The speed of the new version of the export script has significantly decreased after converting the exported ONNX to an engine

[txzshine]

Hello, sorry to trouble you

The model I use is trained based on yolov8s-seg

Using the old version of the export script, with the submission ID: d770e58, export the ONNX file, and then convert it to an engine. The frame rate is 21 FPS

When using the latest version of the export script to export ONNX, there is a warning:
[W shape_type_inference.cpp:1968] Warning: The shape inference of TRT::EfficientNMSX_TRT type is missing, so it may result in wrong shape inference for the exported graph. Please consider adding it in symbolic function. (function UpdateReliable)

There is a warning when converting to engine runtime:
WARNING: [TRT]: TensorRT encountered issues when converting weights between types and that could affect accuracy. WARNING: [TRT]: If this is not the desired behavior, please modify the weights or retrain with regularization to adjust the magnitude of the weights. WARNING: [TRT]: Check verbose logs for the list of affected weights. WARNING: [TRT]: - 84 weights are affected by this issue: Detected subnormal FP16 values. WARNING: [TRT]: - 31 weights are affected by this issue: Detected values less than smallest positive FP16 subnormal value and converted them to the FP16 minimum subnormalized value. 0:14:26.183229981 3939 0xaaaabd64a720 INFO nvinfer gstnvinfer.cpp:682:gst_nvinfer_logger:<primary-inference> NvDsInferContext[UID 1]: Info from NvDsInferContextImpl::buildModel() <nvdsinfer_context_impl.cpp:2141> [UID = 1]: serialize cuda engine to file: /opt/nvidia/deepstream/deepstream-7.0/sources/deepstream_python_apps_1.2.0/apps/txz_deepstream/models/yolov11/seg-coal/coal5-seg.onnx_b1_gpu0_fp16.engine successfully WARNING: [TRT]: The getMaxBatchSize() function should not be used with an engine built from a network created with NetworkDefinitionCreationFlag::kEXPLICIT_BATCH flag. This function will always return 1.

With exactly the same operation, the new version of the export script exports an ONNX conversion engine that only runs at 10fps

What is the cause of this issue, and how can it be resolved?