TDA4VM: how to get the performance(FPS) about the TVM/NEO-AI-DLR + TIDL Heterogeneous Execution？

when i compile the model by tvm,and get the artifacts_**_target,include the deploy_lib.so and subgraph0_net.

and add the code as below on the 'test_tidl_j7_deploy.py'.

and run on evm ,get the time cost is below.

but the model run on C71/MMA(except the not support layer op) just cost  about ~10ms.

so that I doubt it is accurate.

and i would like to know how to test the whole process time cost? how to get the partial time on arm core, optimization by tvm ?

thanks for any advice.

root@j7-evm:/opt/tvm/tests/python/relay/ti_tests#
root@j7-evm:/opt/tvm/tests/python/relay/ti_tests# python3 test_tidl_j7_deploy.py
Input image file: ./test.jpg
2020-11-19 18:13:14,606 INFO Could not find libdlr.so in model artifact. Using dlr from /usr/lib/python3.8/site-packages/dlr/libdlr.so
2020-11-19 18:13:14,606 INFO Could not find libdlr.so in model artifact. Using dlr from /usr/lib/python3.8/site-packages/dlr/libdlr.so
[18:13:14] ../src/dlr_tvm.cc:66: No metadata found
APP: Init ... !!!
MEM: Init ... !!!
MEM: Initialized DMA HEAP (fd=6) !!!
MEM: Init ... Done !!!
IPC: Init ... !!!
IPC: Init ... Done !!!
REMOTE_SERVICE: Init ... !!!
REMOTE_SERVICE: Init ... Done !!!
APP: Init ... Done !!!
   248.343436 s:  VX_ZONE_INIT:Enabled
   248.343460 s:  VX_ZONE_ERROR:Enabled
   248.343465 s:  VX_ZONE_WARNING:Enabled
   248.346302 s:  VX_ZONE_INIT:[tivxInit:71] Initialization Done !!!
   248.348719 s:  VX_ZONE_INIT:[tivxHostInit:48] Initialization Done for HOST !!!
[18:13:15] ../src/dlr.cc:162: No metadata file was found!
time: 0.5211689472198486 second
ONNX_LaneDet dlr runtime execution finished
ONNX_LaneDet execution finished
[[[[0 0 0 ... 0 0 0]
   [0 0 0 ... 0 0 0]
   [0 0 0 ... 0 0 0]
   ...
   [0 0 0 ... 0 0 0]
   [0 0 0 ... 0 0 0]
   [0 0 0 ... 0 0 0]]]]
root@j7-evm:/opt/tvm/tests/python/relay/ti_tests#