NVIDIA / NVIDIA/DALI

How many videos may I decode with one GPU at the same time ?

Open
#2,231 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

enhancement question Video
Dominant language
C++
Stars
5.8k
Forks
678
Avg merge
3d 1h
Merged PRs (30d)
27

Description

I'm tryint to use DALI to accelerate my project. There are 6 videos to decode and each video contians 30 frames.

If I build the pipe and decode them one by one, it takes around 4.8 seconds. Then I set both batch size and num_threads to 6. The running time becomes 2.8 seconds.

It seems that only copy time is saved and the GPU still decodes the videos sequentially.

I wonder if I can decode all the 6 videos parallelly? Then the processing time will be around 4.8/6 ≈ 0.8 second?

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with DALI's video decoding pipeline and the reported batch size and num_threads settings. Compare the one-by-one and batched decoding cases described in the issue, then determine whether parallel GPU decoding is supported. Done means documenting the observed concurrency behavior and any supported configuration or limitation.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp, python
Domain
machine-learning, performance
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.