You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When running python train_DEFORM.py --DLO_type="DLO1" on the workstation, the terminal stays in an "evaluating" state for a long time. How can I speed up the computation to get training results faster?
By default, how long does it take to complete the calculations?
We tested the code on an AMD Threadripper Pro 7995 CPU, and the NumPy version runs at around 100 Hz. I also tried using an NVIDIA RTX 4090 GPU, and it is indeed slow.
We’re currently preparing a faster version of the code, and we aim to release it around mid-February.
We have a faster version (DEFT, available in another repo) specifically designed for branched DLOs. However, I understand migrating that version for single-branched DLOs may not be straightforward
Activity
YizhouChen7045 commentedon Jan 15, 2025
Hi,
Thank you for your interest in our work!
We tested the code on an AMD Threadripper Pro 7995 CPU, and the NumPy version runs at around 100 Hz. I also tried using an NVIDIA RTX 4090 GPU, and it is indeed slow.
We’re currently preparing a faster version of the code, and we aim to release it around mid-February.
BearSmith2024 commentedon Jan 16, 2025
Thank you for providing those details and your efforts in working towards an enhanced version of the code.
okmaker commentedon May 16, 2025
Hello, author! Your project is amazing. I've been studying DER and found this. Is the faster - version code ready? The current one is too slow.
YizhouChen7045 commentedon May 19, 2025
Thanks for your interest in our work!
We have a faster version (DEFT, available in another repo) specifically designed for branched DLOs. However, I understand migrating that version for single-branched DLOs may not be straightforward
Please allow me some time to work on this.
okmaker commentedon May 20, 2025
@YizhouChen7045
Thank you for your reply. I've been greatly inspired by your project and hope you will achieve more outstanding results in the future!