NVIDIA / Fuser

A Fusion Code Generator for NVIDIA GPUs (commonly known as "nvFuser")

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

IdModel: Self mapping

naoyam opened this issue · comments

T0[I0]
T1 = broadcast(T0, {false, true})
T2 = broadcast(T0, {true, false})
T3 = mul (T1, T2)

This is not supported as the two domains of T3 are mapped with each other. There will be only a single loop group for both domains, which indicates there's only one for-loop, but that's not what the loop structure should look like.