[tracker1]

Not really, unless some future hardware supports such a beast, could be an interesting approach for AI/WebGPU interfaces.

It's more for service-oriented applications currently... you pass stuff in, get stuff out.