feat(host,web): multi-GPU selection — GPU inventory + preference API, web-console GPU card
apple / swift (push) Successful in 1m9s
ci / rust (push) Successful in 1m50s
ci / web (push) Successful in 56s
ci / docs-site (push) Successful in 57s
decky / build-publish (push) Successful in 11s
android / android (push) Successful in 3m13s
apple / screenshots (push) Successful in 5m32s
deb / build-publish (push) Successful in 3m15s
docker / build-push (--build-arg FEDORA_VERSION=44, ci, ci/fedora-rpm.Dockerfile, punktfunk-fedora44-rpm) (push) Successful in 5s
windows-host / package (push) Successful in 7m35s
docker / build-push (., web/Dockerfile, punktfunk-web) (push) Successful in 31s
docker / build-push (ci, ci/fedora-rpm.Dockerfile, punktfunk-fedora-rpm) (push) Successful in 4s
docker / build-push (ci, ci/rust-ci.Dockerfile, punktfunk-rust-ci) (push) Successful in 4s
docker / build-push (docs-site, docs-site/Dockerfile, punktfunk-docs) (push) Successful in 4s
ci / bench (push) Successful in 4m53s
rpm / build-publish (bazzite, punktfunk-fedora-rpm) (push) Successful in 9m58s
rpm / build-publish (fedora-44, punktfunk-fedora44-rpm) (push) Successful in 9m54s
docker / deploy-docs (push) Successful in 18s

- new crate::gpu (compiled on all platforms so the OpenAPI doc stays
  platform-independent): DXGI / sysfs GPU inventory with reboot-stable ids
  (PCI vendor:device + occurrence — LUIDs are per-boot), persisted auto/manual
  preference (<config>/gpu-settings.json, atomic temp+rename with in-memory
  rollback), one selection with precedence console preference >
  PUNKTFUNK_RENDER_ADAPTER > max VRAM and graceful fallback when the preferred
  GPU is absent, plus a live "in use" record (RAII session guard wrapped around
  every encoder open_video returns)
- fix: windows_gpu_vendor derived the encoder backend from DXGI adapter 0
  instead of the selected render adapter — on a hybrid box (e.g. Intel iGPU at
  index 0 + NVIDIA dGPU) the backend could disagree with the GPU the capture
  ring / IddCx render pin sit on. The NVENC 4:4:4 probe now also runs on the
  selected adapter (was: OS default), the codec/4:4:4 probe caches are keyed
  per selected GPU (were process-lifetime OnceLocks), and an explicit
  PUNKTFUNK_ENCODER conflicting with the selected GPU's vendor warns up front
- mgmt API: GET /api/v1/gpus (inventory + mode + preferred + next-session
  selection with reason + in-use GPU/backend/session-count) and
  PUT /api/v1/gpus/preference (validates mode/gpu_id before writing);
  openapi.json regenerated; the vdisplay render pin now also engages for a
  console preference (not just the env pin)
- web console: GPU card on the Host page — list with vendor + VRAM,
  Automatic / Prefer controls, Preferred / Next session / "In use · backend"
  badges, missing-preferred-GPU warning and env-pin note; en + de messages
- Linux: a matched manual preference picks the VAAPI render node and the
  NVENC-vs-VAAPI auto choice; auto mode is exactly the previous behavior

Validated live on the hybrid laptop (RTX 3500 Ada + Intel Arc Pro, which
enumerates twice — the occurrence ids disambiguate): enumerate, prefer,
bad-id 400, restart persistence, auto-restore keeping the stored pick.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
2026-07-02 13:57:18 +02:00
parent 40fefd73ca
commit 019f2677a7
17 changed files with 1881 additions and 200 deletions
@@ -50,10 +50,13 @@ const fn fourcc(a: u8, b: u8, c: u8, d: u8) -> u32 {
(a as u32) | ((b as u32) << 8) | ((c as u32) << 16) | ((d as u32) << 24)
}
/// The render node a VAAPI/DRM device should open. `PUNKTFUNK_RENDER_NODE` pins it on a multi-GPU
/// box; the default is correct on a single-GPU host.
/// The render node a VAAPI/DRM device should open, from [`crate::gpu::linux_render_node`]: a
/// matched web-console GPU preference pins it, else `PUNKTFUNK_RENDER_NODE`, else the single-GPU
/// default.
fn render_node() -> CString {
let p = std::env::var("PUNKTFUNK_RENDER_NODE").unwrap_or_else(|_| "/dev/dri/renderD128".into());
let p = crate::gpu::linux_render_node()
.to_string_lossy()
.into_owned();
CString::new(p).unwrap_or_else(|_| CString::new("/dev/dri/renderD128").unwrap())
}
@@ -254,6 +254,11 @@ pub fn probe_can_encode_444(_vendor: WinVendor, _codec: Codec) -> bool {
}
pub fn probe_can_encode(vendor: WinVendor, codec: Codec) -> bool {
// Deliberately NOT pinned to the selected render adapter (unlike `nvenc::probe_can_encode_444`):
// the system-input probe passes no hwdevice, and the AMF/QSV runtimes only ever bind their own
// vendor's silicon — on a mixed-vendor box the probe lands on the right GPU by construction.
