Skip to content

perf: 5.9x speedup in procstat post-processing via metric caching #36

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
atheurer merged 1 commit into from
Jul 16, 2026
Merged

perf: 5.9x speedup in procstat post-processing via metric caching #36

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
31 changes: 23 additions & 8 deletions procstat-post-process.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,6 +27,8 @@ def process_interrupts(fork_idx, num_forks, log_file, cpu_topo):
prev_irq_counts = {}
first_cpu_idx = 0
last_cpu_idx = 0
cpu_topo_cache = {}
metric_idx_cache = {} # (irq, cpu) -> idx, avoids repeated label lookup

try:
fh, _ = open_read_text_file(log_file)
Expand Down Expand Up @@ -75,14 +77,27 @@ def process_interrupts(fork_idx, num_forks, log_file, cpu_topo):
time_diff_sec = (curr_timestamp_ms - prev_timestamp_ms) / 1000
if time_diff_sec > 0:
ints_sec = irq_diff / time_diff_sec
package, die, core, thread = get_cpu_topology(cpu, cpu_topo)
names = {
"package": package, "die": die, "core": core,
"thread": thread, "cpu": cpu, "irq": irq,
"type": irq_type, "desc": irq_desc,
}
sample = {"value": ints_sec, "end": curr_timestamp_ms}
metrics.log_sample(str(fork_idx), desc, names, sample)
cache_key = (irq, cpu)
if cache_key in metric_idx_cache:
metrics.log_sample_by_idx(
metric_idx_cache[cache_key], ints_sec, curr_timestamp_ms
)
elif ints_sec != 0:
# Skip zero-value cold-path registrations: a metric
# that is zero on its first occurrence will always be
# purged by finish_samples anyway. Registering it just
# to purge it wastes memory and finish_samples time.
if cpu not in cpu_topo_cache:
cpu_topo_cache[cpu] = get_cpu_topology(cpu, cpu_topo)
package, die, core, thread = cpu_topo_cache[cpu]
names = {
"package": package, "die": die, "core": core,
"thread": thread, "cpu": cpu, "irq": irq,
"type": irq_type, "desc": irq_desc,
}
sample = {"value": ints_sec, "end": curr_timestamp_ms}
idx = metrics.log_sample(str(fork_idx), desc, names, sample)
metric_idx_cache[cache_key] = idx

prev_irq_counts.setdefault(irq, {})[cpu] = curr_count

Expand Down
Loading