The reporter I talked to really seemed to know this area, so it was fun to chat with him. However, it's interesting how a 30 minute chat turns into a single quote! The part that I talked about that didn't make it in was about how these two machines take very different approaches, with different pros and cons. The Shaw machine ("Anton") is very impressive on many fronts. However, it's also pretty impressive in the end just how much a single high end GPU can do science-wise compared to their special purpose supercomputer. I expect that FAH's 8,000 GPU's could give that machine a good run for its money
However, in the end, I've always been a firm believer in cooperation rather than competition, so hopefully we can find some places for collaboration, much like we have done with the IBM Blue Gene team.
