The flaw, in my opinion, is not necessarily in the benchmark but in the claims and observations around the benchmark. For instance the blog flat out says: > No

The flaw, in my opinion, is not necessarily in the benchmark but in the claims and observations around the benchmark. For instance the blog flat out says: > Nope, we work with float32. It seems that is not true. The blog should make clear that it isn't straightforward or possible to get cupy to do this.