Janusz Lisiecki
Janusz Lisiecki
The fix has been merged. Please check the next nightly build or 1.39 release once it is available.
1.39 and 1.40 are public. Closing.
Hi @runitao, Thank you for reaching out and for the complete repro. I can confirm it is a bug. I will get back to you soon once I learn more.
https://github.com/NVIDIA/DALI/pull/5682 should provide a proper fix.
Hi @quanvuong, Thank you for reaching out. Currently, the exact functionality you are asking for is not fully exposed. What you can try out is: - instantiate the iterator with...
Hi @quanvuong, > (on a 8 H100 node, starting the all the worker pools can takes 20 minutes). This is surprising and not expected. Can you share[ a self contained...
Hi @quanvuong, Can you bisect which part of the external source callback is slow (start with the callback that just calls `np.ones` for example and then gradually extend it adding...
Hi @ivandariojr, Thank you for reaching out. The utilization plots you showed are good place to start more thorough analysis. I recommend capturing the profile using [NSIght System](https://developer.nvidia.com/nsight-systems/) to learn...
Hi @sdpetrides, Thank you for checking this out. Does it happen for all your video files or only some? Does it repro with any of the files from https://github.com/NVIDIA/DALI_extra/tree/main/db/video?
Hi @Ash-one, Thank you for reaching out. Regarding access, I think using tar would be beneficial here as it minimizes the number of file access operations. Since you want to...