Radeon RX 6800 XT: AMD's return to the high-end segment
1.12.2020
Translation: machine translated
With the Radeon RX 6800 XT, AMD is finally targeting the high-end sector again. And the attack is successful: the card can keep up with the RTX 3080 in games and even outperform it in some cases.
Arthur Morgan rides through the fictional United States in "Red Dead Redemption 2". It looks brilliant. But it's not the marvellous landscapes that catch my eye. It's the FPS counter. The impressive values are not conjured up on the screen by an Nvidia graphics card, but by the AMD Radeon RX 6800 XT. I can't help but smile. AMD is back.
The GPU in detail
Navi 21, as the GPU is called, is still manufactured by TSMC using the 7 nm process. Compared to the 5000 series, it has a significantly larger GPU die with 519 mm². There are 72 compute units (CUs) on the RX 6800 XT, each with 64 stream processors, which corresponds to a total of 4608 shaders.
The clock rate has been significantly increased for the 6000 series. The 6800 XT has a gaming clock of 2015 MHz and a boost clock of up to 2250 MHz. By comparison, the gaming clock of the RX 5700 XT is 1755 MHz.
AMD has also increased the memory capacity: 16 GB GDDR6 memory is on the board. The memory clocks at 16 Gbps and employees a 256-bit memory interface, resulting in a total bandwidth of 512 GB/s. Added to this is the 128 MB Infinity Cache, which increases the effective memory bandwidth without excessive power consumption.
The 6800 XT has a total board power (TBP) of 300 W. This corresponds to an increase of 75 W compared to the RX 5700 XT. What stands out is how much AMD has optimised the performance: RDNA 2 is said to work 30 per cent more efficiently than RDNA.
In contrast to the dedicated RT cores at Nvidia, Navi 21 has one Ray Accelerator per CU. These ray accelerators are said to offer ten times higher ray tracing performance than with shaders alone. The advantage of this architecture according to AMD: RDNA 2 can reserve more die area for functions that support both ray tracing and non-ray tracing gaming performance. This also allows higher clock frequencies to be achieved. (such as Infinity Cache), and achieve higher clock speeds for a given power budget.
Infinity Cache
In simple terms, Infinity Cache is on-die cache memory. It is located between the smaller L1 and L2 caches, which are also located on the graphics processor itself, and the wider GDDR6 on the card. Infinity Cache bridges the gap between the small L1 and L2 caches and the VRAM. You can therefore casually regard the Infinity Cache as an L3 cache, which quickly supplies the GPU with data and reduces the utilisation of the frame buffer.
AMD has chosen this approach in order to be able to use expensive and performance-hungry memory with a 512-bit memory interface. Infinity Cache is designed to deliver more than twice the bandwidth of a 384-bit GDDR6 interface with a minimal increase in performance. All in all, AMD claims up to 2.4 times higher bandwidth per watt.
Smart Access Memory
Smart Memory Access - SAM for short - gives the new Zen 3-based Ryzen processors of the 5000 series full access to the GPU memory. However, this requires the CPU to be paired with a Radeon RX 6000 series graphics card. Until recently, a motherboard with a 500-series chipset was also required. However, there are now already 400 series boards that support SAM.
Normally, CPU access to the GPU memory is limited to a block of 256 MB. Smart Access Memory removes this restriction and efficiently utilises the combined memory of the CPU and GPU. This reduces buffering and latency. Which should deliver five to ten per cent more FPS, depending on the game and resolution.
Design and other features
The RX 6800 XT is kept simple. Only the Radeon logo is illuminated in AMD red. The front of the card is also framed in red. In contrast to the RX 5000 series, the RX 6800 XT is no longer cooled by a radial fan, but by three axial fans. Two 8-pin connectors supply the card and the board with power.
At 26.70 centimetres in length, it is rather short for a card with three fans. However, the thickness of 5 centimetres or 2.5 slots is likely to cause space problems in more compact systems.
In terms of connectivity, one HDMI 2.1 and two DisplayPort 1.4 ports are available. USB-C is also on board. Your power supply should have at least 650 W for the card.
I am testing the card on my DimasTech Easy V3.0 benchtable with the following components:
Unfortunately, I don't have any suitable comparison data for the RX 5700 XT. I used different versions of the benchmarks when I tested it at the time and unfortunately the card is no longer available to me for retesting. However, as the 5700 XT is designed for the mid-range segment anyway, it is not really comparable with the high-end 6800 XT card.
Synthetic gaming benchmarks, volume and temperatures
In the synthetic benchmarks from 3D Mark, I only give the graphics score. The benchmark is very inconsistent with the overall score, which is why the result is distorted. I run the benchmarks three times each and select the best result. Here are the results of the Time Spy and Fire Strike benchmarks:
In the Fire Strike benchmarks, the RX 6800 XT clearly outperforms its direct competitor, the RTX 3080, by 18 and 14 per cent respectively. The RTX 3090 also outperforms the AMD card in the Fire Strike benchmark. In Fire Strike Ultra, it is only three per cent behind Nvidia's flagship. In the Time Spy benchmark, the wind shifts in Nvidia's favour. Here, the RTX 3080 is on average around four per cent ahead of the RX 6800 XT. The RTX 3090 even outperforms the AMD card by around 21 per cent. The card remains very quiet at 39 dB.