// Only a two-same-vendor-GPU box could probe the wrong card (accepted; results are cached per
// selected GPU in `windows_codec_support`, so a fix here slots in without churn).
if ffmpeg::init().is_err() {
return false;
}
@@ -958,34 +958,59 @@ impl Drop for NvencD3d11Encoder {
/// so the host advertises the chroma it can really encode (honest downgrade to 4:2:0 on a card without it).
pub fn probe_can_encode_444(codec: Codec) -> bool {
use windows::Win32::Foundation::HMODULE;
use windows::Win32::Graphics::Direct3D::{D3D_DRIVER_TYPE_HARDWARE, D3D_FEATURE_LEVEL_11_0};
use windows::Win32::Graphics::Direct3D::{
D3D_DRIVER_TYPE_HARDWARE, D3D_DRIVER_TYPE_UNKNOWN, D3D_FEATURE_LEVEL_11_0,
};
use windows::Win32::Graphics::Direct3D11::{
D3D11CreateDevice, D3D11_CREATE_DEVICE_BGRA_SUPPORT, D3D11_SDK_VERSION,
};
use windows::Win32::Graphics::Dxgi::{CreateDXGIFactory1, IDXGIAdapter1, IDXGIFactory4};
if codec != Codec::H265 {
return false;
}
// SAFETY: a self-contained probe owning every handle it creates. `D3D11CreateDevice` (HARDWARE
// driver, NULL adapter) fills `device` or returns Err (→ false). `open_encode_session_ex` opens an
// NVENC session against that device's raw pointer (valid while `device` is held) or errors (→ false,
// tearing nothing down). `get_encode_caps` reads one scalar cap into `val` via the loaded API table.
// `destroy_encoder` frees the session exactly once; `device`/its context drop with the COM wrappers.
// No handle escapes this call and nothing runs concurrently.
// SAFETY: a self-contained probe owning every handle it creates. `CreateDXGIFactory1`/
// `EnumAdapterByLuid` return owned COM objects or err (→ default-adapter fallback).
// `D3D11CreateDevice` (explicit adapter + UNKNOWN driver type, or NULL adapter + HARDWARE)
// fills `device` or returns Err (→ false). `open_encode_session_ex` opens an NVENC session
// against that device's raw pointer (valid while `device` is held) or errors (→ false, tearing
// nothing down). `get_encode_caps` reads one scalar cap into `val` via the loaded API table.
// `destroy_encoder` frees the session exactly once; `device`/its context drop with the COM
// wrappers. No handle escapes this call and nothing runs concurrently.
unsafe {
// Probe on the SELECTED render adapter — the GPU the session will actually encode on
// (web-console preference / PUNKTFUNK_RENDER_ADAPTER / max VRAM). The OS default adapter
// (NULL) can be the *other* GPU on a hybrid box, answering for hardware we won't use.
let adapter: Option<IDXGIAdapter1> = crate::win_adapter::resolve_render_adapter_luid()
.and_then(|luid| {
let factory: IDXGIFactory4 = CreateDXGIFactory1().ok()?;
factory.EnumAdapterByLuid(luid).ok()
});
let mut device: Option<ID3D11Device> = None;
if D3D11CreateDevice(
None,
D3D_DRIVER_TYPE_HARDWARE,
HMODULE::default(),
D3D11_CREATE_DEVICE_BGRA_SUPPORT,
Some(&[D3D_FEATURE_LEVEL_11_0]),
D3D11_SDK_VERSION,
Some(&mut device),
None,
None,
)
.is_err()
{
let created = match &adapter {
Some(a) => D3D11CreateDevice(
a,
D3D_DRIVER_TYPE_UNKNOWN,
HMODULE::default(),
D3D11_CREATE_DEVICE_BGRA_SUPPORT,
Some(&[D3D_FEATURE_LEVEL_11_0]),
D3D11_SDK_VERSION,
Some(&mut device),
None,
None,
),
None => D3D11CreateDevice(
None,
D3D_DRIVER_TYPE_HARDWARE,
HMODULE::default(),
D3D11_CREATE_DEVICE_BGRA_SUPPORT,
Some(&[D3D_FEATURE_LEVEL_11_0]),
D3D11_SDK_VERSION,
Some(&mut device),
None,
None,
),
};
if created.is_err() {
return false;
}
let Some(device) = device else { return false };