
         return true;
     }
 

include/ck/tensor_operation/gpu/grid/gridwise_gemm_wmma.hpp

@@ -264,6 +264,13 @@ struct GridwiseGemm_k0mk1_k0nk1_mn_wmma
         }
 
         // TODO: also check validity of all components (blockwise-copy, threadwise-copy, etc)
+        constexpr long_index_t TwoGB = (long_index_t{1} << 31);
+
+        if(!(a_grid_desc_k0_m_k1.GetElementSpaceSize() * sizeof(ADataType) <= TwoGB &&
+             b_grid_desc_k0_n_k1.GetElementSpaceSize() * sizeof(BDataType) <= TwoGB))
+        {
+            return false;
+        }
         return true;
     }