dispatchThreadgroups(indirectBuffer:indirectBufferOffset:threadsPerThreadgroup:) solves this. does anyone know why threadsPerThreadgroup: executeState.threadExecutionWidth literally freezes my mac if I use it with the indirectBuffer dispatch? (using MTLSize(1,1,1) works fine); while executeState.threadExecutionWidth works well if I dispatch normally.

Trying to understand how to encode this into an indirect command buffer...

same