Built on Metal. Not Ported to It.

Most VJ apps that run on a Mac today were not born on a Mac. They have decades of cross-platform code underneath them — OpenGL, DirectX, Vulkan, GLSL — and Metal got bolted on later. That's a perfectly reasonable engineering choice if you have a Windows-paying user base to keep happy. But it has a cost: every frame, the engine is translating from a portable abstraction to whatever the host GPU expects. On Apple Silicon, every cycle spent in translation is a cycle not spent on shader work.

The pattern shows up in vendor statements you can read for yourself. Resolume's own community forum has long acknowledged that Metal on Mac is a compatibility target rather than the engine's home turf. VDMX shipped a Metal-backed pipeline starting with VDMX6 — improving M-series performance significantly — but the engine traces its lineage back to a Cocoa/OpenGL era. TouchDesigner runs on Mac through MoltenVK, the Khronos-blessed Vulkan-to-Metal translation layer; it works, and it works well, but it is by definition a translation. Modul8 v3, still in active sale at the time of writing, requires Rosetta on Apple Silicon and has been explicitly called out by garageCube as incompatible with macOS Tahoe.

RenderWave doesn't carry any of that history. Every kernel is an MTLComputeCommandEncoder dispatch. Every shader compiles ahead-of-time for the specific M-series GPU it'll run on. Every texture in the render graph is zero-copy from a CVPixelBuffer — unified memory is exploited, not worked around. The audio analyzer uses Accelerate / vDSP directly. The signing-and-licensing path is Swift. The UI is SwiftUI.

This isn't a marketing distinction. It's the reason RenderWave can hit sub-3ms frame budgets at 4K60 on an M2 with the audio engine running, MIDI feedback active, and a Syphon output enabled. Cross-platform engines on the same hardware spend more of their per-frame budget on translation overhead, regardless of how well they're optimized otherwise.

Vendor architecture claims summarized from public forum posts, GitHub repos, and official support pages. See the citation block below for primary sources.

RenderWave renders the shader pipeline at native 4K, then upscales to 8K using Apple's MetalFX framework. Two MetalFX modes are exposed: spatial upscaling (lower latency, single-frame) and temporal upscaling (higher quality, uses motion vectors from the shader pipeline). The choice is automatic based on the active output and what the source shader supports.

MetalFX upscaling is a first-party Apple API designed for exactly this kind of workload — render less, present more. On M3 and newer GPUs it makes 8K live output practical from a laptop. On M2 it gives you a comfortable 4K render budget while still feeding an 8K display target.

Primary reference: developer.apple.com/documentation/metalfx.

On ProMotion-capable Macs and external displays, RenderWave syncs to the active refresh rate via CAMetalDisplayLink with adaptive frame pacing. The render loop targets the display's preferred frame rate — 120Hz on ProMotion, 60Hz on standard displays, 144Hz on supported third-party panels — without burning power chasing a refresh the panel can't show.

Variable refresh rate is supported on displays that report it. For projector / club setups with a fixed 60Hz output, RenderWave clamps cleanly and uses the saved GPU headroom for more shader stack depth instead of dropped frames.

Primary reference: developer.apple.com/documentation/quartzcore/cametaldisplaylink.

Lower frame-time variance means the show doesn't hitch on the drop. Cross-platform engines tend to spike when an effect chain crosses an internal translation boundary — a particular blend mode, a feedback loop, a state change. RenderWave's pipeline doesn't have those boundaries because there's no translation step.

Lower power draw means longer battery on a festival run. Sub-10 watt sustained at 4K60 on M3 Max means an unplugged MacBook Pro can do a real set without throttling. Cross-platform engines burn more headroom on translation, so you trip thermal limits sooner.

More shader headroom means more effects in your stack. On an M3 Max, RenderWave's frame budget is large enough to run a 3D raymarched scene + bloom + feedback + RGB split + Syphon out without breaking sweat. That's a layered look you can't get on a laptop with a translation tax.

No Rosetta means no compatibility quirks. Modul8 v3 users have lived with Rosetta on Apple Silicon for years, and have now hit the wall on Tahoe per garageCube's own statements. RenderWave skips all of that — it's the same arm64 binary all the way down, and macOS 26 is a supported target.