diff --git a/cmake/package_lite.cmake b/cmake/package_lite.cmake
index 2ad6cd4f9e0..b1f79f02010 100644
--- a/cmake/package_lite.cmake
+++ b/cmake/package_lite.cmake
@@ -6,6 +6,7 @@ set(CODEGEN_ROOT_DIR ${RUNTIME_PKG_NAME}/tools/codegen)
 set(CONVERTER_ROOT_DIR ${RUNTIME_PKG_NAME}/tools/converter)
 set(OBFUSCATOR_ROOT_DIR ${RUNTIME_PKG_NAME}/tools/obfuscator)
 set(CROPPER_ROOT_DIR ${RUNTIME_PKG_NAME}/tools/cropper)
+set(TEST_CASE_DIR ${TOP_DIR}/mindspore/lite/test/build)
 
 if(SUPPORT_TRAIN)
     set(RUNTIME_DIR ${RUNTIME_PKG_NAME}/train)
@@ -195,6 +196,10 @@ if(PLATFORM_ARM64)
     if(MSLITE_ENABLE_TOOLS)
         install(TARGETS ${BENCHMARK_NAME} RUNTIME DESTINATION ${BENCHMARK_ROOT_DIR} COMPONENT ${RUNTIME_COMPONENT_NAME})
     endif()
+    if(MSLITE_ENABLE_TESTCASES)
+        install(FILES ${TOP_DIR}/mindspore/lite/build/test/lite-test DESTINATION ${TEST_CASE_DIR}
+                COMPONENT ${RUNTIME_COMPONENT_NAME})
+    endif()
 elseif(PLATFORM_ARM32)
     if(SUPPORT_NPU)
         install(FILES ${DDK_LIB_PATH}/libhiai.so DESTINATION ${RUNTIME_DIR}/third_party/hiai_ddk/lib
diff --git a/mindspore/lite/test/run_ut_gpu.sh b/mindspore/lite/test/run_ut_gpu.sh
index 6911619d99f..1b3011e5057 100644
--- a/mindspore/lite/test/run_ut_gpu.sh
+++ b/mindspore/lite/test/run_ut_gpu.sh
@@ -6,10 +6,6 @@ echo ${basepath}
 # Example:sh run_ut_gpu.sh -r /home/temp_test -d "8KE5T19620002408"
 while getopts "r:d:" opt; do
     case ${opt} in
-        r)
-            lite_test_path=${OPTARG}
-            echo "lite_test_path is ${OPTARG}"
-            ;;
         d)
             device_id=${OPTARG}
             echo "device_id is ${OPTARG}"
@@ -32,9 +28,7 @@ echo 'run gpu ut logs: ' > ${run_gpu_ut_log_file}
 ut_gpu_config=${basepath}/ut_gpu.cfg
 
 function Run_gpu_ut() {
-    cd ${lite_test_path} || exit 1
-
-    cp -a ${lite_test_path}/lite-test ${ut_test_path}/lite-test || exit 1
+    cp -a ${basepath}/build/lite-test ${ut_test_path}/lite-test || exit 1
     cp -r ${basepath}/ut/src/runtime/kernel/opencl/test_data ${ut_test_path} || exit 1
 
     # adb push all needed files to the phone
diff --git a/mindspore/lite/test/ut_gpu.cfg b/mindspore/lite/test/ut_gpu.cfg
index e716bd543c6..6c7bb24a750 100644
--- a/mindspore/lite/test/ut_gpu.cfg
+++ b/mindspore/lite/test/ut_gpu.cfg
@@ -1,4 +1,5 @@
 TestOpenCL_Transpose.*
+TestOpenCL_DepthwiseConv2d.*
 TestOpenCL_StridedSlice.1D
 TestOpenCL_StridedSlice.2D
 TestOpenCL_StridedSlice.3D
@@ -6,7 +7,6 @@ TestOpenCL_StridedSlice.4D
 TestOpenCL_StridedSlice.4D_stride2
 TestOpenCL_StridedSlice.4D_to_3D
 TestOpenCL_StridedSlice.test1
-TestOpenCL_Stack.*
 TestOpenCL_Split.input3_axis0
 TestOpenCL_DepthToSpace.*
 TestOpenCL_SpaceToDepth.*
@@ -36,8 +36,6 @@ TestOpenCL_Conv2D.test1
 TestOpenCL_Conv2D.test2
 TestOpenCL_Conv2D.test3
 TestOpenCL_Conv2D.test3_batch2
-TestOpenCL_Concat.*
-TestOpenCL_BatchNorm.*
 TestOpenCL_BatchToSpaceND.*
 TestOpenCL_Arithmetic.ElementwiseAdd
 TestOpenCL_Arithmetic.ScalarMul
@@ -48,5 +46,13 @@ TestOpenCL_Arithmetic.BroadcastFloorMod
 TestOpenCL_Arithmetic.FloorMod
 TestOpenCL_Arithmetic.ElementwiseDiv
 TestOpenCL_ArithmeticSelf.*
-TestOpenCL_ArgMinMax.*
 TestOpenCL_Activation.*
+TestOpenCL_Concat.input2_axis0
+TestOpenCL_Concat.input2_axis0_shape1
+TestOpenCL_Concat.input2_axis1_Align
+TestOpenCL_Concat.input6_axis1_Align
+TestOpenCL_Concat.input6_axis2_Align
+TestOpenCL_Concat.input3_axis1_UnAlign
+TestOpenCL_Concat.input4_axis3_UnAlign
+TestOpenCL_Concat.input5_axis3_UnAlign
+TestOpenCL_Concat.input6_axis3_UnAlign