Very slow resizing

[Gingeh]

Screencast_20251124_234726.webm

It also crashed at the end of the recording after letting go of drag, with the following panic message:

thread 'main' (57919) panicked at /home/[redacted]/.cargo/registry/src/index.crates.io-1949cf8c6b5b557f/wgpu_context-0.1.2/src/surface_renderer.rs:215:14:
failed to get surface texture: Timeout

More details in case any of this is useful:
commit: 76958f4
invocation command: cargo run --profile production --package readme
compiler: rustc 1.93.0-nightly (c23ed3ef2 2025-11-23)
OS: Fedora 43
GPU: 1050 Ti

[nicoburns]

Hmm... I haven't has such a problem, but I suspect this is platform specific. Are you using X or Wayland? And have you seen similar issues with other WGPU apps?

[nicoburns]

May be related to Inlyne-Project/inlyne#419

[Gingeh]

Are you using X or Wayland?

Wayland (forgot to mention, oops)

have you seen similar issues with other WGPU apps?

Just checked with a bevy app and it does occur, albeit with much less slowdown.
I suppose that makes it a general wgpu issue instead.

[nicoburns]

gfx-rs/wgpu#7670 seems relevant

[niminypiminy]

had similar issue before, but if you ditch OpenGL and force Vulkan, this problem goes away unless gpu is old.

#[cfg(target_os = "linux")]
std::env::set_var("WGPU_BACKEND", "vulkan");

On windows, vulkan is also usually the better choice. It uses significantly less memory (almost 40% less) because it avoids driver-side caching.

[Gingeh]

Seems like my GPU is too old for that 😔 (not surprised)

[niminypiminy]

Yeah. a continuous stream of resize events will process slower on older hardware. Most likely cannot destroy/recreate the swapchain images fast enough for this kind of hypermodern stack. stutter becomes more apparent.