Previous | Next --- Slide 7 of 79
Back to Lecture Thumbnails

@RX Spectacular!!!


I have a general question about how the industry reacts when technology like this hits the market. The Titan X (the GPU used in the rendering of this video) is clearly massively powerful; clocking in (hehe) at 3072 CUDA cores and 1000 MHz base clock (which I guess is a bit slower than the 980x). Even with a slightly slower clock rate than the 980, with 3072 cuda cores the Titan X is insane, and clearly has the potential to speedup any sufficiently parallel program.

That being said, when it hit the market did all of the HFT firms immediately buy as many as humanly possible and upgrade their infrastructure and code to support it? I'd imagine that if some of them didn't, then the ones who chose to upgrade would be laughing all the way to the bank. Am I wrong about that? How quickly does the HFT (or any industry that requires super low latency) generally start adopting new hardware and software like this?