Skip to content

Pin audiodsp f763ec3: a compiled graph renders on the audio pump's core - #175

Merged
bdbarnett merged 2 commits into
mainfrom
audiodsp-graph-on-pump
Oct 27, 2026
Merged

bdbarnett merged 2 commits into
mainfrom
audiodsp-graph-on-pump

Conversation

@bdbarnett

@bdbarnett bdbarnett commented Oct 27, 2026 •

Copy link
Copy Markdown
Contributor

Moves the audiodsp pin to f763ec3 (PyDevices/audiodsp#317). An audiograph.Graph can now be played as a source the audio pump pulls, so on esp32 it renders on the pump's core rather than the interpreter's, and set() reaches it through a queue drained at block boundaries.

Checked on an ESP32-P4 board with the WIFI6_DEV_KIT_H2 variant, built from the audiodsp branch that merged as f763ec3: a two-voice Poly6 played for 32 minutes with the LVGL panel open and a WebSocket client changing parameters, and no audio block was starved.

The second commit fixes CI. ubuntu-latest moved from 24.04 to 26.04, but the cached CircuitPython toolchain (deps/circuitpython-venv, ~/.espressif) was still keyed only on deps.lock, so the new image restored a venv built for the old image's Python. CircuitPython's make then failed with No module named 'minify_html'. Both that cache and the firmware release's cache now include the runner's Python version in their key.

@bdbarnett
bdbarnett merged commit e0d43a3 into main Oct 27, 2026
27 checks passed
@bdbarnett
bdbarnett deleted the audiodsp-graph-on-pump branch October 27, 2026 06:06
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Development

Successfully merging this pull request may close these issues.

1 participant