During the autumn, Nvidia consolidated its graphics card management with Geforce RTX, which in addition to raising the performance bar introduced the possibility of ray tracing in real time. Now comes the next part of the saga in the form of a new flagship in the segment “expensive” – Nvidia Titan RTX.
Specifications: Titanium RTX and Geforce RTX 2000
Titan RTX | RTX 2080 Ti | RTX 2080 | RTX 2070 | |
---|---|---|---|---|
Technical | 12 nm TSMC | 12 nm TSMC | 12 nm TSMC | 12 nm TSMC |
Circuit | TU102 | TU102 | TU104 | TU106 |
Circuit surface | 754 mm2 | 754 mm2 | 545 mm2 | 445 mm2 |
Transistors | 18.6 billion | 18.6 billion | 13.6 billion | 10.6 billion |
Architecture | Turing | Turing | Turing | Turing |
CUDA cores | 4 608 st. | 4 352 st. | 2 944 st. | 2 304 st. |
Texture units | 288 st. | 272 st. | 184 st. | 144 st. |
Raster units | 96 st. | 88 st. | 64 st. | 64 st. |
Clock frequency | 1 350 MHz | 1 350 MHz | 1 515 MHz | 1 410 MHz |
GPU Boost | 1 770 MHz | 1 545 MHz | 1 710 MHz | 1 620 MHz |
Computational power | 16 312 GFLOPS | 13 448 GFLOPS | 10 068 GFLOPS | 7 465 GFLOPS |
Memory bus | 384-bit | 352-bit | 256-bit | 256-bit |
Memory amount | 24 GB GDDR6 | 11 GB GDDR6 | 8 GB GDDR6 | 8 GB GDDR6 |
Memory frequency | 14 000 MHz | 14 000 MHz | 14 000 MHz | 14 000 MHz |
Memory bandwidth | 672 GB/s | 616 GB/s | 448 GB/s | 448 GB/s |
Power supply | 8+8-pin | 8+8-pin | 8+6-pin | 8-pin |
TDP | 280 W | 250 W | 215 W | 175 W |
Rec. Award | 2 499 USD | 999 USD | 699 USD | 499 USD |
* Values apply to Nvidia’s Founder’s 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 the Geforce RTX 2080 Ti, the Titan RTX builds around the TU102 circuit, but like the Quadro RTX 8000 in its full-featured design. In other words, all 72 SM clusters are active, giving a total of 4,608 CUDA cores, 288 texture units, 576 Tensor cores and 72 RT cores.
In addition to more computing devices than the RTX 2080 Ti, the model comes with a higher specified GPU Boost, 1,770 MHz or 135 MHz higher than the RTX 2080 Ti in Nvidia’s overclocked Founders Edition. This for a theoretical performance of 16,312 GFLOPS or up nearly 15 percent.
This is flanked by a full-scale 384-bit memory bus with GDDR6 memory at an efficient clock frequency of 14,000 MHz, which gives a theoretical maximum bandwidth of 672 GB / s. The memory capacity is also significantly increased to 24 GB, from “only” 11 GB with the RTX 2080 Ti.
The significantly higher specifications also mean higher power consumption, where Nvidia officially states 280 W. This is 20 W over the RTX 2080 Ti Founders Edition. Despite this, the graphics card appears to retain the same radiator design, except for its gilded exterior.
In terms of picture connections, they are the same as before, as the model is equipped with three Displayports, an HDMI and a USB Type-C according to the cable standard Virtuallink. The latter is used for VR, where a single cable must be able to handle both image, current and data.
Like many previous Titan models, it can be played on and is probably the world’s fastest graphics card in that respect. However, the target group for the model is not primarily players with well-stocked wallets, but perhaps mainly creators and those who are not prepared to take the step over to Nvidia’s Quadro series.
Together with Titan RTX, Nvidia is launching a matching Nvlink bridge, with the same gilded color scheme as the graphics card itself. This delivers a bi-directional bandwidth of 50 GB / s (total 100 GB / s) and makes it possible to merge GDDR6 memory from two graphics cards into a common pool of a total of 48 GB.
Nvidia Titan RTX will be launched later in December and has a recommended price tag of SEK 28,490 directly from Nvidia’s online store.