fix: show Apple GPU power at idle, document the base-vs-Pro/Max CPU split

A base M4 looked like it "had no GPU" because we dropped a zero-valued GPU rail
and the M4's GPU idles at a few mW. The GPU energy channel is always valid on
Apple Silicon, so report power.gpu even at ~0W. Keep the >0 gate on CPU/ANE,
where 0 genuinely means unavailable on Pro/Max.

The CPU asymmetry the fleet showed is real, not our bug: base M-series expose
per-domain CPU power via IOReport, Pro/Max/Ultra zero the CPU/ANE channels in
both IOReport and powermetrics (verified: CPU Power: 0 mW on an M3 Max). So
power.cpu is a real number on base dies and an explained dash on Pro/Max. Wrote
it up in guide 13 with the tested SoC + macOS versions.

Author: kinncj

CHANGELOG.md

@@ -7,6 +7,18 @@ and the project follows [Semantic Versioning](https://semver.org/spec/v2.0.0.htm
 
 ## [Unreleased]
 
+## [2.4.3] - 2026-07-02
+
+### Fixed
+- **Apple GPU power is reported at idle instead of dropped.** The collector
+  skipped a zero-valued GPU rail, so a base M4 — whose GPU idles at a few mW —
+  looked like it "had no GPU", while a Pro/Max (with a small non-zero idle GPU)
+  showed one. The GPU energy channel is always valid on Apple Silicon, so
+  `power.gpu` is now reported even at ~0 W. The CPU asymmetry is *real* and
+  unchanged: base M-series expose per-domain CPU power, Pro/Max/Ultra do not
+  (verified — `powermetrics` also reports `CPU Power: 0 mW`), so `power.cpu`
+  stays Unavailable-with-reason there. Documented in the macOS guide.
+
 ## [2.4.2] - 2026-07-01
 
 ### Fixed

app/internal/helper/collect.go

@@ -143,15 +143,19 @@ func powerMetric(name string, w float64) domain.Metric {
 func assembleApplePower(cpu, gpu, ane, gpuUtil float64, ioOK bool, smcPkg float64, smcOK bool, pm []domain.Metric) []domain.Metric {
 	var out []domain.Metric
 	if ioOK {
+		// CPU and ANE: a 0 reading means the channel is *unavailable* on Pro/Max
+		// (verified — powermetrics also reports 0 there), so only report a real
+		// value; the caller fills the Unavailable-with-reason otherwise.
 		if cpu > 0 {
 			out = append(out, powerMetric("power.cpu", cpu))
 		}
-		if gpu > 0 {
-			out = append(out, powerMetric("power.gpu", gpu))
-		}
 		if ane > 0 {
 			out = append(out, powerMetric("power.npu", ane))
 		}
+		// GPU: the energy channel is always valid on Apple Silicon, so report it
+		// even at idle (~0 W) rather than dropping the rail — otherwise a base M4,
+		// whose GPU idles at a few mW, looks like it "has no GPU".
+		out = append(out, powerMetric("power.gpu", gpu))
 		if gpuUtil >= 0 {
 			out = append(out, domain.Metric{Name: "gpu.util", Unit: "percent", Status: domain.StatusOK, Gauge: gpuUtil})
 		}

app/internal/helper/smc_power_test.go

@@ -30,6 +30,16 @@ func TestAssembleApplePower_SMCWinsOverPhantomIOReport(t *testing.T) {
 	}
 }
 
+// GPU power is always a valid IOReport channel on Apple Silicon, so it must be
+// reported even at idle (~0 W) — dropping it made a base M4 look like it "has no
+// GPU" while a Pro/Max (whose CPU/ANE channels genuinely read 0) showed the GPU.
+func TestAssembleApplePower_ShowsIdleGPU(t *testing.T) {
+	got := byName(assembleApplePower(0.5, 0 /*gpu idle*/, 0, 3, true /*ioOK*/, 20, true, nil))
+	if m, ok := got["power.gpu"]; !ok || m.Status != domain.StatusOK || m.Gauge != 0 {
+		t.Fatalf("power.gpu = %+v (present=%v), want a 0 W reading at idle", m, ok)
+	}
+}
+
 // Without SMC, a sub-watt IOReport sum must not shadow a real powermetrics package
 // reading (the old mergeByName poisoning).
 func TestAssembleApplePower_PowermetricsBeatsPhantomIOReport(t *testing.T) {

docs/guides/13-privileged-macos.md

@@ -47,6 +47,36 @@ This ordering exists because of the Pro/Max quirk below.
 > counter at all — neither IOReport nor `powermetrics` reports it. `power.cpu`
 > reads `unavailable` there. A hardware limit, not a misconfiguration.
 
+## Why a base M4 shows CPU power but an M3 Max doesn't (and both show GPU)
+
+This trips people up, so to be explicit — it's **the chip tier, not the macOS
+version, and not a Heimdall bug**:
+
+| Rail | Base (M1–M4) | Pro / Max / Ultra |
+|---|---|---|
+| `power.cpu` | **reported** — IOReport exposes per-domain CPU energy | **`unavailable`** — the CPU (and ANE) energy channels read **0** |
+| `power.gpu` | reported (idles at a few mW) | reported |
+| `power.total` | SMC `PSTR`, whole-system | SMC `PSTR`, whole-system |
+
+On **Pro/Max/Ultra** dies the SoC's "Energy Model" simply does not surface
+per-domain **CPU** or **ANE** power — both IOReport *and* `powermetrics` return
+`CPU Power: 0 mW` even with the cores pegged. So `power.cpu` is honestly
+`unavailable` (`Pro/Max: no per-domain CPU power`), while the true whole-machine
+draw still comes through `power.total` (SMC). Base dies expose the per-domain CPU
+figure, so they show a real `power.cpu`.
+
+**GPU** power is a valid IOReport channel on *every* Apple Silicon SoC, so it is
+always reported — including a few milliwatts at idle. (Heimdall used to drop a
+zero-valued GPU rail, which made an idle base M4 look like it "had no GPU"; since
+v2.4.3 the idle GPU shows as `0 W` instead.)
+
+Verified across two SoCs and macOS builds:
+
+| Machine | SoC | macOS (Darwin) | `power.cpu` | `power.gpu` |
+|---|---|---|---|---|
+| Mac mini | Apple **M4** (base) | 26.6 / Darwin 25.6 | real (~0.5 W idle) | reported (idle → 0 W) |
+| MacBook Pro | Apple **M3 Max** | 27.0 / Darwin 27.0 | `unavailable` (Pro/Max) | reported |
+
 ## Build note — IOReport needs CGO
 
 IOReport (and SMC) are reached through cgo. The local `make build-tui` enables