runWorkgroup method

void runWorkgroup(
  1. (int, int, int) group,
  2. KernelInputs inputs
)

Runs every invocation of the workgroup at group over inputs.

A binding the kernel reads and inputs does not hold is a StateError naming it, as a GPU refuses a dispatch with a binding missing; one it holds and the kernel does not declare is ignored.

Implementation

void runWorkgroup((int, int, int) group, KernelInputs inputs) {
  final (gx, gy, gz) = group;
  _compiled.runWorkgroup(gx, gy, gz, _resources(inputs));
}