Nvidia Turing GeForce 2080 (Ti) architecture review

Graphics cards 1048 Page 1 of 11 Published by



NVIDIA Turing architecture overview
NVIDIA GeForce RTX 2080 and 2080 Ti

In this article, we'll take a deep dive into the technology and specifications that is the GeForce RTX series graphics cards and Turin GPUs. This is an architectural deep dive into the Turing graphics processors, and of course, we can share with you all specifications, you know the nitty-gritty stuff including all details you want to know about the pending GeForce RTX 2070, 2080 and 2080 Ti. We've got a lot of new stuff to cover as the Turing architecture of the new GPUs offers a fundament change and maybe even a shift in the graphics card arena as next to your normal shading engine. You heard all about it, NVIDIA has added RT (Raytracing) cores, as well as Tensor (AI), cores onto the new GPUs, and these are active on the three announced products. Will Turing be the start of the next 20 years of gaming graphics? Well, that all depends on the actual adoption rate in the software houses, the guys and girls that develop games and a dozen or so DirectX Raytraced compatible games are in development an equal amount of them will make use of deep learning DLSS running utilizing the Tensor cores. Turing predominantly is all about adding Hybrid Raytracing in the rendering pipeline though. That has got to be my opening paragraph for this article. So hey all, and welcome to the first in a long row of RTX reviews. We'll start off with a technical overview of the reference cards and GPUs, of course. It is the 14th of September 2018, that wait is almost over. But before we begin, let's quickly have an overview as to what three graphics cards NVIDIA actually has announced. 

GeForce RTX 2070

NVIDIA has announced three GeForce RTX graphics cards, starting at 499 USD you are going to see the GeForce RTX 2070. I immediately need to address pricing. Starting at means the cards from the board partners. These (should) start at 499 USD. The reference aka founders editions will cost you 599 USD. The GeForce RTX 2070 is a graphics card rated at a 185 Watt. The GeForce RTX 2070 based on the TU106, including the some RT Cores, Turing Tensor Cores, and all of the architectural changes made to the Turing SM. This card does not offer NVLink, and thus does not have SLI support. The GeForce RTX 2070 is based on the full implementation of the TU106 GPU, which contains three GPCs, 36 SMs, and eight 32-bit memory controllers (256-bit total). Each SM includes 64 shader processors, 96 KB L1 data cache/shared memory cache, and four texture units. The full TU106 GPU contains 10.8 Billion transistors and includes 2304 CUDA Cores, 288 Tensor Cores, and 36 RT Cores. It has a 1410 MHz base clock and a 1710 MHz boost clock for the standard founder version. The card has 8GB of GDDR6 memory running a 256-bit wide bus (448 GB/s).

GeForce RTX 2080

You will see two products launched in the first launch wave on September the 20th 2018, the GeForce RTX 2070 we just talked about will follow later in October. Next up, the GeForce RTX 2080. This graphics card series will not have the full shader count and GPU as the 2080 Ti, in fact, it even has its own GPU called the TU104. This card will get 2944 shader processors (also referred to as Stream or Cuda cores) active based on that GU104 GPU running a base clock of 1515 MHz with Boost frequencies running upwards to 1800 MHz. The TU104 chip contains six GPCs, 48 SMs, and eight 32-bit memory controllers (256-bit total). Each SM includes 64 shader processors, 256 KB register file, 96 KB L1 data cache/shared memory cache, and four texture units. The full TU104 chip contains 13.6 Billion transistors and includes 3072 shader processors, 368 Tensor Cores, and 48 RT Cores. So yes, the RTX 2080 is cut down. One x8 NVLink link is activated providing 25 GB/sec of bandwidth in each direction (50 GB/sec total bandwidth). Mind you that these will different here and there, AIB partners obviously have different factory tweaked products. The GDDR6 memory will get tied towards a 256-bit bus, and depend on the clock frequency, we are looking at 448 GB/sec. It is a product that will start at 699 USD for the AIB partner products, and 799 for the founder editions. This product will have a 225 Watt TDP.


GeForce RTX 2080 Ti

The new consumer flagship graphics card from NVIDIA is the GeForce RTX 2080 Ti, this product series has been fitted with the TU102 GPU. This dawg of a GPU will have 4352 active shader processors, which is substantial when compared with the GeForce GTX 1080 Ti. The product will get 11GB GDDR6 graphics memory, and with those 11 GB you will get a 352-bit wide memory bus. The GPU has six Graphics Processing Clusters (GPCs), 36 Texture Processing Clusters (TPCs), and 72 Streaming Multiprocessors (SMs). Each SM contains 64 CUDA Cores, eight Tensor Cores, four texture units, and 96 KB of L1/shared memory which can be configured for various capacities depending on the compute or graphics workloads. Ray tracing acceleration is performed by a new RT Core processing, the TU102 has 72 of them with 576 tensor cores and 96 ROP units. For clock frequencies, we're looking at a 1350 MHz base frequency, with Turbo allowance towards 1635 MHz. Keep some margin in mind for the board partner clock frequencies. the AIB products should start at 999 USD, the founders' edition will start at 1199 USD.

That said and done, let's have a closer peek at the products, its architecture, and precise specifications. Next page, please. 

Share this content
Twitter Facebook Reddit WhatsApp Email Print