WIP: colour NR v2 reads its env knobs once (per-pixel env reads made the linear mode 11x slower); joint stage timings with LIGHTCRAFT_PROFILE

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

crates/pipeline/src/local.rs

@@ -211,8 +211,20 @@ pub fn nr_v2_enabled() -> bool {
     std::env::var("LIGHTCRAFT_NR_V2").map_or(true, |v| v != "0")
 }
 
-fn knob(name: &str, default: f32) -> f32 {
-    std::env::var(name).ok().and_then(|v| v.parse().ok()).unwrap_or(default)
+/// A prototype tuning knob from the environment, read once per process (they are read in per-pixel code).
+fn knob(name: &'static str, default: f32) -> f32 {
+    use std::collections::HashMap;
+    use std::sync::{Mutex, OnceLock, PoisonError};
+    static READ: OnceLock<Mutex<HashMap<&'static str, Option<f32>>>> = OnceLock::new();
+    let map = READ.get_or_init(|| Mutex::new(HashMap::new()));
+    let mut m = map.lock().unwrap_or_else(PoisonError::into_inner);
+    m.entry(name).or_insert_with(|| std::env::var(name).ok().and_then(|v| v.parse().ok())).unwrap_or(default)
+}
+
+/// [`nrc_linear`], read once (it is asked per pixel).
+fn nrc_linear_cached() -> bool {
+    static ON: std::sync::OnceLock<bool> = std::sync::OnceLock::new();
+    *ON.get_or_init(|| knob("NRV2_LINEAR", 1.0) > 0.5)
 }
 
 /// Most samples the noise level of the new colour noise reduction is measured from. The CPU and
@@ -363,7 +375,7 @@ pub fn nrc_eps(samples: &[[f32; 3]], k: f32, pct: f32) -> [f32; 3] {
 /// kept exactly) with a threshold from a noise model `σ²(Y) = α·Y + β`, instead of square-root differences
 /// with one threshold for the whole image.
 pub fn nrc_linear() -> bool {
-    knob("NRV2_LINEAR", 1.0) > 0.5
+    nrc_linear_cached()
 }
 
 /// Experimental joint guide; kept opt-in while native-resolution CPU cost is evaluated.

crates/pipeline/src/nrc_joint.rs

@@ -231,9 +231,19 @@ fn denoise_grid(img: &mut Rgb32f, params: &NrcParams, step: usize, detail: f32,
             *p = p.map(|v| v / count.max(1.0));
         }
     });
+    let lap = |what: &str, t: &mut Option<std::time::Instant>| {
+        if let Some(t0) = t {
+            eprintln!("      joint {what}: {:.0} ms", t0.elapsed().as_secs_f64() * 1e3);
+            *t = Some(std::time::Instant::now());
+        }
+    };
+    let mut tt = start;
+    lap("transform", &mut tt);
     let mut noise = noise_model(&original);
+    lap("noise model", &mut tt);
     noise.tolerance = (1.5 - detail).clamp(0.5, 1.5);
     let guide = pilot(&original, &noise);
+    lap("pilot", &mut tt);
     if std::env::var_os("NRV2_DEBUG").is_some() {
         eprintln!("    nr joint: {w}x{h}, Y/R-Y/B-Y alpha {:?} beta {:?}, supported Y {:?}", noise.model.alpha, noise.model.beta, noise.ymax);
     }
@@ -305,6 +315,7 @@ fn denoise_grid(img: &mut Rgb32f, params: &NrcParams, step: usize, detail: f32,
             }
         });
         std::mem::swap(&mut paths, &mut scratch);
+        lap(&format!("scale {delta}"), &mut tt);
         delta = delta.saturating_mul(2);
     }
     drop(scratch);