Repository navigation
perf(processors): allocation-free early-exit contains_nan guard in all nopython kernels - #220
Merged
Merged
Conversation
There was a problem hiding this comment.
🟡 Changes recommended
contains_nan hard-codes Numba caching (cache=True) and therefore bypasses the existing DSPEED_CACHE/numba_defaults.cache control mechanism used elsewhere.
Once you've addressed the issues Copilot identified, you can request another Copilot review.
Pull request overview
This PR introduces a Numba-jitted utils.contains_nan() helper to provide an allocation-free, early-exit NaN scan, and uses it to replace np.isnan(...).any() guards inside many nopython processor kernels to reduce per-call overhead.
Changes:
- Add
contains_nan()insrc/dspeed/utils.pyas an optimized NaN guard for nopython kernels. - Replace
np.isnan(w).any()waveform/array guards across many processors withcontains_nan(...). - Update processor imports to use the new helper.
File summaries
| File | Description |
|---|---|
| src/dspeed/utils.py | Adds contains_nan() Numba helper for allocation-free NaN checks. |
| src/dspeed/processors/windower.py | Use contains_nan for waveform NaN guard. |
| src/dspeed/processors/wf_correction.py | Use contains_nan for waveform/correction NaN guards. |
| src/dspeed/processors/wf_alignment.py | Use contains_nan for waveform NaN guard. |
| src/dspeed/processors/upsampler.py | Use contains_nan for waveform NaN guards. |
| src/dspeed/processors/trap_filters.py | Use contains_nan for waveform NaN guards in trap-family processors. |
| src/dspeed/processors/time_point_thresh.py | Use contains_nan for waveform/threshold NaN guards. |
| src/dspeed/processors/time_over_threshold.py | Use contains_nan for waveform NaN guard. |
| src/dspeed/processors/sort.py | Use contains_nan for waveform NaN guard. |
| src/dspeed/processors/soft_pileup_corr.py | Use contains_nan for waveform NaN guard. |
| src/dspeed/processors/saturation.py | Use contains_nan for waveform NaN guard. |
| src/dspeed/processors/recursive_filter.py | Use contains_nan for input coefficient/waveform NaN guards. |
| src/dspeed/processors/rc_cr2.py | Use contains_nan for waveform NaN guards and output validation. |
| src/dspeed/processors/pulse_injector.py | Use contains_nan for waveform NaN guards. |
| src/dspeed/processors/presum.py | Use contains_nan for waveform NaN guard. |
| src/dspeed/processors/poly_fit.py | Use contains_nan for waveform/parameter NaN guards. |
| src/dspeed/processors/pole_zero.py | Use contains_nan for waveform NaN guards and output validation. |
| src/dspeed/processors/pmt_pulse_injector.py | Use contains_nan for waveform NaN guards. |
| src/dspeed/processors/pad.py | Use contains_nan for waveform-slice NaN guard. |
| src/dspeed/processors/moving_windows.py | Use contains_nan for waveform NaN guards. |
| src/dspeed/processors/min_max.py | Use contains_nan for waveform NaN guards. |
| src/dspeed/processors/log_check.py | Use contains_nan for waveform NaN guard. |
| src/dspeed/processors/linear_slope_fit.py | Use contains_nan for waveform NaN guards. |
| src/dspeed/processors/inl_correction.py | Use contains_nan for INL array NaN guard. |
| src/dspeed/processors/histogram.py | Use contains_nan for waveform NaN guards. |
| src/dspeed/processors/histogram_stats.py | Use contains_nan for weights NaN guards. |
| src/dspeed/processors/get_wf_centroid.py | Use contains_nan for waveform NaN guard. |
| src/dspeed/processors/get_multi_local_extrema.py | Use contains_nan for waveform NaN guard. |
| src/dspeed/processors/fixed_time_pickoff.py | Use contains_nan for waveform NaN guard. |
| src/dspeed/processors/bl_subtract.py | Use contains_nan for waveform NaN guard. |
| src/dspeed/processors/arithmetic.py | Use contains_nan for waveform NaN guard. |
Review details
- Files reviewed: 31/31 changed files
- Comments generated: 2
- Review effort level: Lite
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
Codecov Report❌ Patch coverage is Additional details and impacted files@@ Coverage Diff @@
## main #220 +/- ##
==========================================
+ Coverage 66.74% 66.79% +0.04%
==========================================
Files 61 61
Lines 4066 4111 +45
==========================================
+ Hits 2714 2746 +32
- Misses 1352 1365 +13 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
…ee contains_nan In nopython numba kernels np.isnan(w).any() heap-allocates a boolean array and always scans the full input. contains_nan() (new in utils) allocates nothing, vectorizes via a fixed-width inner loop, and exits at the first NaN. Applied to all nopython kernels (object-mode kernels keep numpy's C implementation). Outputs are bit-identical; ~2x on O(1)-work processors like fixed_time_pickoff, ~10% on scan-heavy ones. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…dd unit tests Addresses PR review: cache=True was hard-coded, bypassing the DSPEED_CACHE/DSPEED_BOUNDSCHECK mechanism used by all processors. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
ggmarshall
force-pushed
the
perf-contains-nan
branch
from
September 1, 2026 01:50
e59946a to
9bf5ae1
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Split out of #217 (per review request there) — middle of the stack. #219 has merged, so this is now a single commit on main: just the NaN-guard change. #217 is stacked on top of this.
Adds
utils.contains_nan()— an allocation-free, early-exit,@guvectorize-compatible NaN scan — and replaces thenp.isnan(w).any()guards in all nopython processor kernels with it.np.isnan(w).any()allocates a boolean array per call and always scans the full waveform;contains_nanallocates nothing, vectorizes, and exits at the first NaN. Measured on real L200 p18 waveform blocks: ~2× on processors whose real work is O(1) (pickoffs, min/max, etc.), ~1.1× on scan-heavy ones.Outputs are bit-identical — verified end-to-end by rebuilding a full 199-channel L200 raw file and comparing column-by-column against a stock-dspeed control run (report: https://claude.ai/code/artifact/6d1c437c-c519-4797-a510-993d6aa80e98).
🤖 Generated with Claude Code