Fact · Dictation on Windows
The whisper.cpp README says that with a dedicated NPU, it can run the Whisper encoder on the NPU through VitisAI, which gives a significant speedup over CPU-only inference.
Checked against its source on October 16, 2026
In the source's own words
“with a dedicated NPU, whisper.cpp can fully offload the Whisper encoder to the NPU via VitisAI, delivering significant speedup over CPU-only inference”
whisper.cpp README, Georgi Gerganov, GitHub, 2026, README, NPU section (VitisAI).
Open the original- What it measures
- The README lists Windows 11 and Ubuntu 24.04 as supported platforms for this path and asks you to install an XRT runtime and FlexML first.
- What it does not show
- This path needs specific AMD hardware and a multi-step setup. The README does not give a speedup factor in the sentence quoted here.