| [01/04/2024-16:29:12] [W] [TRT] onnx2trt_utils.cpp:375: Your ONNX model has been generated with INT64 weights, while TensorRT does not natively support INT64. Attempting to cast down to INT32. | |
| [01/04/2024-16:29:12] [W] [TRT] onnx2trt_utils.cpp:403: One or more weights outside the range of INT32 was clamped | |
| [01/04/2024-16:29:16] [W] [TRT] Calibrator won't be used in explicit precision mode. Use quantization aware training to generate network with Quantize/Dequantize nodes. | |
| [01/04/2024-17:18:11] [W] * Throughput may be bound by Enqueue Time rather than GPU Compute and the GPU may be under-utilized. | |
| [01/04/2024-17:18:11] [W] If not already in use, --useCudaGraph (utilize CUDA graphs where possible) may increase the throughput. | |
| [01/04/2024-17:18:11] [W] * GPU compute time is unstable, with coefficient of variance = 2.14899%. | |
| [01/04/2024-17:18:11] [W] If not already in use, locking GPU clock frequency or adding --useSpinWait may improve the stability. | |