Nvidia isn’t going to let the year end off without at least one more product on shelves, but the new Titan RTX is far from a consumer card. Much like the Titan V from last year, the range has shipped from slightly more powerful, far more expensive consumer-facing cards into more attainable compute chips that are far more useful in research and industry than in your latest PC gaming rig. That doesn’t mean it won’t work there, but you’ll find it tough to justify the $2500 cost for gaming alone.
The Titan RTX is unbelievably powerful though, irrespective of its price. It’s a version of the same TU102 GPU found in the GTX 2080 Ti, albeit the fully featured board and not a slightly pared down one. That puts it more in-line with the Quadro RTX 6000 compute card, which costs a massive $6300. Considering this new Titan RTX will feature the same 24GB of GDDR6 memory and the same number of RT, Tensor and CUDA cores, it’s certainly a far more approachable offering for those needing the extra compute power for deep learning, intensive ray-tracing animation and rendering and more.
The Titan RTX is more powerful than the GTX 2080 Ti as you might expect, with Nvidia projecting around 15% more performance in raw gaming performance. But given that it’s over double the price, it’s clear that this isn’t the market Nvidia is aiming for with this release. Instead, interested parties will find use in the card’s 72 Turing RT and 4,608 CUDA cores, both of which will be valuable in FP32, FP16, INT8 and INT4 calculations that can rapidly decrease the training times of neural networks.
The Titan RTX is launching later this month, with Nvidia expecting it on shelves within the USA before the year rounds out. You’ll be able to find it elsewhere afterwards, although you better have a good reason for throwing down this type of money if you aren’t doing cutting-edge research of some kind.
Last Updated: December 4, 2018