app: multi_core_inverter: perf: move the stimulus Gating to outside the CurlStimulus

the region.contains() logic is much more expensive than the time bounds
check.
this gets an easy 50% perf boost to the ENTIRE simulation

crates/applications/multi_core_inverter/src/main.rs

@@ -137,30 +137,30 @@ struct Params {
     clock_decay: f32, // exp decay half-life
 }
 
-type HoldStim = CurlStimulus<Torus, Gated<f32>>;
-type ReleaseStim = CurlStimulus<Torus, Shifted<Gated<Exp<f32>>>>;
+type HoldStim = Gated<CurlStimulus<Torus, f32>>;
+type ReleaseStim = Shifted<Gated<Exp<CurlStimulus<Torus, f32>>>>;
 fn control_signal_hold(params: Params, region: Torus, cycle: u32, ty: DriveType) -> HoldStim {
     let area = region.cross_section();
     let amp = ty.direction() as f32 * params.input_magnitude / area;
-    let start = params.clock_phase_duration * cycle as f32;
-    // simple square wave:
-    let wave = Gated::new(amp, start, start + params.clock_phase_duration);
-    CurlStimulus::new(
+    let base_stim = CurlStimulus::new(
         region.clone(),
-        wave.clone(),
-    )
+        amp,
+    );
+    // simple square wave:
+    let start = params.clock_phase_duration * cycle as f32;
+    Gated::new(base_stim, start, start + params.clock_phase_duration)
 }
 
 fn control_signal_release(params: Params, region: Torus, cycle: u32, ty: DriveType) -> ReleaseStim {
     let area = region.cross_section();
     let amp = ty.direction() as f32 * params.input_magnitude / area;
-    let start = params.clock_phase_duration * cycle as f32;
-    // decaying exponential wave:
-    let wave = Exp::new_at(amp, start, params.clock_decay);
-    CurlStimulus::new(
+    let base_stim = CurlStimulus::new(
         region.clone(),
-        wave.clone(),
-    )
+        amp,
+    );
+    // decaying exponential wave:
+    let start = params.clock_phase_duration * cycle as f32;
+    Exp::new_at(base_stim, start, params.clock_decay)
 }
 
 
@@ -393,7 +393,7 @@ fn main() {
     assert_eq!(num_cores, 5);
     let mut driver = driver.with_stimulus(stim);
 
-    let prefix = "out/applications/multi_core_inverter/13/";
+    let prefix = "out/applications/multi_core_inverter/14/";
     let _ = std::fs::create_dir_all(&prefix);
     driver.add_state_file(&*format!("{}state.bc", prefix), 6400);
     driver.add_serializer_renderer(&*format!("{}frame-", prefix), 3200, None);

crates/coremem/src/stim.rs

@@ -346,6 +346,17 @@ impl TimeVarying3 for Exp3 {
     }
 }
 
+impl<A: AbstractStimulus> AbstractStimulus for Exp<A> {
+    fn at(&self, t_sec: f32, pos: Meters) -> Fields {
+        let scale = (t_sec * -self.tau).exp();
+        let inner = self.amp.at(t_sec, pos);
+        Fields {
+            e: inner.e * scale,
+            h: inner.h * scale,
+        }
+    }
+}
+
 #[derive(Clone)]
 pub struct Gated<T> {
     inner: T,
@@ -380,6 +391,16 @@ impl<T: TimeVarying3> TimeVarying3 for Gated<T> {
     }
 }
 
+impl<T: AbstractStimulus> AbstractStimulus for Gated<T> {
+    fn at(&self, t_sec: f32, pos: Meters) -> Fields {
+        if (self.start..self.end).contains(&t_sec) {
+            self.inner.at(t_sec, pos)
+        } else {
+            Default::default()
+        }
+    }
+}
+
 #[derive(Clone)]
 pub struct Shifted<T> {
     inner: T,
@@ -405,6 +426,12 @@ impl<T: TimeVarying3> TimeVarying3 for Shifted<T> {
     }
 }
 
+impl<T: AbstractStimulus> AbstractStimulus for Shifted<T> {
+    fn at(&self, t_sec: f32, pos: Meters) -> Fields {
+        self.inner.at(t_sec - self.start_at, pos)
+    }
+}
+
 #[cfg(test)]
 mod test {
     use super::*;