diff --git a/src/tim/vx/ops/custom_base.cc b/src/tim/vx/ops/custom_base.cc
index 2ff185d..d26aa13 100644
--- a/src/tim/vx/ops/custom_base.cc
+++ b/src/tim/vx/ops/custom_base.cc
@@ -100,6 +100,7 @@ vsi_bool op_compute(vsi_nn_node_t* self, vsi_nn_tensor_t** inputs,
       reinterpret_cast<CustomOpBase*>(self->nn_param.client_param);
 
   uint32_t param_num = op_this->param_list_.size();
+  uint32_t input_start = op_this->input_num_ + op_this->output_num_;
 
   std::vector<tim::vx::DataType> input_types;
   for (uint32_t i = 0; i < op_this->input_num_; i++) {
@@ -127,7 +128,7 @@ vsi_bool op_compute(vsi_nn_node_t* self, vsi_nn_tensor_t** inputs,
   snprintf(kernel->info.name, VX_MAX_KERNEL_NAME, "%s", op_this->func_name_);
   kernel->unique_id =
       std::hash<std::string>()(std::string(op_this->func_name_));
-  vx_param_description_t kernel_param_def[param_num];
+  vx_param_description_t kernel_param_def[param_num + input_start];
 
   for (uint32_t i = 0; i < op_this->input_num_; i++) {
     kernel_param_def[i] = {VX_INPUT, VX_TYPE_TENSOR,
@@ -145,7 +146,7 @@ vsi_bool op_compute(vsi_nn_node_t* self, vsi_nn_tensor_t** inputs,
 
   kernel->info.parameters = kernel_param_def;
   kernel->info.enumeration = KERNEL_ID_PLACEHOLDER;
-  kernel->info.numParams = param_num;
+  kernel->info.numParams = param_num + input_start;
   kernel->info.initialize =
       reinterpret_cast<vx_kernel_initialize_f>(op_this->init_kernel_);
 
@@ -162,11 +163,10 @@ vsi_bool op_compute(vsi_nn_node_t* self, vsi_nn_tensor_t** inputs,
 
   auto node = vsi_nn_KernelCreateNodeExt(self->graph, kernel, resource);
   if (node) {
-    uint32_t input_start = op_this->input_num_ + op_this->output_num_;
 
     std::vector<vsi_nn_kernel_node_param_t> node_params(param_num + input_start);
     vsi_nn_kernel_node_param_t* node_params_ptr = node_params.data();
-    vsi_nn_kernel_node_pack_io(node_params_ptr, param_num, inputs,
+    vsi_nn_kernel_node_pack_io(node_params_ptr, param_num + input_start, inputs,
                                op_this->input_num_, outputs,
                                op_this->output_num_);
 
@@ -196,7 +196,7 @@ vsi_bool op_compute(vsi_nn_node_t* self, vsi_nn_tensor_t** inputs,
     }
 
     input_start = op_this->input_num_ + op_this->output_num_;
-    status = vsi_nn_KernelNodePassParam(node, node_params_ptr, param_num);
+    status = vsi_nn_KernelNodePassParam(node, node_params_ptr, param_num + input_start);
     for (uint32_t i = 0; i < param_num; i++) {
       vsi_nn_kernel_scalar_release(&node_params_ptr[input_start + i]);
     }