mitchell/hamlab
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases 14 Wiki Activity

GPUs: fan speed in % per card (nvidia-smi fan.speed, hwmon pwm1); 0% shows

Browse Source 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Mitchell-Johnson
2026-11-14 17:09:10 -08:00
co-authored by Claude Opus 5.5
parent 95bb026b7f
commit afd4688358
5 changed files with 37 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
internal/metrics/gpu.go
+17 -2
View File
@@ -49,6 +49,7 @@ type GPU struct {
PowerW float64 `json:"power_w,omitempty"`
PowerCapW float64 `json:"power_cap_w,omitempty"`
FanRPM int `json:"fan_rpm,omitempty"`
FanPct *float64 `json:"fan_pct,omitempty"` // fan speed as a % of its range: nvidia-smi's fan.speed, or the hwmon's pwm1; nil when the card has no such reading (0 is a stopped fan)
MHz float64 `json:"mhz,omitempty"` // actual clock (0 when power-gated)
ReqMHz float64 `json:"req_mhz,omitempty"` // what the driver asked for
MaxMHz float64 `json:"max_mhz,omitempty"`
@@ -479,6 +480,14 @@ func (d *gpuDev) read(x *GPU, now time.Time) {
x.PowerCapW = p / 1e6
}
x.FanRPM = int(readF(d.hwmon + "/fan1_input"))
if pwm, err := strconv.ParseFloat(strings.TrimSpace(readStr(d.hwmon+"/pwm1")), 64); err == nil {
top := readF(d.hwmon + "/pwm1_max")
if top <= 0 {
top = 255
}
v := round1(pwm / top * 100)
x.FanPct = &v
}
}
d.prevAt = now
}
@@ -762,7 +771,7 @@ func ueventValue(p, key string) string {
// sample (~20 ms and ~75 ms on a GTX 965M), no root needed. The PCI address selects the
// card, so two NVIDIA GPUs stay apart.
func (d *gpuDev) readNvidia(x *GPU) {
if out, ok := runOut("nvidia-smi", "--id="+d.bus, "--query-gpu=utilization.gpu,utilization.encoder,utilization.decoder,memory.used,memory.total,clocks.sm,clocks.max.sm,temperature.gpu,power.draw,power.limit", "--format=csv,noheader,nounits"); ok {
if out, ok := runOut("nvidia-smi", "--id="+d.bus, "--query-gpu=utilization.gpu,utilization.encoder,utilization.decoder,memory.used,memory.total,clocks.sm,clocks.max.sm,temperature.gpu,power.draw,power.limit,fan.speed", "--format=csv,noheader,nounits"); ok {
parseNvidiaQuery(x, out)
}
if out, ok := runOut("nvidia-smi", "pmon", "-c", "1", "-s", "um", "--id="+d.bus); ok {
@@ -781,7 +790,8 @@ func nvNum(s string) (float64, bool) {
}
// parseNvidiaQuery fills a GPU from one --query-gpu line in the field order readNvidia asks
// for: gpu%, encoder%, decoder%, mem used MiB, mem total MiB, SM MHz, max SM MHz, °C, W, W cap.
// for: gpu%, encoder%, decoder%, mem used MiB, mem total MiB, SM MHz, max SM MHz, °C, W, W cap,
// fan % ("[N/A]" on a card with no fan of its own: a laptop's, a passive datacentre card).
func parseNvidiaQuery(x *GPU, line string) {
f := strings.Split(strings.TrimSpace(line), ",")
if len(f) < 10 {
@@ -817,6 +827,11 @@ func parseNvidiaQuery(x *GPU, line string) {
if v, ok := nvNum(f[9]); ok {
x.PowerCapW = v
}
if len(f) > 10 {
if v, ok := nvNum(f[10]); ok {
x.FanPct = &v
}
}
}
// parseNvidiaPmon reads `nvidia-smi pmon -c 1 -s um`: a header naming the columns, then one
internal/metrics/gpu_test.go
+14
View File
@@ -182,6 +182,20 @@ func TestParseNvidiaQuery(t *testing.T) {
if g.Temp != 57 || g.TempSrc != "gpu" || g.PowerW != 23.4 || g.PowerCapW != 0 {
t.Errorf("temp/power (cap must stay 0 for [N/A]): %+v", g)
}
if g.FanPct != nil {
t.Errorf("no fan field must leave the fan unset: %+v", g)
}
var fan, nofan GPU
parseNvidiaQuery(&fan, "99, 0, 0, 20000, 24576, 1695, 2100, 83, 403.99, 420.00, 34\n")
parseNvidiaQuery(&nofan, "5, 0, 0, 500, 2048, 300, 1097, 45, 8.0, [N/A], [N/A]\n")
if fan.FanPct == nil || *fan.FanPct != 34 || nofan.FanPct != nil {
t.Errorf("fan: %v, [N/A]: %v", fan.FanPct, nofan.FanPct)
}
var stopped GPU
parseNvidiaQuery(&stopped, "0, 0, 0, 500, 24576, 210, 2100, 35, 20.0, 420.00, 0\n")
if stopped.FanPct == nil || *stopped.FanPct != 0 {
t.Errorf("a stopped fan must read 0, not nothing: %v", stopped.FanPct)
}
var short GPU
parseNvidiaQuery(&short, "garbage")
if short.UsageSrc != "" {
internal/metrics/sensors.go
+3
View File
@@ -181,6 +181,9 @@ func flatten(s Sample, want func(id string) bool) []sensors.Sensor {
if g.FanRPM > 0 {
f.add(p+".fan", g.Name+" fan", "fan", "rpm", float64(g.FanRPM), t, none, lbl...)
