immediate_address_space
WGSL language feature
Specification: WGSL § immediate_address_space
Check navigator.gpu?.wgslLanguageFeatures?.has('immediate_address_space')
before choosing a shader that uses this language feature. It is available automatically
when supported; requires immediate_address_space; documents the dependency
and rejects the shader on unsupported implementations. This language token is not a
requiredFeatures entry for requestDevice(); any additional
device requirements are described below.
A module-scope var<immediate> declaration exposes a small block of read-only shader data. The application supplies its bytes with setImmediates() while encoding rendering or compute commands, instead of creating a uniform-buffer binding for that data. Each entry point may statically access at most one immediate variable.
Per-draw indices, small flags, and per-dispatch parameters are typical uses. Check the language token together with the adapter’s maxImmediateSize, request the size your device needs, and configure the pipeline layout’s immediateSize. Data layout and command offsets must match the shader declaration and WebGPU alignment rules.
Language support alone does not guarantee that a particular device or pipeline can accept the amount of data you need. Keep the block small and treat it as command state, not persistent writable shader memory. A uniform buffer, optionally using dynamic offsets to select per-draw records, provides a fallback when immediate data is unavailable or too small.
Availability by month
| Month | With feature | Eligible reports | Availability |
|---|---|---|---|
| 2026-08 | 63 | 106 | 59.4% |
| 2026-09 | 22 | 36 | 61.1% |
By browser (last 30 days)
| Browser | Availability | n |
|---|---|---|
| chrome | 85.6% | 90 |
| edge | 100.0% | 8 |
| firefox | 0.0% | 44 |
Availability = share of first-party, non-quarantined reports with WebGPU on a hardware
(non-fallback) adapter
whose browser reports this token in navigator.gpu.wgslLanguageFeatures.
This measures language support in the browser, not a GPU hardware capability.
Developer-leaning sample.
Full methodology.