monster/modelbeast
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

docs: CUDA worker — add Win11 + WSL2 path (recommended, no reinstall)

Browse Source 
T490s already runs Win11, so keep it: WSL2 gives clean Linux CUDA env with eGPU
passthrough where the Linux-first 3D repos build properly. Bare-metal Ubuntu kept
as the leaner Path B. Covers mirrored WSL networking, headless/lid + Windows Update
settings for an always-on worker.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
MODELBEAST 2026-07-13 11:01:36 +10:00
parent 27ce69782d
commit 8fc7b407f3
1 changed files with 51 additions and 43 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

94
docs/CUDA-WORKER.md
View File

@ -1,63 +1,71 @@
# CUDA Worker Setup — T490s + 2080 Ti eGPU on the tailnet # CUDA Worker Setup — T490s + 2080 Ti eGPU on the tailnet
Goal: a headless Linux box with the RTX 2080 Ti (11GB) eGPU, live on Tailscale, that Goal: a headless box with the RTX 2080 Ti (11GB) eGPU, live on Tailscale, that
MODELBEAST (on the M3 Ultra) offloads the CUDA-only jobs the Mac can't run — MODELBEAST (on the M3 Ultra) offloads the CUDA-only jobs the Mac can't run —
gaussian-splat → mesh extraction (SuGaR/2DGS), research-grade video mocap gaussian-splat → mesh extraction (SuGaR/2DGS), research-grade video mocap
(GVHMR/WHAM/TRAM), NeRF/tiny-cuda-nn, nvdiffrast/PyTorch3D. (GVHMR/WHAM/TRAM), NeRF/tiny-cuda-nn, nvdiffrast/PyTorch3D.
**Why a ThinkPad and not the Intel iMac:** no T2 chip = no Secure-Boot/Thunderbolt **Why the T490s and not the Intel iMac:** no T2 chip = no Secure-Boot/Thunderbolt
fight; laptop idles ~6–10W (+ ~20W eGPU) vs ~55–75W for the iMac; battery = built-in fight; laptop idles ~6–10W (+ ~20W eGPU) vs ~55–75W for the iMac; battery = built-in
UPS. The 2080 Ti is EOL-ish (Turing: no bf16/fp8, 11GB) — it's a *compatibility* UPS. The 2080 Ti is EOL-ish (Turing: no bf16/fp8, 11GB) — a *compatibility* box, not
box, not a performance one. It will not beat the M3 Ultra at anything the Mac a performance one. It won't beat the M3 Ultra at anything the Mac already does; it
already does; it only unlocks the CUDA-locked tail. only unlocks the CUDA-locked tail.
## Do this at the shop (once) **A physical GPU can't be passed to a VM on a Mac/Windows consumer hypervisor** — so
the CUDA OS must have direct hardware access. On this ThinkPad that's either native
Windows (already installed) or WSL2 (a real Linux env Windows gives GPU access to).
NOT a Linux VM on macOS.
1. **Ubuntu 24.04 LTS**, bare metal (dual-boot Win11 or wipe — bare-metal, NOT a VM; ---
macOS/Windows hypervisors can't pass a physical GPU through, so a Linux VM can't
do CUDA on the eGPU).
2. **Thunderbolt authorize the eGPU** (plug in the Razer Core first): ## Path A — Keep Windows 11 + WSL2 (RECOMMENDED: Win11 already runs, no reinstall)
```bash
sudo apt install bolt Windows drives the eGPU; WSL2 gives a clean Ubuntu where the Linux-first 3D repos
boltctl list # find the enclosure UUID build properly (native-Windows builds of their custom CUDA kernels are a pain).
boltctl enroll --policy auto <UUID>
1. **NVIDIA Windows driver** (Studio or Game Ready) with the 2080 Ti plugged in.
Confirm Device Manager → Display adapters shows the 2080 Ti. This one driver is
all WSL2 needs — do NOT install a driver inside WSL.
2. **WSL2 + Ubuntu** (admin PowerShell): `wsl --install -d Ubuntu` → reboot.
3. **CUDA toolkit INSIDE WSL** (the `wsl-ubuntu` package — toolkit only, no driver),
per NVIDIA's "CUDA on WSL" guide. Then verify: `nvidia-smi` inside Ubuntu lists
the 2080 Ti. (If empty: update the Windows driver; ensure eGPU present at boot.)
4. **Reach WSL from the tailnet cleanly** — create `C:\Users\<you>\.wslconfig`:
``` ```
[wsl2]
3. **NVIDIA driver + CUDA:** networkingMode=mirrored
```bash
sudo ubuntu-drivers autoinstall # or the latest official driver branch
