GeForce RTX 3070 and Radeon RX 6700 XT are two next-generation premium graphics cards from Nvidia and AMD. With ray tracing support and the ability to handle heavy 4K games with good performance, the two GPUs differ little in performance.
A large difference favors Nvidia in games with active ray tracing. GeForce also stands out for the availability of DLSS, an image enhancement solution that has not yet found a parallel in the Radeon. Below, Techidence compares the two cards in detail.
Nvidia GeForce RTX 3070 vs AMD Radeon RX 6700 XT
|Specifications||GeForce RTX 3070||Radeon RX 6700 XT|
|Launch||February 2021||March 2021|
|Nuclei||5,888 CUDA||2,560 stream processors|
|Speed||1,500 to 1,730 MHz||up to 2,581 MHz (2,424 MHz in games)|
|RAM||8GB GDDR6||12GB GDDR6|
|Memory speed||14 Gb/s||16 Gb/s|
|Memory interface||256 bits||192 bits|
|Bandwidth||448 GB/s||384 GB/s|
|TGP||220 Watts||230 Watts|
Of the two cards, the RX 6700 XT is more recent on the market. AMD’s hardware consists of a 40 compute unit graphics processor with a total of 2,560 stream processors divided among them. AMD’s GPU has a speed of up to 2,581 MHz in the reference design but offers a sustained speed of 2,424 MHz during gaming.
In the case of the RTX 3070, the graphics processor is composed of 5,888 CUDA processing cores and reaches a base speed of 1,500 MHz, with a turbo capable of taking the reference card to 1,730 MHz.
Both cards have some important specializations, such as support for ray tracing acceleration using hardware. In Nvidia’s case, there are 46 RT cores dedicated to the task, while the RX 6700 XT has 40 AMD’s Ray Accelerators. As with the GPU core count, it is not possible to equalize the two brands’ solutions: Nvidia’s RT cores are not equal to AMD’s Ray Accelerators, even though they perform the same task.
A very important differentiator of GeForce is the presence of Tensor cores for artificial intelligence. It is this dedicated hardware that is behind the support for DLSS, image enhancement technology that uses AI to improve the graphics and performance of supported games. AMD does not yet have a comparable alternative to DLSS.
As for RAM, the RTX 3070 comes with 8 GB of GDDR6 connected to the graphics processor via a 256-bit interface, capable of providing 448 GB/sec (gigabytes per second) of information exchange between GPU and card memory. On the Radeon, there are 12 GB of GDDR6 memory that, connected by a 192-bit connection, exchanges 384 GB/s.
In AMD’s case, the nominally lower bandwidth is offset by the generous, high-speed cache that the card applies. Called Infinity Cache, the technology reserves 96 MB of fast-access memory for the processor.
Overall the two cards are comparable. In some scenarios, such as games that support DLSS or where the user makes a point of ray tracing, the advantage is big for GeForce.
The specialized website Tom’s Hardware did a battery of tests between the two cards running games in ultra at 1080p, 1440p, and 4K and detected that the performance is, in general, very similar between the two products. There are a few games where one or the other excels over its rival, but these were exceptions in the comparison that considered 13 different titles. While AMD has an advantage in something like Forza Horizon 4, Nvidia is better in Strange Brigade or Final Fantasy XIV, for example.
The results are different when the comparisons use rendering modes with ray tracing enabled. Here, the Nvidia card is far superior, in some cases delivering twice the performance when compared to the Radeon in the same game and settings.
According to AMD, the Radeon has a TBP of 230 Watts, while the RTX 3070 registers 220 Watts. Unlike traditional TDP ratings, these values indicate the typical consumption rate of the card, signaling how much power the card will need to run at its maximum capabilities, discounting the consumption represented by the cooling system and eventual lighting.
To get an idea of the cost of this on your energy bill, you need to consider the time frame by multiplying the TBP value by the total number of hours the board has been in use. The bill won’t be precise – your Radeon or GeForce won’t spend all its time at 220 Watts – but it can serve as a basis to get a sense of how much the equipment costs in the long run.
From the hardware point of view, both are quite similar: although they can handle 4K, they are cards with a profile more directed towards 1440p games, especially considering the long term. Another similarity between the two is the presence of ray tracing acceleration and support for technologies such as FreeSync and G-Sync – the latter exclusive to GeForce.
Something that sets the two cards apart is the DLSS offering in Nvidia’s product, as well as some special features in the card’s driver and toolkit. The RTX 3070 allows streamers to capture and remove the background of scenes while playing, for example, without the need for a chroma key.
Price and availability
Both cards are found in the market, and between the two, the RTX 3070 appeared in our searches at a slightly higher value. The RTX 3070 can be found at $2,056.42 currently on Amazon. Depending on the version of the card, it is possible to find higher prices in more specific editions.
The AMD card was seen for $899.99 on Amazon. The model in question is PowerColor Red Devil, and just like GeForce, you may come across editions of the 6700 XT from other brands at higher values.
Either of the two cards is a great option for those with an eye on configuring a powerful PC in 2021 without going to the extreme of the prices charged on even more powerful cards, such as the RX 6800 and RTX 3080. From a practical point of view, both are more than enough to face recent games in ultra and with good performance – even in 4K.
Although it is hard to ignore that the RX 6700 XT is cheaper, it is worth considering that the price difference (considering the budget of those who are looking for a powerful PC) is not so great. The GeForce has better ray-tracing performance and, in addition, the Nvidia unit offers a significant differential: DLSS support.