Created by: TheodoreG

I saved a inference model using resnet50 from paddle model. But when I run following inference code:

model_path = "./resnet50_infer_mode   
place = fluid.CUDAPlace(0)
exe = fluid.Executor(place)
img_np_array = np.random.random(size=(1, 3, 224, 224)).astype('float32')
[inference_program, feed_target_names, fetch_targets] = fluid.io.load_inference_model(model_path, exe)
results = exe.run(inference_program,feed={feed_target_names[0]: img_np_array},fetch_list=fetch_targets)
print(results)

but I find that when LD_LIBRARY_PATH include only cudnn v6 library, it shows

*** Aborted at 1540288996 (unix time) try "date -d @1540288996" if you are using GNU date ***
PC: @                0x0 (unknown)
*** SIGSEGV (@0x0) received by PID 1274 (TID 0x7fcb0da1c700) from PID 0; stack trace: ***
    @     0x7fcb0d1eb160 (unknown)
    @                0x0 (unknown)
Segmentation fault

When LD_LIBRARY_PATH include cudnn v7, it acts normal. Could anyone explain that to me?