ggml-org/whisper.cpp

NPU support in whisper.cpp

Open

#1,557 opened on 2023年11月27日

GitHub で見る
 (18 comments) (12 reactions) (0 assignees)C++ (49,693 stars) (5,535 forks)batch import
good first issueperformanceresearch🔬

説明

Christmas is coming soon, and I want to take some time to research something interesting, such as edge low-power inference. Although current whisper.cpp can run on Raspberry Pi, the inference performance cannot achieve real-time transcription. Fortunately, there are now some development boards that use processors with NPUs, which can be used to achieve real-time transcription of large models. My primary goal is to first support RK3566 and RK3588.

Roadmap:

  • MatMul offloading
  • Conv-Gelu offloading
  • LayerNorm offloading ...

Reference:

https://github.com/rockchip-linux/rknpu2

コントリビューターガイド

技術スタック
cpp
領域
machine learningperformance
Issue 種別
feature
難度新しい貢献者にとっての実装難度です。1 は非常に小さな変更、5 は専門的な作業です。
5
推定時間経験ある貢献者が調査、実装、テスト、pull request 準備にかけるおおよその時間範囲です。
over 1 week
活動状況Issue が今どれだけ取り組みやすいかを示します。新しい、活発、古い、ブロック中、またはメンテナー入力待ちなどです。
active
明確さ期待される変更、受け入れ条件、次の手順がどれだけ明確かを示します。
mostly clear
前提条件
C++ programmingNeural network inference basicsNPU architecture knowledgeRK3566/RK3588 familiarity
初心者向け度初回貢献者にどれだけ取り組みやすいかを 1-100 で推定したスコアです。
20
調査方針
The issue proposes adding NPU support for RK3566 and RK3588, with a roadmap for offloading MatMul, Conv Gelu, and LayerNorm. Research should focus on integrating with the Rockchip RKNN toolkit (rknpu2) and modifying the whisper.cpp backend to delegate these operations to the NPU. Investigate existing ggml backend implementations for similar hardware and adapt the build system. Review the linked discussion on Raspberry Pi for performance baselines.