NVIDIA GeForce RTX 2070 Ti Specs Allegedly Leak, 2560 CUDA Cores And 14Gbps GDDR6
Remember when NVIDIA teased that "something super is coming" from its GeForce division? The initial speculation was a new graphics card, or perhaps multiple new GPUs. Then a few days later, however, NVIDIA announced a new platform and certification program for creators, called NVIDIA Studio, and it seemed as though that was the "super" surprise. Perhaps not, though.
Rumors of NVIDIA launching additional GeForce RTX cards have not let up. As it relates to that, specs for an alleged GeForce RTX 2070 Ti variant are making the rounds.
RTX2070Ti 2560SP 14Gbps ~1770MHz
— kimi7kop (@kimi7kop) June 3, 2019
confirmed
Forget the grain of salt, you might want a spoonful for this one. The rumor originates from Twitter user kimi7kop, who notes in his profile that his "source is not perfectly reliable." To be fair, that disclaimer pretty much applies to any leak or rumor, but it still casts doubt about the information presented (especially since it's supposedly "confirmed").
That said, a GeForce RTX 2070 Ti addition to the RTX family is at least plausible. Going back to Pascal, there exists a GeForce GTX 1070 Ti, and so it seems only a matter of time before Turing applies the same treatment to the GeForce RTX 2070.
As the rumor goes, the GeForce RTX 2070 will have 2,560 CUDA cores, 14Gbps GDDR6 memory, and a 1,770MHz boost clock. Not surprisingly, those specs would position the card between the GeForce RTX 2070 and GeForce RTX 2080 (we're quite the detectives!).
Here's how the specs compare to NVIDIA's existing RTX lineup (Founders Edition variants, where available)...
- GeForce RTX 2080 Ti: 4,352 CUDA cores, 11GB GDDR6 (14Gbps), 1,635MHz boost clock
- GeForce RTX 2080: 2,944 CUDA cores, 8GB GDDR6 (14Gbps), 1,800MHz boost clock
- GeForce RTX 2070 Ti: 2,560 CUDA cores, ?? GDDR6 memory (14Gbps), 1,770MHz boost clock
- GeForce RTX 2070: 2,304 CUDA cores, 8GB GDDR6 memory (14Gbps), 1,710MHz boost clock
- GeForce RTX 2060: 1,920 CUDA cores, 6GB GDDR6 memory (14Gbps), 1,680MHz boost clock
The GeForce RTX 2060 is the only one of the existing RTX cards that does not have an overclocked Founders Edition variant. For all of the rest, those clockspeeds are representative of the FE models, which are slightly faster than the reference designs (all other specs are the same between FE and reference).
Here is how the leaked specs compare to the reference versions...
- GeForce RTX 2080 Ti: 4,352 CUDA cores, 11GB GDDR6 (14Gbps), 1,545MHz boost clock
- GeForce RTX 2080: 2,944 CUDA cores, 8GB GDDR6 (14Gbps), 1,710MHz boost clock
- GeForce RTX 2070 Ti: 2,560 CUDA cores, ?? GDDR6 memory (14Gbps), 1,770MHz boost clock
- GeForce RTX 2070: 2,304 CUDA cores, 8GB GDDR6 memory (14Gbps), 1,620MHz boost clock
- GeForce RTX 2060: 1,920 CUDA cores, 6GB GDDR6 memory (14Gbps), 1,680MHz boost clock
What about price? Let's have a look at one more set of comparisons...
- GeForce RTX 2080 Ti Founders Edition: $1,199
- GeForce RTX 2080 Ti Reference: $999
- GeForce RTX 2080 Founders Edition: $799
- GeForce RTX 2080 Reference: $699
- GeForce RTX 2070 Ti: ???
- GeForce RTX 2070 Founders Edition: $599
- GeForce RTX 2070 Reference: $499
- GeForce RTX 2060: $349
There is not a lot of wiggle room between the GeForce RTX 2070 and GeForce RTX 2080, especially if the supposed GeForce RTX 2070 Ti variant comes in both FE and reference flavors. Either way, NVIDIA probably has some shuffling to do, for pricing of a new model to make sense. Hopefully that is the case, because if so, it will make the RTX family more affordable, at least for some of the models.
None of this is confirmed (by NVIDIA), so it's a wait-and-see thing. How long will we have to wait? If an announcement looms, we suspect it will be made either at E3 next week, or during Gamescom in August.
