test inference_api_test when run in windows-inference ci (#37710)

* test inference_api_test when run in windows-inference ci

* test if test failed the code run correctly

* put the failed test back

python/paddle/fluid/tests/unittests/ir/inference/test_trt_pool_op.py

@@ -27,9 +27,9 @@ from paddle.fluid.core import AnalysisConfig
 class TensorRTPoolTest(InferencePassTest):
     def setUp(self):
         self.bs = 1
-        self.channel = 3
-        self.height = 8
-        self.width = 8
+        self.channel = 2
+        self.height = 2
+        self.width = 2
         self.pool_size = 2
         self.pool_type = 'max'
         self.pool_stride = 1

tools/windows/run_unittests.sh

@@ -53,10 +53,6 @@ if [ -f "$PADDLE_ROOT/added_ut" ];then
         echo "========================================"
         exit 8;
     fi
-    if nvcc --version | grep 11.2; then
-        echo "Only test added_ut temporarily when running in CI-Windows-inference of CUDA 11.2."
-        exit 0;
-    fi
 fi
 set -e
 
@@ -107,7 +103,6 @@ disable_win_trt_test="^test_trt_convert_conv2d$|\
 ^test_trt_convert_emb_eltwise_layernorm$|\
 ^test_trt_convert_pool2d$|\
 ^test_trt_conv3d_op$|\
-^test_trt_matmul_quant_dequant$|\
 ^test_trt_subgraph_pass$|\
 ^test_trt_convert_dropout$|\
 ^test_trt_convert_hard_sigmoid$|\
@@ -121,6 +116,16 @@ disable_win_trt_test="^test_trt_convert_conv2d$|\
 ^test_trt_convert_matmul$|\
 ^test_trt_convert_scale$"
 
+# /*==================Fixed Disabled Windows GPU inference_api_test unittests==============================*/
+disable_win_inference_api_test="^test_analyzer_capi_exp_pd_config$|\
+^trt_quant_int8_yolov3_r50_test$|\
+^test_trt_dynamic_shape_ernie$|\
+^test_trt_dynamic_shape_ernie_fp16_ser_deser$|\
+^lite_resnet50_test$|\
+^test_trt_dynamic_shape_transformer_prune$|\
+^lite_mul_model_test$|\
+^paddle_infer_api_copy_tensor_tester$"
+
 # /*============================================================================*/
 
 # /*==================Fixed Disabled Windows CPU OPENBLAS unittests==============================*/
@@ -180,6 +185,7 @@ long_time_test="^test_gru_op$|\
 ^test_transformer$|\
 ^test_imperative_auto_mixed_precision$|\
 ^test_imperative_optimizer_v2$|\
+^test_trt_matmul_quant_dequant$|\
 ^test_strided_slice_op$"
 
 if [ ${WITH_GPU:-OFF} == "ON" ];then
@@ -331,6 +337,24 @@ function show_ut_retry_result() {
 set +e
 
 export FLAGS_call_stack_level=2
+
+if nvcc --version | grep 11.2; then
+    echo "Only test added_ut and inference_api_test temporarily when running in CI-Windows-inference of CUDA 11.2."
+    export CUDA_VISIBLE_DEVICES=0
+    tmpfile=$tmp_dir/$RANDOM
+    inference_api_test=^$(ls "paddle/fluid/inference/tests/api" | sed -n 's/\.exe$//pg' | awk BEGIN{RS=EOF}'{gsub(/\n/,"$|^");print}' | sed 's/|\^$//g')
+    (ctest -R "$inference_api_test" -E "$disable_win_inference_api_test" --output-on-failure -C Release -j 2 | tee $tmpfile ) &
+    wait;
+    collect_failed_tests
+    set -e
+    rm -f $tmp_dir/*
+    if [[ "$failed_test_lists" != "" ]]; then
+        unittests_retry
+        show_ut_retry_result
+    fi
+    exit 0;
+fi
+
 if [ "${WITH_GPU:-OFF}" == "ON" ];then
     run_unittest_gpu $cpu_parallel_job 10
     run_unittest_gpu $tetrad_parallel_job 4