Nvidia GeForce RTX 2070

STRONG POINTS
Very effective to play in WQHD any graphic options enabled.
Energetic efficiency.
The promise of hybrid rendering.
VirtualLink output.
WEAK POINTS
Still little visibility on raytracing performance.

 CHARACTERISTICS

Puce	TU106
Fréquence GPU	1410 MHz
Amount of memory	8 Go
Memory type	GDDR6
Memory frequency	1750 MHz
heatsink	double height

GeForce RTX 2070 is at the core of the RTX 20 series of Nvidia graphics cards. Historically, cards with the suffix "70" are the most prominent players because sold at a rate much more content than their counterparts "80" and "80 Ti". This model is intended primarily for users of WQHD monitors or those who want to play in Full HD while pushing the most graphic details. It remains to be seen how this new model is positioned, which also offers compatibility with hybrid rendering games, combining classic rendering and raytracing.

Presentation

The GeForce RTX 2070 uses a TU106 chip, a GPU smaller than the TU104 and TU102 which equip respectively GeForce RTX 2080 and GeForce RTX 2080 Ti. This GPU is, like all GeForce RTX 20 GPUs, a Turing architecture chip. We will not go back in detail on the novelties brought by this architecture, but to summarize the key points. Turing brings a new method of distribution of computing units, units dedicated to the IA (Tensor Cores), units dedicated to the acceleration of rendering raytracing (RT Cores) as well as an optimization of the different levels of cache.

GeForce RTX 2080 Ti	GeForce RTX 2080	GeForce RTX 2070	Titan Xp	GeForce GTX 1080 Ti	GeForce GTX 1080	GeForce GTX 1070
Architecture	Turing	Pascal
GPU	TU102	TU104	TU106	GP102	GP104
Gravure	12 nm	16 nm
Unités de calcul	4 352	2 944	2 304	3 840	3 584	2 560	1 920
Unités Tensor	544	368	288	-
Unités RT	68	46	36	-
Unités texture	272	184	144	240	224	160	120
Unités de rendu	88	64	96	88	64
Fréq. base	1 350 MHz	1 515 MHz	1 410 MHz	1 405 MHz	1 480 MHz	1 607 MHz	1 506 MHz
Fréq. Boost	1 545 - 1 635 MHz	1 710 - 1 800 MHz	1 620 - 1 710 MHz	1 582 MHz	1 733 MHz	1 582 MHz
Mémoire	11 Go GDDR6	8 Go GDDR6	12 Go GDDR5X	11 Go GDDR5X	8 Go GDDR5X	8 Go GDDR5
Fréq. mémoire	1 750 MHz	1 426 MHz	1 375 MHz	1 250 MHz	2 000 MHz
Bus	352 bits	256 bits	384 bits	352 bits	256 bits
Bande passante	616 Go/s	448 Go/s	548 Go/s	484 Go/s	320 Go/s	256 Go/s
TDP	250 - 260 W	215 - 225 W	175 - 185 W	250 W	180 W	150 W
Tarif lancement	1 259 €	849 €	639 €	1 349 €	824 €	789 €	499 €

GPU Boost 4.0: a slight evolution for more control

Nvidia has nevertheless revised its frequency adjustment system and passes GPU Boost in its fourth iteration. With GPU Boost 4.0, there is still talk, as on GPU Boost 3.0, to adjust the operating frequency according to the applied load, a threshold of power consumption and the temperature of the chip. The change is in the temperature management, which has an intermediate temperature threshold before which the frequency passes to its nominal value. According to Nvidia, this would maintain a higher frequency than previously when the cooling system of the graphics card is efficient.

Performance in games

As usual, we run our video game card tests inserted into a closed computer. We then let the board warm up for several tens of minutes before making our measurements so that the frequency stabilizes - this is higher in the first few minutes, when the board is not hot enough. Under these conditions, our MSI GeForce RTX 2070 Armor sees its frequency oscillate between 1680 and 1740 MHz for an average of 1700 MHz. It's about 50 MHz less on average than what we see on the GeForce GTX 1070 and it's comparable to the GeForce GTX 1080.

Nvidia primarily targets the GeForce RTX 2070 to monitors in WQHD definition, at best. To play in 4K, it is always up-to-the-end that you have to turn - GeForce RTX 2080 and, especially, GeForce RTX 2080 Ti being better equipped for such a definition.

On average, in WQHD, the MSI GeForce RTX 2070 Armor is 38% faster than a GeForce GTX 1070 reference. It is also 19% faster than a GeForce GTX 1070 Ti and even 11% faster than a GeForce GTX 1080 Founders Edition - which brings it to the level of the best overclocked custom GeForce GTX 1080s, which are up to 10% faster than the reference model. The GeForce RTX 2080 still remains 23% faster on average on our panel of games. On the competition side, the Radeon RX Vega 64 is 19% behind average.

If we switch to Full HD, the gain is more measured since an average of 29% compared to the GTX 1070, 11% compared to the GTX 1070 Ti and 2% compared to the GeForce GTX 1080. The Radeon RX Vega 64 stays behind with a 19% lower average.