inference_results_v3.0
inference_results_v3.0 copied to clipboard
Running the benchmark on Xavier AGX
I am currently trying to execute the different benchmarks on the NVIDIA Jetson Xavier AGX.
I have successfully built the container with make prebuild
and make build
. This has launched the container and everything is fine. I then run python3 -m scripts.custom_systems.add_custom_system
to add the system, which I give "Xavier" as the custom name. This makes for the error that only Orin is supported currently. This error is by-passable by vimming into the Makefile.docker
and adding the following lines to the top:
IS_SOC=1
SOC_SM=87
As per the documentation, running make build
will build the harnesses, why I run the following command to execute a benchmark:
make run_harness RUN_ARGS="--benchmark=resnet50, scenarios=Offline"
This results in the following error:
Detected system ID: KnownSystem.Xavier
Traceback (most recent call last):
File "/usr/lib/python3.8/runpy.py", line 194, in _run_module_as_main
return _run_code(code, main_globals, None,
File "/usr/lib/python3.8/runpy.py", line 87, in _run_code
exec(code, run_globals)
File "/work/code/main.py", line 233, in <module>
main(main_args, DETECTED_SYSTEM)
File "/work/code/main.py", line 104, in main
load_config_fn(benchmarks, scenarios)
File "/work/code/main.py", line 54, in populate_config_registry
ConfigRegistry.load_configs(benchmark, scenario)
File "/work/configs/configuration.py", line 123, in load_configs
importlib.import_module(f"{base_module}.custom")
File "/usr/lib/python3.8/importlib/__init__.py", line 127, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
File "<frozen importlib._bootstrap>", line 1014, in _gcd_import
File "<frozen importlib._bootstrap>", line 991, in _find_and_load
File "<frozen importlib._bootstrap>", line 975, in _find_and_load_unlocked
File "<frozen importlib._bootstrap>", line 671, in _load_unlocked
File "<frozen importlib._bootstrap_external>", line 848, in exec_module
File "<frozen importlib._bootstrap>", line 219, in _call_with_frames_removed
File "/work/configs/bert/Offline/custom.py", line 8, in <module>
class ORIN(OfflineGPUBaseConfig):
File "/work/configs/configuration.py", line 207, in _do_register
raise KeyError("Config for {} is already registered.".format("/".join(map(str, keyspace))))
KeyError: 'Config for Benchmark.BERT/Scenario.Offline/KnownSystem.Orin/WorkloadSetting(HarnessType.Custom, AccuracyTarget.k_99, PowerSetting.MaxP) is already registered.'
make: *** [Makefile:45: run_harness] Error 1
This is somewhat odd, as I am specifically attempting to run the ResNet50 benchmark, however the error seems to stem from BERT
. Is there any way for me to, albeit in a hacky fashion, to actually execute the ResNet50 benchmark on the Xavier?