`groupshared` is not supported in slangtorch.

[saedrna]

Hi there, groupshared float4x4 mtx; in mesh.slang does not compile (as below). I am using the The latest slangtorch version in pip. I am not familiar with slang lang, but it seems that groupshared is similar to block-wise shared memory in CUDA. I think this should be absolutely supported by CUDA.

renderutils\mesh.slang(51): error 31201: modifier 'groupshared' is not allowed here.       
    groupshared float4x4 mtx;

I am not sure if this is the version difference, because recently slangpy has been renamed to slangtorch.

Han