[ggmarshall]

Split out of #217 as requested

linear_slope_fit / linear_slope_diff: the Welford update loops did full-array arithmetic on the length-1 output views (mean[:] += ...), which allocates a heap temporary per sample per output. Indexing the outputs as scalars (mean[0]) removes the allocations: 10× per processor call (173 → 17 µs/waveform) on real L200 p18 waveform blocks, benchmarked through the production build_dsp path. Since these processors are called many times per germanium chain, this is the bulk of the whole-chain speedup measured on #217 (ICPC 11.2 s → 6.9 s per channel-file for the full stack).

Numerical note: linear_slope_diff outputs, and linear_slope_fit's mean/std on near-zero float32 data, shift within the last float32 ulps relative to the old array-expression path (≤1.5e-3 ADC absolute on L200 QC quantities, which are near-zero). Verified end-to-end by rebuilding a full 199-channel L200 raw file and comparing column-by-column against a stock-dspeed control run; all other outputs bit-identical. Full report: https://claude.ai/code/artifact/6d1c437c-c519-4797-a510-993d6aa80e98

🤖 Generated with Claude Code

[Copilot]

🟢 Approval recommended

The change is localized and consistent with existing gufunc scalar-output conventions (using [0] indexing) while preserving the prior computation flow.

Pull request overview

This PR removes per-sample heap allocations in the hot Welford update loops inside linear_slope_fit and linear_slope_diff by indexing the scalar (length-1) gufunc outputs as scalars (out[0]) instead of performing full-array arithmetic on them. This aligns with the PR’s stated goal of eliminating temporary arrays in the tight loop to achieve the reported processor-level speedup.

Changes:

• Update Welford mean/variance updates in linear_slope_fit to operate on mean[0]/stdev[0] scalars rather than array expressions.
• Update linear_slope_diff to compute residuals using slope[0]/intercept[0] and accumulate into mean[0]/rms[0] scalars.
• Replace in-place np.sqrt(out, out) usage with scalar out[0] = np.sqrt(out[0]) to avoid array-expression temporaries.

File summaries

File	Description
src/dspeed/processors/linear_slope_fit.py	Removes per-sample temporary allocations in Welford update loops by switching to scalar indexing on length-1 gufunc outputs/inputs.

Review details

• Files reviewed: 1/1 changed files
• Comments generated: 0
• 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]

Codecov Report

❌ Patch coverage is 0% with 10 lines in your changes missing coverage. Please review.
✅ Project coverage is 66.74%. Comparing base (7dc457e) to head (a5d6c59).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/dspeed/processors/linear_slope_fit.py	0.00%	10 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #219   +/-   ##
=======================================
  Coverage   66.74%   66.74%           
=======================================
  Files          61       61           
  Lines        4066     4066           
=======================================
  Hits         2714     2714           
  Misses       1352     1352           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.