The prices for graphics cards are in a moderate descent. Reason enough to see which model is perfect for your wallet: AMD RX 6700 XT vs Nvidia RTX 3070.

The time of great sadness in the graphics card market is slowly coming to an end. The hype for cryptocurrencies is gradually fading, making graphics cards uninteresting for crypto miners. A boon for weary gamers looking for a new image accelerator.

The all-important question for most buyers is which graphics card offers the best value for money. As usual, the two top dogs AMD and Nvidia compete for the throne in their respective price segments.

The upper middle class: RX 6700 XT vs RTX 3070

In the upper middle class around 500 euros, the AMD RX 6700 XT and the Nvidia RTX 3070 face each other. The AMD image accelerator currently (as of June 2, 2022) costs around 550 euros, the cheapest RTX 3070 is available to buyers from 620 euros plus shipping costs. Makes a whopping price difference of 70 euros. But it should be said that the recommended retail price of both cards is only 20 US dollars apart. At launch, AMD charged $479 for the RX 6700 XT and Nvidia $499 for the RTX 3070.

Both models naturally have a different architecture. The RX 6700 XT uses AMD’s RDNA 2 architecture. Some gamers already know them from the Playstation 5 and the Xbox Series X. According to AMD, what makes them special is the so-called infinity cache. The cache is the component of the graphics card that the graphics card processor uses to retrieve data from the main memory. In theory, the higher the bandwidth and the larger the cache, the more data the chip can handle. The RX 6700 XT gives AMD 96 MB “Infinity Cache”, while the RTX 3070 has “only” 6 MB L2 cache. Rumor has it that Nvidia is equipping the graphics cards in its 4000 series with a much larger cache.

A large, high-bandwidth cache helps handle higher resolutions and details in games when the chip can compute the data fast enough. AMD relies on the Navi 22 chip for the RX 6700 XT. Of course, it is smaller than the manufacturer’s top models. It only has 40 compute units and thus exactly 2,560 shader units, i.e. arithmetic units. For comparison: The RTX 3070 naturally uses Nvidia’s Ampere architecture in conjunction with the GA104 chip. With 5,888, it offers more than twice as many shader units.

However, by no means alone do they decide how powerful a graphics card is. Because it also depends on how fast the card clocks. And this is where the tide turns: In the reference design, the RTX 3070 clocks at up to 1.7 gigahertz in boost mode. The standard clock is 1.5 gigahertz. For comparison, the RX 6700 XT: Here the boost clock is 2.5 gigahertz and the standard clock in games is 2.4 gigahertz. The RTX 3070 has many computing units that clock frugally and the AMD RX 6700 XT has a few that really make steam.

What is better now? Well, in theory, the RTX 3070 has an FP32 value of 20.3 TeraFlops and the RX 6700 XT has 12.4 TeraFlops. Clear victory for Nvidia. With the FP32 value, manufacturers indicate the theoretically possible computing power of a graphics processor. However, you must not make the mistake of equating this with the gaming performance of the graphics card. Because it doesn’t just depend on the theoretical performance, but is a combination of cache, shader units, clock rate and graphics card memory.

And this is exactly where there is another difference between the cards, because the RTX 3070 only has 8 gigabytes of GDDR6 memory, while the RTX 6700 XT boasts 12 gigabytes of GDDR6 memory. Already at the launch, critics complained that the RTX 3070 did not have enough memory, because it helps the graphics chip to process particularly large amounts of data. This is especially important when playing games in 4K resolution. However, it should be said that neither of the two cards can handle 4K resolution in high detail in new games. Absolute high-end models are needed here. 8 gigabytes of memory are sufficient for QHD and the two cards are also designed for the resolution. What the graphics cards also differ in, but not noticeably, is the power consumption. The RTX 3070 works with 220 watts and the RX 6700 XT with 230 watts TDP.

The gaming performance

The crucial question for most interested parties is which of the two graphics cards offers the better performance in games. The answer is that the RTX 3070 fires more frames per second (9 percent) onto the monitor in most titles than the RX 6700 XT, as pcgameshardware reports. The same applies if you want to enjoy the new light calculation ray tracing in games.

AMD is clearly at a disadvantage here, but this is also due to the fact that AMD gives its cards significantly fewer computing units for light calculation than Nvidia does with its RTX 3070. Anyone who wants to enjoy ray tracing to the fullest should definitely use an Nvidia card.

The RX 6700 XT can still come up with a special feature and it is called Smart Access Memory. The function allows the processor to access the graphics card memory, which increases performance in some games. But this is only possible with a Ryzen processor and PCIe 4.0. However, the feature does not compensate for the general difference in performance between the two cards.

Apart from AMD or Nvidia fan camps, the RTX 3070 is undoubtedly the more powerful graphics card. If the prices are only 20 euros apart, as in the RRP, you can reach for the RTX 3070. However, the cards currently cost 70 euros, which corresponds to a price difference of around 12 percent and for around 9 percent more performance. The RX 6700 XT is therefore the better choice for price-conscious buyers. In addition, Ryzen users get a nice feature with the RX 6700 XT with AMD’s Smart Access Memory to squeeze the last bit of performance out of the PC. But that is not the deciding factor in the purchase.


This article contains so-called affiliate links. There is more information here.