All right, so here are the latest news I could gather around the web. Frankly talking we are less than 48 hours from NVIDIA’s big GeForce 20 series launch event at Gamescom, so leaks at this stage at this stage are usually accurate.
NVIDIA GeForce RTX 2070 Specs Leaked – 2304 CUDA Cores & 8GB GDDR6
As opposed to the GeForce RTX 2080 which has been confirmed to feature 23 Turing SMs with a total of 2944 CUDA cores, the RTX 2070 will come in at only 18 Turing SMs for a total of 2304 CUDA cores. One very interesting tidbit that has also come out is that the RTX 2070 will in fact feature 8GB of GDDR6 memory, rather than 7GB as was previously believed. This was claimed by two separate sources according to videocardz.
Based on the above specifications, we can extrapolate that the card will likely feature a 256-bit GDDR6 memory interface for a total of 448GB/s of bandwidth, assuming 14gbps GDDR6 memory is employed or 384GB/s of bandwidth if the company opts for slightly slower 12gbps GDDR6. Alternatively, NVIDIA may again be employing a form of memory segmenting like it did with the GTX 970. Although with the immense infamy that solution has garnered it’s not likely we will see it again.
GeForce 20 series lineup, according to AdoredTV
NVIDIA GeForce RTX 2080 – $500-$700 US (50% Faster Than 1080)
NVIDIA GeForce RTX 2070 – $300-$500 US (40% Faster Than 1070)
NVIDIA GeForce GTX 2060 – $200-$300 US (27% Faster Than 1060)
NVIDIA GeForce GTX 2050 – $100-$200 US (50% Faster Than 1050 Ti)
NVIDIA GeForce RTX 2070 – Performance, Price & Launch Date
Unlike the RTX 2080 and 2080Ti launching next Monday, the RTX 2070 will not hit the shelves before September. Nothing surprising since Nvidia will try its best to bank on its newest architecture as it usually does.
Estimated price is about 400 USD at launch and should outperform the current GTX 1080 by about 8% to 10%, give or take.
According to wccftech.com , 18 Turing SMs running at the same clock speeds of NVIDIA’s RTX Quadro series, which is about ~1.7GHz would generate roughly 8 TFLOPS of single precision compute performance. That’s roughly right around what the GTX 1080 is capable of.
Although it’s worthy of note that improvements to the inner-workings of the graphics architecture allow it to make more use of its TFLOPS than Pascal, which would put the RTX 2070 ahead of the GTX 1080 as previous leaks have suggested. Not to mention the built-in acceleration for next-generation lighting engines using ray-tracing technology.
Turing is said to reach 2GHz clock speeds with a bit of overclocking, with that factored in the RTX 2070 could very well end up being a bargain GTX 1080 Ti once overclocked. In combination with an attainable price tag of $400 or right around it, the RTX 2070 will very likely be the go-to graphics card in the Turing line-up.
NVIDIA GeForce 20 Series Preliminary Specifcations
| Graphics Card Name | NVIDIA GeForce GTX 2050 | NVIDIA GeForce GTX 2060 | NVIDIA GeForce RTX 2070 | NVIDIA GeForce RTX 2080 | NVIDIA GeForce RTX 2080 Ti |
|---|---|---|---|---|---|
| GPU Architecture | Turing GPU | Turing GPU | Turing GPU | Turing GPU | Turing GPU |
| Process | 12nm NFF | 12nm NFF | 12nm NFF | 12nm NFF | 12nm NFF |
| CUDA Cores | TBD | TBD | 2304 Cores | 2944 Cores | 4352 Cores |
| TMUs/ROPs | TBD | TBD | TBD | TBD | TBD |
| Cache | TBD | TBD | TBD | TBD | 6 MB L2 Cache |
| Base Clock | TBD | TBD | TBD | TBD | TBD |
| Boost Clock | TBD | TBD | TBD | TBD | TBD |
| Compute | TBD | TBD | TBD | TBD | TBD |
| Memory | 4 GB GDDR5 | 5 GB GDDR6 | 8 GB GDDR6 | 8 GB GDDR6 | 11 GB GDDR6 |
| Memory Speed | 8.00 Gbps | 12 Gbps | 12-14 Gbps | 14 Gbps | 14 Gbps |
| Memory Interface | 128-bit | 192-bit | 256-bit | 256-bit | 352-bit |
| Memory Bandwidth | 128 GB/s | 288 GB/s | 384-448 GB/s | 448 GB/s | 616 GB/s |
| TDP | TBD | TBD | ~180W | ~210W | 250W |
| Price | ~$150 US | ~$300 US | ~$400 US | $699 US | $899 US |
| Launch | November 2018? | October 2018? | September 2018 | August 2018 | August 2018 |