I have an application that relies heavily of FP64 performance and large matrix operations, so we're looking to offload some of this work to the GPU. We don't need ECC or constant computing, so the Tesla series seems unnecessary. I was looking into the Titian black, which seems to beat the most of the Tesla lineup in FP64, and it's a whole lot cheaper. Are there any gotchas or things that I'm overlooking with this card? It seems like it'd suit our purpose well.