Allow controlling the graphics backend & power preference through standard wgpu env vars #1332
Conversation
Yes, this means we're pretty much down to the default. At least this gives a nice place for all these comments.
Wumpf
added
enhancement
| // * Some Windows VMs only provide DX12 drivers, observed with Parallels on Apple Silicon | ||
| // * May run into Linux issues that warrant trying out the GL backend. | ||
| // | ||
| // For changing the backend we use standard wgpu env var, i.e. WGPU_BACKEND. |
There was a problem hiding this comment.
The wgpu docs are quite lacking here in what WGPU_BACKEND expects, but reading the code it seems like it would be:
WGPU_BACKEND=gl,metal ?
Please add a comment here so that the next reader can just use the env-var instead of having to dig into the wgpu code!
There was a problem hiding this comment.
oh. yes I'll sneak that in some future PR. Also I'll create a PR for our troubleshooting doc page.
For everyone stumbling in here. This makes us support:
WGPU_BACKEND with a comma separated list of the backends you want to use: vulkan, metal, dx12, dx11, or gl).
WGPU_POWER_PREF with the power preference to choose when a specific adapter name isn't specified - high or low
There was a problem hiding this comment.
This PR introduces two new environment variables that control the behavior of Rerun:
WGPU_BACKEND and WGPU_POWER_PREF.
We should document them!
It would make sense to me if rerun --help would list them, and that's indeed where RERUN_TRACK_ALLOCATIONS is already documented (in crates/rerun/src/run.rs). Please add the above environment variables to that list, and maybe throw in RERUN while your at it, since we seems to have forgotten that one :)
Didn't realize that these env-vars are opt-ins from
wgpu::util(kinda makes sense thinking about it!)This might help us with troubleshooting in tickets like #1324
Depicted is my Ubuntu vm running with GL - note how this completely meses up gamma handling -.-
Checklist
CHANGELOG.md(if this is a big enough change to warrant it)