syphon-wgpu

High-level, zero-copy Syphon integration for wgpu applications on macOS. Publish your rendered frames to other apps, or receive frames as wgpu textures — GPU-to-GPU, no CPU readback.

Built on syphon-core and syphon-metal.

macOS only. Requires Syphon.framework at link/run time. Targets wgpu 29.

Highlights

• SyphonWgpuOutput — publish a Bgra8Unorm wgpu texture each frame. publish() returns a PublishStatus so silent CPU fallbacks are impossible.
• SyphonWgpuInput — receive frames as wgpu textures, push-based via connect_with_channel() (no polling) or by UUID via connect_by_info().
• Native BGRA end-to-end — no format conversion overhead.
• Triple-buffered IOSurface pool to avoid GPU stalls.

Output (server)

use syphon_wgpu::{SyphonWgpuOutput, PublishStatus};

let mut output = SyphonWgpuOutput::new("My App", &device, &queue, 1920, 1080)
    .expect("failed to create Syphon output");

match output.publish(&render_texture, &device, &queue) {
    PublishStatus::ZeroCopy      => { /* GPU-to-GPU blit */ }
    PublishStatus::CpuFallback   => log::warn!("CPU fallback — check Metal setup"),
    PublishStatus::NoClients     => { /* no receivers connected */ }
    PublishStatus::PoolExhausted => log::warn!("increase pool_size"),
}

Input (client, push-based)

use syphon_wgpu::SyphonWgpuInput;

let mut input = SyphonWgpuInput::new(&device, &queue);
let rx = input.connect_with_channel("My App")?;

while rx.recv().is_ok() {
    if let Some(texture) = input.receive_texture(&device, &queue) {
        // texture is Bgra8Unorm, GPU-blitted zero-copy
    }
}
# Ok::<(), syphon_core::SyphonError>(())

Requirements

• macOS 10.13+ with a Metal-capable GPU
• Syphon.framework on the framework search path (typically /Library/Frameworks)
• wgpu 29

Feature flags

• logging — cap log output at warn in release builds.
• max_perf — compile out all logging in release builds.

License

MIT. Part of the syphon-rs workspace.