}
if g.FanPct != nil {
f.add(p+".fan_pct", g.Name+" fan", "fan", "%", *g.FanPct, t, pct, lbl...)
}
if g.MaxMHz > 0 {
f.add(p+".freq", g.Name+" clock", "frequency", "MHz", g.MHz, t, upTo(g.MaxMHz), lbl...)
}
web/src/Top.tsx
+2 -2
View File
@@ -624,7 +624,7 @@ function GPUBox({ g, values, tick, advance, now, cap }: { g: GPU; values: number
<span className="t-fact" title={g.req_mhz ? `${Math.round(g.req_mhz)} MHz requested${g.max_mhz ? `, ${Math.round(g.max_mhz)} MHz max` : ''}` : undefined}>{Math.round(g.mhz ?? 0)}{g.max_mhz ? ` / ${Math.round(g.max_mhz)}` : ''} MHz</span></div> : null}
{(g.power_w !== undefined && g.power_w > 0) || g.power_cap_w ? <div className="t-row t-sub"><span>Power</span><span className="spacer" />
<span className="t-fact">{(g.power_w ?? 0).toFixed(1)}{g.power_cap_w ? ` / ${Math.round(g.power_cap_w)}` : ''} W</span></div> : null}
{g.fan_rpm ? <div className="t-row t-sub"><span>Fan</span><span className="spacer" /><span className="t-fact">{g.fan_rpm} rpm</span></div> : null}
{g.fan_rpm || g.fan_pct !== undefined ? <div className="t-row t-sub"><span>Fan</span><span className="spacer" /><span className="t-fact">{[g.fan_pct !== undefined ? `${Math.round(g.fan_pct)}%` : '', g.fan_rpm ? `${g.fan_rpm} rpm` : ''].filter(Boolean).join(' · ')}</span></div> : null}
{g.mem_total ? <div className="t-row t-sub"><span>Mem</span><span className="t-pct">{Math.round(((g.mem_used ?? 0) / g.mem_total) * 100)}%</span><Meter pct={((g.mem_used ?? 0) / g.mem_total) * 100} grad="used" /><span className="t-amt">{bytes(g.mem_used ?? 0, 1)}</span></div>
: g.mem_used ? <div className="t-row t-sub"><span>Mem</span><span className="spacer" /><span className="t-fact" title="resident memory of the processes that have it open">{bytes(g.mem_used, 1)}</span></div> : null}
</div>
@@ -645,7 +645,7 @@ function GPUBox({ g, values, tick, advance, now, cap }: { g: GPU; values: number
// Nothing on this GPU can be read — no busy counter, engine, temperature, clock, power, memory,
// nor a process with it open: the Mac's Asahi, today. Its box would be a graph that can never
// draw beside a column of dashes, so it is one line instead. A client opening it brings the box back.
const unmeasured = (g: GPU) => !g.usage_src && !g.engines && !(g.temp && g.temp > 0) && !g.mhz && !g.max_mhz && !g.power_w && !g.power_cap_w && !g.fan_rpm && !g.mem_total && !g.mem_used && g.clients.length === 0
const unmeasured = (g: GPU) => !g.usage_src && !g.engines && !(g.temp && g.temp > 0) && !g.mhz && !g.max_mhz && !g.power_w && !g.power_cap_w && !g.fan_rpm && g.fan_pct === undefined && !g.mem_total && !g.mem_used && g.clients.length === 0
function GPULine({ g }: { g: GPU }) {
return (
<div className="t-row t-gpu-line">
web/src/api.ts
+1 -1
View File
@@ -86,7 +86,7 @@ export interface GPUClient { pid: number; name: string; usage: number; engines?:
export interface GPU {
name: string; driver: string; bus?: string; integrated?: boolean
usage: number; usage_src?: 'idle' | 'busy' | 'clients'; engines?: Record<string, number>
temp?: number; temp_src?: 'gpu' | 'package'; power_w?: number; power_cap_w?: number; fan_rpm?: number
temp?: number; temp_src?: 'gpu' | 'package'; power_w?: number; power_cap_w?: number; fan_rpm?: number; fan_pct?: number
mhz?: number; req_mhz?: number; max_mhz?: number; mem_used?: number; mem_total?: number
clients: GPUClient[]; clients_ok: boolean; stalled?: boolean
}