We continue our acquaintance with the new series of GeForce RTX graphics accelerators. Next in line is NVIDIA’s gaming flagship in the face of the GeForce RTX 2080 Ti. It is the most powerful Turing architecture-based graphics card that raises the bar for performance and supports all the latest NVIDIA technologies. In this review, we will explore the capabilities of the GeForce RTX 2080 Ti Founders Edition, check the overclocking potential and compare gaming performance with the GeForce RTX 2080 and GeForce GTX 1080 Ti.
Since the architecture and technical features of video cards were discussed in a separate review, we briefly recall the main points. At the heart of the GeForce RTX 2080 Ti is the TU102 GPU, slightly trimmed in compute units. The top-end video adapter has 4352 CUDA stream computing cores, 544 tensor units for deep learning operations, 88 RT units for ray tracing acceleration, 272 texture units and 88 render units. 11 GB of GDDR6 memory communicate with the processor via a 352-bit bus. The effective memory frequency is equivalent to 14 GHz. For the GPU, a base frequency of 1350 MHz is declared with an average Boost Clock of 1545 MHz or 1635 MHz.
The different frequencies require a little explanation. Boost at 1635 MHz is declared for solutions manufactured by NVIDIA itself, which belong to the Founders Edition line. And simple models from other manufacturers will be content with the 1350/1545 MHz formula. At the same time, many versions from partners have already been announced, where there is factory overclocking to the same 1635 MHz or higher values. Depending on the frequency, a different TDP level is also announced – 260 W for the Founders Edition and 250 W for the simplest GeForce RTX 2080 Ti. The recommended price of the Founders Edition is also higher, while the video card is immediately equipped with a cooler with two fans, and not a standard “turbine”.
GeForce RTX 2080 Founders Edition
Visually, the flagship video card resembles the GeForce RTX 2080. Both are made in the same design with a combination of black and silver elements.
GeForce RTX 2080 Ti occupies the usual two slots. The total length is almost 27 cm. The GeForce RTX 2080 Ti Founders Edition looks somewhat simpler than the old GeForce GTX 10 models, but in general outlines, some continuity and a familiar industrial style in design are recognized.
The reverse side is completely covered with a silvery metal plate. It has smooth rounded edges, forming a common design with the body.
The connector for connecting a second video card in SLI mode is closed with a special plug. Data exchange takes place via the high-speed NVLink interface.
In the corner there are two eight-pin connectors for connecting external power.
The large GeForce RTX inscription on the side face is highlighted in green. It is possible to control the backlight through specialized utilities for working with video cards.
The panel with connectors has a small perforation. There are five connectors themselves: three DisplayPort ports with support for the 1.4a standard, one HDMI 2.0b and VirtualLink (USB Type-C).
Let’s take a closer look at the cooling system. The bottom of the cooler looks familiar. This is a massive black base that is in contact with all the hot elements on the board. In the base window there is a protruding area of the copper evaporation chamber, which removes heat from the GPU. All wires are collected in one loop, which goes somewhere inside the structure. It provides the operation of the backlight and fans.
Various thermal interfaces are used. Greenish thermal pads are mainly involved. They provide contact with memory chips, transistors and tantalum capacitors. The chokes use a viscous blue substance, but white thermal padding is used for the center elements.
The blue thermal interface is also visible at the point of contact between the heatsink and the cooler base.
As for the complete analysis of cooling, this is a non-trivial task. The structure of the cooler is such that a number of screws are hidden under the front black panel with the name GeForce RTX 2080 Ti, and the panel itself is held on with glue. We decided to limit ourselves to a general inspection and the official image of the radiator. It can be seen that this is a massive structure. The heat pipes are replaced by the evaporation chamber.
The back plate is also involved in cooling. There are several thermal pads on the inner surface that provide contact with the board in the area under the memory chips and the power unit. The rest of the surface is covered with a dielectric black film. It is noteworthy that one small thermal pad (on the far right in the picture) is superimposed directly on this film, and clearly does not provide direct contact with the metal surface.
The printed circuit board has a complex design, with both surfaces dotted with elements. There are 13 power phases in total.
The GPU power system has 10 phases.
Full marking of the processor TU102-300A-A1-K1. This is a massive crystal with an area of 754 square meters. mm.
11 GB of memory is made up of Micron chips marked 8PA77 D9WCW.
Base core clock 1350 MHz with Boost Clock 1635 MHz. The effective frequency of the GDDR6 memory corresponds to 14000 MHz.
For monitoring and overclocking, the EVGA Precision X1 utility was used. It has a separate fan control function, backlight settings and is supported by NVIDIA Scanner to determine the optimal overclocking.
The video card was warmed up by gaming applications. The tests were carried out on an open stand at an indoor temperature of 21–22 °C. When loaded with a long 14-minute Final Fantasy XV Windows Edition test at 4K with Ultra settings, the core temperature reached 80-81 ° C with a maximum fan speed of about 2200 rpm. GPU frequencies in this hard test were close to 1700 MHz or slightly higher, with occasional drawdowns up to 1680 MHz. Rare peak values were noted before 1965 MHz.
In a number of games, the frequency level could be slightly higher. For example, in Deus Ex: Mankind Divided, frequencies were kept at the level of 1785-1800 MHz, and the fan speed was 2% lower.
The noise is moderate, but after the excellent performance of the GeForce RTX 2080 Founders Edition from the older graphics card, we expected something more impressive. In harsher conditions, heat may increase, but this should be compensated by a slight increase in rpm. It should be noted that the temperature limit of the video card is at 84 ° C, so heating will have a minimal effect on Boost. And in general, we have lower operating temperatures compared to older top models with turbine cooling.
After testing in NVIDIA Scanner, we got +162 MHz, but in the end we used a 170 MHz core overclock.
After a number of tests, we settled on a base value of 1520 MHz, which gives an average Boost Clock of 1805 MHz. The memory has been accelerated to 15560 MHz. The power limit has been increased to the maximum.
With these settings in Final Fantasy XV at maximum quality, the core frequencies ranged from 1920-1980 MHz.
To keep operating temperatures within 80 °C, we had to fix the fan speed at 80% of the maximum, which is associated with a very high noise level.
In some games, the mids were kept at 2000 MHz or higher, as in Deus Ex: Mankind Divided. Peak Boost values reached 2130 MHz.
Based on the results of the review, an interesting observation can be made that the GeForce RTX 2080 Ti Founders Edition does not differ much from the GeForce GTX 1080 Ti Founders Edition in terms of GPU frequencies in Boost mode. And this is true both for nominal and for overclocking. The performance benefits will become clear from testing.