I serialized and saved TensorRT int8 engine file for 1080ti and 2080ti. For 1080ti, engine file equals 65 Mb, for 2080ti engine file size equals 125 mb. GPU architecture different but, why models have different size ?
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| different tensorRT vesrion int8 engine is not the same | 2 | 1062 | January 25, 2019 | |
| int8 mode is different between 1080ti with 2080ti | 0 | 586 | September 3, 2019 | |
| TRT engine file bigger than original Darknet YOLO V3 model | 3 | 696 | January 30, 2023 | |
| File size vs. batch size for TensorRT serialized engine files | 1 | 675 | September 30, 2020 | |
| The size of tensorrt's engine is different | 3 | 1321 | October 22, 2018 | |
| Difference between TRT engine file size for FP16 & 32 | 1 | 1298 | November 23, 2018 | |
| Serialized trt engine file is much larger than the original model file? | 1 | 1053 | September 18, 2020 | |
| The same onnx file produces engine files with different size under same configurations | 4 | 900 | May 12, 2022 | |
| Strange changes in file size when deployed with tensorrt | 9 | 1200 | May 30, 2026 | |
| TX2 NX ONNX Convert TensorRT Engine | 1 | 768 | March 29, 2021 |