filename without .wgsl extension (e.g. "sscal"),
or an array of filenames whose source gets concatenated into one module — WGSL has no
#include, so this is how one shader's helper functions (e.g. f64add.wgsl's decode/
encode/computeSum) get reused from another (e.g. dasum.wgsl) without duplicating them.
OptionalentryPoint: string = "main"
which @compute function to run; only needed when
concatenating shaders whose entry points aren't both named "main"
Returns a cached
GPUComputePipelinefor the given shader, compiling it on first use. Pipelines are cached per device so reinitialization (new device) always recompiles.