Skip to content

gracefully handle and warn about zero-gap between events for frequency estimate #87

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
WarningRan merged 1 commit into from
Mar 25, 2026
Merged

gracefully handle and warn about zero-gap between events for frequency estimate #87

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 14 additions & 1 deletion src/aiu_trace_analyzer/pipeline/normalize.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -106,6 +106,13 @@ def __init__(self, soc_frequency: float, ignore_crit: bool = False,
data={"count": 0},
is_error=True
),
TraceWarning(
name="zero_gap_time",
text="Detected {d[count]} identical timestamps between consecutive events. "
"This suggests a timestamp issue in the trace and "
"causes unreliable interval-based frequency calculation.",
data={"count": 0}
),
TraceWarning(
name="epoch_start",
text="OVC: reference epoch had to be updated {d[count]} times with newly detected earlier timestamps. "
Expand Down Expand Up @@ -338,7 +345,13 @@ def frequency_stats(self, event: TraceEvent) -> None:
if self.prev_event_data[qid][self._INTERVAL_KEY].count > 0:
gap_cycles = int(event["args"][ts_a]) - self.prev_event_data[qid][self._INTERVAL_KEY].get_start_cycle()
gap_time = event["ts"] - self.prev_event_data[qid][self._INTERVAL_KEY].get_start_ts()
gap_freq = float(gap_cycles) / gap_time
# Handle zero gap_time to avoid division by zero
if gap_time > 0:
gap_freq = float(gap_cycles) / gap_time
else:
# If gap_time is zero, fall back to duration-based frequency
self.warnings["zero_gap_time"].update()
gap_freq = dur_freq
else:
gap_freq = dur_freq
self.prev_event_data[qid][self._INTERVAL_KEY].update(
Expand Down