As shown in NVIDIA Data Center Deep Learning Product Performance, the latest performance of Resnet50v1.5 in MxNet reached 2751img/sec. But it doesn’t indicate the version of the container and server. And the link of ResNet-50 v1.5 is not updated., so I tried the lastest mxnet:21.02-py3 which reached 1734 img/sec in a single A100 as issue shows in DeepLearningExamples.So I’m confused whether the hardware I’m using is too different or the NGC version is not the right one.
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Status of fp16 resnet-50 training? | 2 | 1363 | October 12, 2021 | |
| Containers v17.12 Now Available! | 0 | 1079 | December 6, 2017 | |
| Maximum Performance of ResNet50 model for NVIDIA T4 in TensorRT using trtexec | 1 | 438 | October 5, 2020 | |
| Xavier NX inference speed | 5 | 1877 | March 9, 2022 | |
| Training 3x Slower with nvcr.io/nvidia/pytorch:23.08-py3 Compared to 22.04-py3 | 0 | 34 | September 11, 2024 | |
| Docker DeepStream 5.0 runs quite slow | 4 | 1136 | October 12, 2021 | |
| Very slow 1.8s/image on Faster RCNN ResNet101 and wrongly perceived 1 image as 153216 images !!! | 0 | 835 | July 11, 2019 | |
| Fujitsu Breaks ImageNet Record with V100 Tensor Core GPUs | 0 | 228 | August 21, 2022 | |
| Containers v19.01 Now Available! | 0 | 699 | January 25, 2019 | |
| Containers v18.10 Now Available! | 0 | 796 | October 31, 2018 |