[zanussbaum]

i tried using workgroup shared memory and found it slower than just recomputing everything in each thread although i may have been doing something dumb

i'm excited to try subgroups though: https://developer.chrome.com/blog/new-in-webgpu-128#experime...