sudo apt install nvidia-cuda-toolkit
sudo reboot
nvidia-smi # must list the 2080 Ti over Thunderbolt
``` ```
If it doesn't appear: confirm the eGPU was connected at boot, re-check `boltctl`, (Win11 22H2+; makes WSL share the host's network so Tailscale/SSH reach it.)
ensure the `thunderbolt` module loads before `nvidia_drm`. 5. **Tailscale** on the Windows host: install, sign in, `tailscale ip -4` → note it.
6. **Headless Windows settings:** Power → "when I close the lid: do nothing" + never
sleep; enable **OpenSSH Server** (Settings → Optional Features); set Windows Update
active hours wide / pause auto-reboots so it never reboots mid-job.
7. Access model: MODELBEAST SSHes to the Windows host over Tailscale and runs CUDA
tools via `wsl bash -lc "..."`, or SSH straight into WSL with mirrored networking.
4. **Tailscale:** ## Path B — Bare-metal Ubuntu (leaner/lower-power, if you'd rather wipe Win11)
```bash
curl -fsSL https://tailscale.com/install.sh | sh
sudo tailscale up --ssh # --ssh lets the M3 Ultra SSH in over the tailnet
tailscale ip -4 # note this IP → put it in MODELBEAST settings
```
5. **Headless / low-power:** run lid-closed → 1. Ubuntu 24.04 LTS bare metal.
`sudo sed -i 's/#HandleLidSwitch=suspend/HandleLidSwitch=ignore/' /etc/systemd/logind.conf` 2. `sudo ubuntu-drivers autoinstall && sudo apt install nvidia-cuda-toolkit && sudo reboot`; `nvidia-smi` must show the 2080 Ti.
then `sudo systemctl restart systemd-logind`. Enable Wake-on-LAN in BIOS + `ethtool` 3. `curl -fsSL https://tailscale.com/install.sh | sh && sudo tailscale up --ssh`.
if you want it to sleep and wake on demand. 4. Lid-closed headless: set `HandleLidSwitch=ignore` in `/etc/systemd/logind.conf`.
(No Thunderbolt-authorization dance needed on a non-Mac — it enumerates on boot.)
## MODELBEAST integration (I'll build this when the box exists) ---
## MODELBEAST integration (I'll build this once the box is on the tailnet)
The operator framework already shells out to tools; a **remote-CUDA operator** wraps The operator framework already shells out to tools; a **remote-CUDA operator** wraps
that pattern: rsync the input to `cuda_host:~/work/`, `ssh cuda_host` to run the CUDA that: rsync the input to the worker, `ssh` (→ `wsl` on Path A) to run the CUDA tool
tool in its venv, rsync results back, register as assets. Planned operators: in its env, rsync results back, register as assets. Add a `cuda_host` setting (tailnet
- `sugar_mesh` — splat `.ply` → textured mesh (closes the Brush→mesh gap; the Mac can't extract meshes from splats). IP + user) to the vault. Planned:
- `gvhmr_mocap` — video → world-grounded SMPL → FBX (research-grade, replaces the cloud/manual path). - `sugar_mesh` — splat `.ply` → textured mesh (closes the Brush→mesh gap; the Mac can't extract meshes from splats). ~5–6GB, fits 11GB.
- `gvhmr_mocap` — video → world-grounded SMPL → FBX (research-grade; replaces the cloud/manual path). Light.
Add a `cuda_host` setting (the tailnet IP + user) to the vault so these operators know where to reach it. Fits in 11GB: SuGaR/2DGS/gsplat (~5-6GB), GVHMR/WHAM (light). Does NOT fit: FLUX, 20B+ models — those stay on the Mac. Does NOT fit 11GB: FLUX, 20B+ models — those stay on the Mac.
## Reality check ## Reality check
Only worth it because the card's already owned. If you were buying, a used RTX 3090 Worth it only because the card's already owned. If buying, a used RTX 3090 (24GB,
(24GB, bf16) would be strictly better and unlock FLUX/video too. Turing's runway is bf16) is strictly better and unlocks FLUX/video too. Turing's runway is shrinking
shrinking (newer 3D wheels drop 20-series), so treat this as "use the free card for (newer 3D wheels drop 20-series). Use the free card for the CUDA-only leftovers, not
the CUDA-only leftovers," not a long-term foundation. as a foundation.