AMD Radeon VII and exams from around the world. Can he compete with Nvidia? – Živě.cz

At CES this year, AMD introduced a new border graphic card Radeon VII, which is also the first graphic chart with a 7nm skip. The latest models in Nvidia are still the responsibility of the 12m production process.

But while Nvidia has been a long time in the decline of a long time, AMD has been experiencing # 39; promises a new generation of graphics chart for more competition and a harder battle.

Waiting for Navi

Radeon VII is still built on GCN's fifth generation in the latest Vega 20 (the second one of Vega). This is not yet the generation of Navi, which may be coming this year, and will bring many new technologies and developments.

The new end-end model includes 7nm skipper made in TSMC 13.2 billion transisers and 3,840 total stream processes, with a basic frequency of 1.4 GHz and up to 1,750 MHz in turbo mode. A & # cart; use HBM2 fast memory with 1 TB / s by a measure of 4,096 measurements and abilities 16 GBSouth-West This is the main advantage of Nvidia's electronics, which offers only 11GB of memory, which may have been some of the games on the edge. At the same time it implies restrictions for general calculations that are required.

Theoretical computing power of the 13.8 TFLOPS card in SP and 3.46 TFLOPS in DP, which is a huge jump over the Radeon RX Vega 64, can only handle 794 GFLOPS in the DP. AMD has made a small decision to buy the card for customers who want to calculate and calculate.

Its TDP is 300 W and its description is & # 39; including cooling by three fans. What is important is the good airflow in the cabinet, because it is essential to get away from the cabinet with the amount of heat the card will have; generated during the crowd.

Outputs are generally shared with three showcases 1.4 and one HDMI with 4K @ 60 support. Compared to the Nvidia competition, AMD has a " also hitting that her cart is already in her; Supporting the new generation of PCI Express 4.0. External PCI Express power is required to use two osmipins.

There is no ray, very fast disease that can be broken up by Microsoft

As AMD said several times, real-time wind management, as Nvidia puts GeForce RTX's graphics cards, is too tough and does not work hard in the skeleton because it's a # 39 ; means less space for classical computing units.

However, Nvidia also has introduced the DLuring system to incorporate Turing architecture, where tensor skin and specialistic artistic experience can be resolved without any impact on performance as the classic classic system. Thanks to this, DLSS can find the same beautiful or better image, with fps higher.

AMD has also decided to support this method, but in collaboration with Microsoft, it works on a wider form of DirectML, which should be incorporated into the future in DirectX API 12. You might win a larger level, and Nvidia may expect. This new cleaning system should also be on new game consoles that are likely to be built on AMD technologies.

World trials

And this time, the new Radeon VII exams were edited from around the world, so that we can look at its achievements from different angles – both in terms of different games and aspirations and demands and other features.


3DNews, 4Gamer, Anandtech, BitTech, ComptoirHardware, ComputerBase, Engadget, GDM, Golem, Guru3D, HardwareBattle,, HardwareLuxx, HotHardware, Hexus, KitGuru, Lab501, LegitReviews, LesNumeriques, Lowyat, PCGamesHardware, PCPerspective, PCWorld, Phoronix, SweClockers, TechGage, TechPowerUp, TechReport, TomsHardware, Tweak, Tweakers, TweakTown, WASD

Video Recognition:

BitTech, BitWit, BPS Customs, DerBauer, GamersNexus, HardwareCanucks, HardwareUnboxed, JayzTwoCents, JokerProductions, KitGuruTech, LinusTechTips, OptimusTech, PaulsHardware, TechYesCity,

As you can see, most of the games are AMD Radeon VII, similar to the GeForce GTX 1080 Ti. means that its achievement is slightly lower (about 10%) than the GeForce RTX 2080. But AMD is definitely doing computing applications that make better than graphics skills that & # 39 ; competition.

Unfortunately, efficiency is not a very strong part of this module, since the tightness is much worse than the Nvidia modules and this applies to the power usage of more than 16n GeForce RTX 1080 Ti FE. It is only seen that architecture is unique from efficiency and Nvidia has a strong direction on this side.

How is all associated with the price fixed?

High price

AMD added the final price for Radeon VII to $ 699, for example, at it means a price about 19 thousand crownsThe post is the problem, at the same price today, that you are going to do. Buying GeForce RTX 2080, with only 8GB of memory, but offers higher performance, lower energy measurement and lower noise.

So if it's the main player, the new model is not worth the current price and it's best to go to # 39; competition. The only thing is that if you know you need more graphic memory – usually for some calculations. In this case, 16 GB of HBM2 for Radeon VII is the most appropriate option.

AMD would make sure it would be better if a & # 39; price below and down below, where the graphics could compete, for example, GeForce RTX 2070. But this may still be, but the new cards are limited and the sale is still on sold. As soon as the product is made, it may be that a & # 39; price to fall.

Source link