A rumor appears that NVIDIA GeForce RTX 2070 graphics cards will have a total of 2304 CUDA Cores and 8GB of GDDR6 memory
Tomorrow, everything points to us seeing the NVIDIA GeForce RTX 2080 and the NVIDIA GeForce RTX 2080 Ti, but we still need to learn about the mid-range cards aimed at the average user. Within this mid-range will be the GeForce RTX 2070 (it seems likely to support Raytracing with limitations), about which we currently know very little, although Videocardz suggests that this graphics card will have a total of 2304 CUDA Cores, a very interesting figure whose performance implications are still unknown.
The new NVIDIA graphics will have a total of 2304 CUDA Cores
The US media points out that this model would have 2304 CUDA Cores in a non-specific tweet and that it would be accompanied by 8GB of GDDR6 @ 14GHz VRAM memory, which through a 256-bit memory bus will offer us a bandwidth of 448GB / s. It is pointed out that in the end it could be 384GB / s in the case that they use somewhat slower memories, working these at 12GHz.
It should be noted that despite the fact that we know a lot of data, we still have many unknowns to discover and perhaps the most outstanding is at what frequency the GPUs will operate, since at the moment no data has been given on their working frequencies, something that draws powerfully attention. We also do not know the performance which we hope is truly outstanding.
What we also need to know is which benchmark tools will be compatible with Raytracing. So far we have seen simple synthetic benchmarks, the usual ones for any graphics but these are prepared for Raytracing and we need a tool that allows us to quantify how good they are for those who want to play with unprecedented quality and want to enjoy the visual effects through Raytracing. .
