PERF-2 P0: intra-step profiler — StepTimers on the overset solver (RTX_PROFILE; overlap build / predictors / patch BiCGSTAB / background Poisson with its setup-vs-iterate split / round exchange / apply / end exchange), PoissonSolution carries setup_ns and iterate_ns, the harness times advance / restore / snapshot / load sampling / force / FEA predictor and prints the wall split of the coupled phase; no clock is read when profiling is off; the reclassified/step progress denominator fixed (was ×4)

Co-Authored-By: Claude Fable 5.1 <[email protected]>
Claude-Session: https://claude.ai/code/session_01YJPeT6WA2e7YvAnS875AHL

crates/specialized/rtx-cfd/src/solvers/incompressible/poisson.rs

@@ -310,6 +310,10 @@ pub struct PoissonSolution {
     pub residual: f64,
     /// `residual < tolerance` at exit.
     pub converged: bool,
+    /// Wall time of the setup (hierarchy, fine level, components) [ns].
+    pub setup_ns: u64,
+    /// Wall time of the CG iteration (including the V-cycles) [ns].
+    pub iterate_ns: u64,
 }
 
 /// Which inner solver a projection uses for its pressure-correction system.
@@ -761,6 +765,7 @@ fn solve_pcg_with<T: MgScalar>(
         "invalid PoissonProblem: {:?}",
         problem.validate()
     );
+    let t_entry = std::time::Instant::now();
 
     let mut hier = Hierarchy::<T>::build(problem, params);
     let fine = Level::<f64>::new(problem.clone());
@@ -771,6 +776,8 @@ fn solve_pcg_with<T: MgScalar>(
             iterations: 0,
             residual: 0.0,
             converged: true,
+            setup_ns: t_entry.elapsed().as_nanos() as u64,
+            iterate_ns: 0,
         };
     }
 
@@ -818,6 +825,8 @@ fn solve_pcg_with<T: MgScalar>(
         |p: &[f64], r: &mut [f64], fine: &Level<f64>| -> f64 { fine.residual(&b, p, r) };
 
     let anchor = anchor.filter(|&a| a < n && fine.active[a]);
+    let setup_ns = t_entry.elapsed().as_nanos() as u64;
+    let t_iter = std::time::Instant::now();
     let finish = |p: &mut [f64], iterations: usize, residual: f64| {
         // Level of each singular component: the anchor's component is
         // shifted so p[anchor] == 0, every other singular component to mean
@@ -839,6 +848,8 @@ fn solve_pcg_with<T: MgScalar>(
             iterations,
             residual,
             converged: residual < tolerance,
+            setup_ns,
+            iterate_ns: t_iter.elapsed().as_nanos() as u64,
         }
     };