[kanaffa12345]

there is no way a general purpose compiler will figure this out. op is probably talking about something like halide or tvm or torchscript jit.

[astrange]

https://polly.llvm.org is a research project that could possibly maybe do this.

[kanaffa12345]

a polyhedral compiler wouldn't find this either - polyhedral compilation is for finding optimal schedules for loop nests i.e., the order in which independent (wrt dataflow) iterations run. as far as i know you, a transpose can't be expressed in the polyhedral model.