Add checks for correctness of vector lenghts in DL GEMM

include/ck/tensor_operation/gpu/device/impl/device_gemm_dl.hpp

@@ -273,6 +273,9 @@ struct DeviceGemmDl : public DeviceGemm<ALayout,
               block_2_ctile_map_{},
               M01_{M01},
               N01_{N01},
+              M_raw{M},
+              N_raw{N},
+              K_raw{K},
               a_element_op_{a_element_op},
               b_element_op_{b_element_op},
               c_element_op_{c_element_op}
@@ -314,6 +317,10 @@ struct DeviceGemmDl : public DeviceGemm<ALayout,
         index_t M01_;
         index_t N01_;
 
+        index_t M_raw;
+        index_t N_raw;
+        index_t K_raw;
+
         // TODO: unused since gridwise_gemm_dl_v1r3 does NOT support prologue for the time being.
         AElementwiseOperation a_element_op_;
         BElementwiseOperation b_element_op_;
@@ -485,6 +492,50 @@ struct DeviceGemmDl : public DeviceGemm<ALayout,
 
     static bool IsSupportedArgument(const Argument& arg)
     {
+        // Make sure that the M, N, K dimensions before padding are divisible by respective vector
+        // lengths.
+        if constexpr(is_same<tensor_layout::gemm::RowMajor, ALayout>::value)
+        {
+            constexpr auto A_K_vec_length =
+                ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1::At(0) *
+                ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1::At(3);
+            if(arg.K_raw % A_K_vec_length != 0)
+            {
+                return false;
+            }
+        }
+        else
+        {
+            constexpr auto A_M_vec_lenght =
+                ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1::At(1) *
+                ABlockTransferSrcVectorTensorLengths_K0_M0_M1_K1::At(2);
+            if(arg.M_raw % A_M_vec_lenght != 0)
+            {
+                return false;
+            }
+        }
+
+        if constexpr(is_same<tensor_layout::gemm::RowMajor, BLayout>::value)
+        {
+            constexpr auto B_N_vec_lenght =
+                BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1::At(1) *
+                BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1::At(2);
+            if(arg.N_raw % B_N_vec_lenght != 0)
+            {
+                return false;
+            }
+        }
+        else
+        {
+            constexpr auto B_K_vec_length =
+                BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1::At(0) *
+                BBlockTransferSrcVectorTensorLengths_K0_N0_N1_K1::At(3);
+            if(arg.K_raw % B_K_vec_length != 0)
+            {
+                return false;
+            }
+        }
+
         if(ck::get_device_name() == "gfx906" || ck::get_device_name() == "gfx1030" ||
            ck::get_device_name() == "gfx1100" || ck::get_device_name() == "gfx1101" ||
            ck::get_device_name() == "gfx1102")