Add `"depleted"` operator trace handling

triton-inference-server / dali_backend

The Triton backend that allows running GPU-accelerated data pre-processing pipelines implemented in DALI's python API.

https://docs.nvidia.com/deeplearning/dali/user-guide/docs/index.html

MIT License

120 stars 28 forks source link

Add `"depleted"` operator trace handling #191

Closed szalpal closed 1 year ago

szalpal commented 1 year ago

[x] Wait for DALI 1.26 release

This PR duplicates #187 .

Should the VideoInput operator be properly working with ExternalSource in the same pipeline, DALI Backend needs to properly handle the "depleted" operator trace. This trace, defined in DALI, specifies when a given operator needs more data at the input. Refers to https://github.com/NVIDIA/DALI/pull/4794

dali-automaton commented 1 year ago

CI MESSAGE: [8550912]: BUILD STARTED

dali-automaton commented 1 year ago

CI MESSAGE: [8550912]: BUILD PASSED