Back when I was a customer (before Photon was released, also during) they had a very good tuning, in the order of around 2x faster for the workloads we had at t

Back when I was a customer (before Photon was released, also during) they had a very good tuning, in the order of around 2x faster for the workloads we had at the time (very large graph computation and a “simple” filtering)