Generated command for ssd-mobilenet fails

[mahmoodn]

Using Nvidia codes in V2.0, I see that the generated command for ssd-mobilenet doesn't work while make run works. In fact in the output of make run, I see the generated command. However, when I paste the command and try to run it, it fails. Please see the generated command below:

[2023-02-12 10:25:41,504 __init__.py:43 INFO] Running command: ./build/bin/harness_default --plugins="build/plugins/NMSOptPlugin/libnmsoptplugin.so" \
--logfile_outdir="/work/build/logs/2023.02.12-10.23.52/mahmood2022_TRT/ssd-mobilenet/Offline" \
--logfile_prefix="mlperf_log_" --performance_sample_count=1024 \
--gpu_batch_size=512 --map_path="data_maps/coco/val_map.txt" \
--tensor_path="build/preprocessed_data/coco/val2017/SSDMobileNet/int8_linear" \
--use_graphs=false --gpu_engines="./build/engines/mahmood2022/ssd-mobilenet/Offline/ssd-mobilenet-Offline-gpu-b512-int8.lwis_k_99_MaxP.plan" \
--mlperf_conf_path="measurements/mahmood2022_TRT/ssd-mobilenet/Offline/mlperf.conf" \
--user_conf_path="measurements/mahmood2022_TRT/ssd-mobilenet/Offline/user.conf" \
--max_dlas=0 --scenario Offline --model ssd-mobilenet --response_postprocess coco

However:

(mlperf) mahmood@mlperf-inference-mahmood-x86_64:/work$ ./build/bin/harness_default \
--plugins="build/plugins/NMSOptPlugin/libnmsoptplugin.so" \
--logfile_outdir="/work/build/logs/2022.07.19-11.08.40/mahmood2022_TRT/ssd-mobilenet/Offline" \
--logfile_prefix="mlperf_log_" --performance_sample_count=1024 --gpu_batch_size=512 \
--map_path="data_maps/coco/val_map.txt" --tensor_path="build/preprocessed_data/coco/val2017/SSDMobileNet/int8_linear" \
--use_graphs=false --gpu_engines="./build/engines/mahmood2022/ssd-mobilenet/Offline/ssd-mobilenet-Offline-gpu-b512-int8.lwis_k_99_MaxP.plan" \
--mlperf_conf_path="measurements/mahmood2022_TRT/ssd-mobilenet/Offline/mlperf.conf" \
--user_conf_path="measurements/mahmood2022_TRT/ssd-mobilenet/Offline/user.conf" \
--max_dlas=0 --scenario Offline --model ssd-mobilenet --response_postprocess coco
...
Starting running actual test.
LoadGen: Failed to open summary file.LoadGen: Failed to open detailed log file.LoadGen: Failed to open accuracy log file.LoadGen: Failed to open trace file.Finished running actual test.
Device Device:0 processed:
  Memcpy Calls: 0
  PerSampleCudaMemcpy Calls: 0
  BatchedCudaMemcpy Calls: 0
&&&& PASSED Default_Harness # ./build/bin/harness_default
terminate called without an active exception
Aborted (core dumped)

I have attached the full outputs. cmd.output.txt make.run.output.txt