I am using the cudnn convolutional operation and I was wondering if there is a configuration in the API to wrap instead of pad the horizontal component?
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| how to only pad the start when performing 1D convolution | 0 | 429 | February 19, 2019 | |
| Specify conv algorithm in backend API | 3 | 499 | October 26, 2023 | |
| Convolutions in cuDNN | 1 | 760 | July 8, 2018 | |
| How to set convolution math type in cuDNN 9.2.0? | 0 | 104 | June 16, 2024 | |
| For Multiple GPU, cudnn Convolution operation Api call by CPU is not completing parallelly(at the same time)) | 1 | 602 | September 27, 2022 | |
| Question about the "SAME" convolution in cuDNN | 0 | 459 | March 11, 2019 | |
| cuDNN convolution with pitched memory | 1 | 750 | June 6, 2022 | |
| What's the rules that cudnn convolution select fp32 vector core and tf32 tensor core? | 1 | 850 | March 31, 2023 | |
| Problem of fusing Convolution and BatchNorm | 0 | 611 | December 29, 2021 | |
| Dose cuDNN support operator fusion or graph fusion | 3 | 782 | October 12, 2021 |