Skip to content

audio tests: the real-clock tone tests hold on a busy host - #744

Merged
techomancer merged 1 commit into
mainfrom
audiotestfix
Nov 15, 2026
Merged

techomancer merged 1 commit into
mainfrom
audiotestfix

Conversation

@techomancer

Copy link
Copy Markdown
Owner

Under a loaded test run the thread standing in for the guest is starved, the card's ring runs dry and the output file has short gaps. The pitch meter counted cycles across the spliced gaps and read about 1% high; it now takes the median cycle length. The bound on frames heard allowed only one stretch of filter ringing; it now allows one per stretch of sound and the 0.5% the stream may stretch while following the card's clock.

Fixes the O2 (MACE) test and the same weakness in the Octane (RAD1) one.

Under a loaded test run the thread standing in for the guest is starved,
the card's ring runs dry and the output file has short gaps. The pitch
meter counted cycles across the spliced gaps and read about 1% high; it
now takes the median cycle length. The bound on frames heard allowed only
one stretch of filter ringing; it now allows one per stretch of sound and
the 0.5% the stream may stretch while following the card's clock.

Fixes the O2 (MACE) test and the same weakness in the Octane (RAD1) one.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
@techomancer
techomancer merged commit 83c8ddd into main Nov 15, 2026
9 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant