tackle some comments

ark/codegen.cpp

@@ -125,14 +125,21 @@ CodeGenerator::Impl::Impl(const Json &plan,
                     it++;
                     continue;
                 }
-                // no need to sync if range is a part of proc_range.
                 if (intersec.size() < range.size()) {
+                    // range is not a part of proc_range, so we need to
+                    // sync range here.
                     size_t state_id = get_state_id(range);
                     body_ss << sync_process_range(range, state_id);
                     if (intersec.size() < proc_range.size()) {
+                        // proc_range is not a part of range, so we need to
+                        // sync proc_range later.
                         need_sync = true;
                     }
                 } else {
+                    // intersec.size() == range.size(), which means that
+                    // range is a part of proc_range. In this case, we don't
+                    // need to sync range here, because we will sync
+                    // proc_range later by setting `need_sync` to true.
                     need_sync = true;
                 }
                 it = unsynced.erase(it);
@@ -373,7 +380,6 @@ std::string CodeGenerator::Impl::resource_group(
         //     if (task_id >= task_e) break;
         //     task_func(_buf, task_id, sram_per_warp);
         //   }
-        //   __syncthreads();
         // }
         // ```
         ss << "  ";

ark/gpu/gpu_kernel_test.cpp

@@ -13,35 +13,7 @@ ark::unittest::State test_gpu_kernel() {
     return ark::unittest::SUCCESS;
 }
 
-//
-// const std::string test_kernel_loop_void =
-//     "__device__ void ark_loop_body(char *_buf, int _iter) {\n"
-//     "  // Do nothing. Print iteration counter.\n"
-//     "  if (threadIdx.x == 0 && blockIdx.x == 0) {\n"
-//     "    if (_iter % 50 == 49) {\n"
-//     "      printf(\".\\n\");\n"
-//     "    } else {\n"
-//     "      printf(\".\");\n"
-//     "    }\n"
-//     "  }\n"
-//     "}\n";
-
-// ark::unittest::State test_gpu_loop_kernel() {
-//     int num_sm = ark::GpuManager::get_instance(0)->info().num_sm;
-//     ark::GpuLoopKernel glk(0, "test_kernel_loop_void", test_kernel_loop_void,
-//                            static_cast<size_t>(num_sm), 1, 0, 0);
-//     glk.compile();
-//     glk.load();
-
-//     glk.launch();
-//     glk.run(100);
-//     glk.stop();
-
-//     return ark::unittest::SUCCESS;
-// }
-
 int main() {
     UNITTEST(test_gpu_kernel);
-    // UNITTEST(test_gpu_loop_kernel);
     return 0;
 }