Nvidia has just announced the newest addition to the Titan graphics card line, the Titan V, the catch here is that it is very much focused on AI and deep learning. It can still be used for gaming but there are other options than buying this $3000 card.
That puts it at a much higher cost than the Titan Xp card but there is a very good reason for that when looking at the specs sheet. It is also running on a new architecture called the Volta GV100 and it is using Tensor cores with 12GB of HBM2 memory. It still has CUDA cores with a whopping count of 5120 active cores. It does have a slightly lower clock speed compared to the Titan Xp and the 1080 Ti but it makes up for the sheer amount of cores that they included.
To put this card into perspective, it can output 14.9 TFLOPS of single-precision FP32 computations, a 31 percent increase compared to the Titan Xp. The HBM2 memory module has a clock speed of 1.7Gbps and 3072-bit memory which performs at 653GB/s of bandwidth. The department that it shines in is AI calculations as it has 640 Tensor cores to perform the calculations and a total of 128 FLOPS of power to perform multiply and addition calculations.
You will not see a huge change in gaming if you decide to slap this card into your system, but there is untapped potential there as different types of coding is required to make use of the new architecture. Never the less, the time will come where games will be able to use AI to process out game environments and that’s where the Titan V will shine.
