shader-f16
Adds 16-bit floating-point types in shaders. Using 16-bit values can reduce memory bandwidth and register pressure compared with full 32-bit floats.
Check your device: WebGPU Capabilities Report.
More details: Developer Docs (MDN) Specification (W3C)
Overall Support
Percentage supporting this feature among reports tested for it.
93.98%
of browsers tested for this feature
Platform Support
Browser Support
| Browser | Support |
|---|---|
| Chrome/Chromium | 93.26% |
| Edge | 93.16% |
| Firefox | 92.84% |
| Opera | 91.32% |
| Other | 94.84% |
| Safari | 99.97% |
| Samsung Internet | 93.73% |
Vendor Support
| Vendor | Support |
|---|---|
| amd | 92.22% |
| apple | 99.99% |
| arm | 100% |
| 0% | |
| img-tec | 100% |
| intel | 95.41% |
| intelr | 100% |
| microsoft | 30% |
| nvidia | 84.67% |
| qualcomm | 71.13% |
| samsung | 100% |
| venus | 100% |
Vendor Architecture Support
| Vendor - Architecture | Support |
|---|---|
| amd - gcn-4 | 29.08% |
| amd - gcn-5 | 99.81% |
| amd - rdna-1 | 99.66% |
| amd - rdna-2 | 99.96% |
| amd - rdna-3 | 100% |
| apple - apple | 100% |
| apple - metal-3 | 100% |
| arm - bifrost | 100% |
| arm - valhall | 100% |
| intel - gen-11 | 100% |
| intel - gen-12lp | 99.91% |
| intel - gen-8 | 28.21% |
| intel - gen-9 | 87.24% |
| intel - xe-2lpg | 100% |
| intel - xe-lpg | 99.85% |
| nvidia - ampere | 98.4% |
| nvidia - blackwell | 98.76% |
| nvidia - kepler | 6.85% |
| nvidia - lovelace | 98.73% |
| nvidia - maxwell | 0% |
| nvidia - pascal | 0.16% |
| nvidia - turing | 98.38% |
| qualcomm - adreno-6xx | 31.27% |
| qualcomm - adreno-7xx | 96.77% |
| qualcomm - adreno-8xx | 97.98% |