rename WITH_INFERENCE_NVTX to WITH_NVTX and fix compile bug (#55219)

* fix WITH_SHARED_IR option type

* rename WITH_INFERENCE_NVTX to WITH_NVTX and fix compile bug

* update

CMakeLists.txt

@@ -270,7 +270,7 @@ option(WITH_PSCORE "Compile with parameter server support" ${WITH_DISTRIBUTE})
 option(WITH_HETERPS "Compile with heterps" OFF)
 option(WITH_INFERENCE_API_TEST
        "Test fluid inference C++ high-level api interface" OFF)
-option(WITH_INFERENCE_NVTX "Paddle inference with nvtx for profiler" OFF)
+option(WITH_NVTX "Paddle with nvtx for profiler" OFF)
 option(PY_VERSION "Compile PaddlePaddle with python3 support" ${PY_VERSION})
 option(WITH_DGC "Use DGC(Deep Gradient Compression) or not" ${WITH_DISTRIBUTE})
 option(
@@ -661,6 +661,10 @@ if(WITH_MIPS)
   add_definitions(-DPADDLE_WITH_MIPS)
 endif()
 
+if(WITH_NVTX AND NOT WIN32)
+  add_definitions(-DPADDLE_WITH_NVTX)
+endif()
+
 if(WITH_LOONGARCH)
   set(WITH_XBYAK
       OFF

cmake/inference_lib.cmake

@@ -372,10 +372,6 @@ else()
   )
 endif()
 
-if(WITH_INFERENCE_NVTX AND NOT WIN32)
-  add_definitions(-DPADDLE_WITH_INFERENCE_NVTX)
-endif()
-
 copy(
   inference_lib_dist
   SRCS ${src_dir}/inference/capi_exp/pd_*.h ${paddle_inference_c_lib}

paddle/fluid/framework/CMakeLists.txt

@@ -664,37 +664,31 @@ cc_library(
   SRCS variable_helper.cc
   DEPS lod_tensor)
 
+set(NAIVE_EXECUTOR_DEPS
+    op_registry
+    denormal
+    device_context
+    scope
+    framework_proto
+    glog
+    lod_rank_table
+    feed_fetch_method
+    graph_to_program_pass
+    variable_helper)
+
 if(TENSORRT_FOUND)
-  cc_library(
-    naive_executor
-    SRCS naive_executor.cc
-    DEPS op_registry
-         denormal
-         device_context
-         scope
-         framework_proto
-         glog
-         lod_rank_table
-         feed_fetch_method
-         graph_to_program_pass
-         variable_helper
-         tensorrt_engine_op)
-else()
-  cc_library(
-    naive_executor
-    SRCS naive_executor.cc
-    DEPS op_registry
-         denormal
-         device_context
-         scope
-         framework_proto
-         glog
-         lod_rank_table
-         feed_fetch_method
-         graph_to_program_pass
-         variable_helper)
+  set(NAIVE_EXECUTOR_DEPS ${NAIVE_EXECUTOR_DEPS} tensorrt_engine_op)
+endif()
+
+if(WITH_NVTX AND NOT WIN32)
+  set(NAIVE_EXECUTOR_DEPS ${NAIVE_EXECUTOR_DEPS} cuda_profiler)
 endif()
 
+cc_library(
+  naive_executor
+  SRCS naive_executor.cc
+  DEPS ${NAIVE_EXECUTOR_DEPS})
+
 cc_library(
   executor_gc_helper
   SRCS executor_gc_helper.cc

paddle/fluid/framework/naive_executor.cc

@@ -28,7 +28,7 @@
 #ifdef PADDLE_WITH_TENSORRT
 #include "paddle/fluid/operators/tensorrt/tensorrt_engine_op.h"
 #endif
-#ifdef PADDLE_WITH_INFERENCE_NVTX
+#ifdef PADDLE_WITH_NVTX
 #include "paddle/fluid/platform/device/gpu/cuda/cuda_profiler.h"
 #endif
 #ifdef PADDLE_WITH_LITE
@@ -57,14 +57,14 @@ void NaiveExecutor::Run() {
   platform::RegisterModelLayout(ops_, place_);
 #endif
   platform::ScopedFlushDenormal flush;
-#ifdef PADDLE_WITH_INFERENCE_NVTX
+#ifdef PADDLE_WITH_NVTX
   platform::CudaNvtxRangePush("model", platform::NvtxRangeColor::Yellow);
 #endif
   for (auto &op : ops_) {
     VLOG(4) << std::this_thread::get_id() << " run "
             << op->DebugStringEx(scope_) << " on scope " << scope_;
     op->SetIsCalledByExecutor(false);
-#ifdef PADDLE_WITH_INFERENCE_NVTX
+#ifdef PADDLE_WITH_NVTX
     platform::CudaNvtxRangePush(op->Type() + "|" + op->OutputVars(true).front(),
                                 platform::NvtxRangeColor::Green);
 #endif
@@ -105,14 +105,14 @@ void NaiveExecutor::Run() {
       }
     }
 
-#ifdef PADDLE_WITH_INFERENCE_NVTX
+#ifdef PADDLE_WITH_NVTX
     platform::CudaNvtxRangePop();
 #endif
     for (auto &func : output_hookfuncs_) {
       func(op.get(), scope_);
     }
   }
-#ifdef PADDLE_WITH_INFERENCE_NVTX
+#ifdef PADDLE_WITH_NVTX
   platform::CudaNvtxRangePop();
 #endif
 }

paddle/fluid/inference/CMakeLists.txt

@@ -118,10 +118,6 @@ if(WITH_PSCORE)
   set(SHARED_INFERENCE_DEPS ${SHARED_INFERENCE_DEPS} fleet ps_service)
 endif()
 
-if(WITH_INFERENCE_NVTX AND NOT WIN32)
-  set(SHARED_INFERENCE_DEPS ${SHARED_INFERENCE_DEPS} cuda_profiler)
-endif()
-
 if(WITH_ONNXRUNTIME)
   set(SHARED_INFERENCE_SRCS
       ${SHARED_INFERENCE_SRCS}