During the early summer, rumors hailed about upcoming launches from both AMD and Nvidia, where the latter was the Geforce RTX Super product series. According to the information, the RTX Super series consists of a trio of Turing-based graphics cards in the form of the launched Geforce RTX 2070 Super and RTX 2060 Super, as well as the upcoming RTX 2080 Super.
Geforce RTX 2070 Super and RTX 2060 Super were released on July 2 and compete mainly against AMD’s brand new Radeon RX 5700 and RX 5700 XT, which were launched on July 7. More powerful Geforce RTX 2080 Super will instead appear on July 23 and offer increased performance through more unlocked CUDA cores and higher frequencies for both memory and graphics circuit.
Now the model is glimpsed in the performance test for Final Fantasy XV, in the resolution 2,560 × 1,440 pixels with the detail settings on “High”. This mode means that Nvidia-specific features such as Hairworks is activated and the performance is therefore not directly comparable with AMD cards. It is clear, however, that the Geforce RTX 2080 Super steps up to 8,736 points, which is just over 8 percent higher than the RTX 2080.
The increase in performance also means that the upcoming card performs better than Pascal-based Titan Xp and on a par with the Titan V model, which has the Volta architecture under the hood. However, the step up to the top models with the current Turing architecture is still a long way off, where the Geforce RTX 2080 Ti and Titan RTX perform 9,928 and 11,504 points respectively.
Specifications: Nvidia Geforce RTX 2080 Super
RTX 2080 Ti | RTX 2080 Super | RTX 2080 | RTX 2070 Super | |
---|---|---|---|---|
Technical | 12 nm TSMC | 12 nm TSMC | 12 nm TSMC | 12 nm TSMC |
Circuit | TU102 | TU104 | TU104 | TU104 |
Circuit surface | 754 mm² | 545 mm² | 545 mm² | 545 mm² |
Transistors | 18.6 billion | 13.6 billion | 13.6 billion | 13.6 billion |
Architecture | Turing | Turing | Turing | Turing |
CUDA cores | 4 352 st. | 3 072 st. | 2 944 st. | 2 560 st. |
Texture units | 272 st. | 192 st. | 184 st. | 160 st. |
Raster units | 96 st. | 64 st. | 64 st. | 64 st. |
Tensor cores | 544 st. | 384 st. | 368 st. | 320 st. |
RT cores | 68 st. | 48 st. | 46 st. | 40 st. |
Clock frequency | 1 350 MHz | 1 650 MHz | 1 515 MHz | 1 605 MHz |
GPU Boost | 1 545 MHz | 1 815 MHz | 1 710 MHz | 1 770 MHz |
Computational power | 13 448 GFLOPS | 11 151 GFLOPS | 10 068 GFLOPS | 9 062 GFLOPS |
Memory bus | 352-bit | 256-bit | 256-bit | 256-bit |
Memory amount | 11 GB GDDR6 | 8 GB GDDR6 | 8 GB GDDR6 | 8 GB GDDR6 |
Memory frequency | 14 000 MHz | 15 500 MHz | 14 000 MHz | 14 000 MHz |
Memory bandwidth | 616 GB/s | 496 GB/s | 448 GB/s | 448 GB/s |
Power supply | 8+8-pin | 8+6-pin | 8+6-pin | 8+6-pin |
SLI connection | NVLink 2.0 x16 | NVLink 2.0 x8 | NVLink 2.0 x8 | NVLink 2.0 x8 |
Outputs | 1 st. HDMI 2.0b | 1 st. HDMI 2.0b | 1 st. HDMI 2.0b | 1 st. HDMI 2.0b |
TDP | 250 W | 250 W | 250 W | 215 W |
Rec. Award | 999 USD | 699 USD | 699 USD | 499 USD |
* Values apply to Nvidia’s Founders Edition of each model, which comes with 90 MHz overclocking. In addition to slightly higher performance, this in all cases leads to an increase in TDP of 10 W.
Like last year’s Geforce RTX 2080, the Super version uses the TU104 graphics circuit, but the number of active CUDA cores increases from 2,944 to 3,072, while 2 RT and 16 Tensor cores are added. In addition, the memory frequency climbs from 14,000 to 15,500 MHz, while the turbo and base frequencies land at 1,650 and 1,815 MHz, respectively. This points to a decent increase in performance, but whether 8 percent is true remains to be seen.