The RX 6800 XT reached a maximum temperature of 75° Celsius in Time Spy. The average was 71° Celsius. Compared to the Asus TUF RTX 3080, these are rather poor values. The card with the Nvidia GPU reached a maximum temperature of 65° Celsius. Since I did the test on the open testbench, higher values are to be expected in a case. Here is the heat development on the back of the PCB during the approximately two-minute Time Spy demo:
As far as the noise development of the two cards is concerned: If I turn the three axial fans up to 100 per cent, they generate 62 dB. This is rather loud and can be annoying. Fortunately, the fans never run at 100 per cent when gaming and even if they do: If the card is in the case, the sound is still somewhat muffled. If I run the Time Spy benchmark with the fans at 100 per cent, I get a Graphics Score of 17,705 points. That's only one per cent more than with the fans on standard. The card reaches a maximum temperature of 61° Celsius and an average of 58° Celsius. The temperature differences therefore don't play a major role in this case.
In order to be able to make a better statement about the cooling performance of the cards, I test them normalised by volume. I lower the percentage output of the fans step by step until I measure 40 dB at a distance of 30 centimetres from the fans. The fans of the RX 6800 XT run at 54 per cent of their maximum power. People consider 40 dB to be quiet. If I run Time Spy like this, I achieve a score of 17,400 points. The card gets a maximum of 73° Celsius and an average of 69°. The card runs less warm than on the standard fan profile, but I can't match the result on standard.
I can't say anything about the card's consumption. I don't have the appropriate tools to provide exact information.
Application: Puget Systems Photoshop Benchmark
The Photoshop benchmark uses the following reference workstation as the basis for calculating the scores:
- Intel Core i9 9900K 8 Core
- NVIDIA GeForce RTX 2080 8GB
- 64GB RAM
- Samsung 960 Pro 1TB
The results of the reference workstation can be used to estimate how well other systems perform. My test benchmark with the Radeon RX 6800 XT achieves the following results:
The RX 6800 XT is in a neck-and-neck race with the RTX 3080 and RTX 3090. Only a few points separate the graphics cards from each other.
Application: Puget Systems Premiere Benchmark
In contrast to the Photoshop benchmark, the test benchmark here does not compete against a reference workstation. With the Puget Systems Premiere Benchmark, the score is calculated relative to the existing frame rate of the test videos. If the test video has an FPS of 29.97 and the system renders it at 29.97 FPS, this means 100 points. If it is only 14.98 FPS, 50 points are awarded.
The benchmark runs media in the formats 4K H.264 with 150 Mbps in 8 bit (59.94 FPS), 4K ProRes 422 16 bit (59.94 FPS) and 4K RED (59.94 FPS). He tested live playback in Adobe Premiere Pro and the export. A value of 100 is the maximum for live playback, as Premiere cannot play back the media faster than specified. For export, on the other hand, over 100 points are feasible, as rendering is not limited to the FPS of the media.
This is how the RTX 3090 compares to the 2080 Ti and 3080:
In Premiere Benmark, the difference between the RTX 3080 and RTX 3090 is greater than in Photoshop: the AMD card is 20 and 28 per cent behind Nvidia respectively.
Blender
The difference to the competition from Nvidia is also significant in the bmw27 Blender benchmark. The RX 6800 XT takes 12 and 16 seconds longer respectively to render the test scene.
The games
The difference to the RTX 3080 is small. However, the RX 6800 XT is slower than the competition from Nvidia in all three resolutions tested. In 1080p it is two per cent, in 1440p five per cent and in 2160p resolution six per cent. All games are tested without SAM so that AMD does not have an unfair advantage.
Except for "Shadow of the Tomb Raider" and "Deus Ex: Mankind Divided", the RTX 3080 is always faster than the RX 6800 XT.
When it comes to ray tracing, AMD can't keep up with Nvidia. The RX 6800 XT is far behind Nvidia in the ray tracing game "Shadow of the Tomb Raider".
And SAM?
As I have a Ryzen 9 3900X on the testbench, I am unfortunately unable to test SAM in this configuration. The feature only works with Ryzen 5000 CPUs, so the results would not be comparable with those of the Nvidia GPUs. And I had to return the GPU after the test. Unfortunately, night tests with these are also not possible. However, I loaded the testbench with the 5800X and ran "Shadow of the Tomb Raider" and "Civilization VI: Gathering Storm" over it. In 1080p resolution, I had around ten per cent more FPS with "Civilization VI: Gathering Storm" and around seven per cent more FPS with "Shadow of the Tomb Raider". This advantage decreases with higher resolutions. At 1440p resolution, it is still six per cent for "Tomb Raider" and four per cent FPS for "Civilization VI". In 2160p resolution, it's only around two per cent for "Civilization VI" and four per cent for "Tomb Raider". SAM is definitely a nice feature that can give you a few more FPS. If that's what you want, I'll run the test games with RX 6800 XT and Ryzen 5800X and SAM enabled in a separate article. Just write it in the comments column.
Conclusion: AMD is finally competitive again
When Nvidia introduced the RTX 30 series, I didn't think it was possible for AMD to (almost) catch up. For years, the Red Team was behind Nvidia. With RDNA and the Radeon RX 5700 XT, AMD was at least in the middle class. As a former ATI graphics card enthusiast, I'm delighted that AMD is finally back in the upper class.
The 6800 XT is first and foremost a gamer card. It holds its own against the RTX 3080 in classic rasterisation games, but when ray tracing comes into play, the strength of Nvidia's dedicated RT cores becomes apparent. However, I can imagine that AMD's ray tracing performance will improve in future games. After all, the next-gen consoles Playstation 5 and Xbox Series X/S also support ray tracing on the RDNA-2 architecture. Many games are likely to be optimised for this in the future.
In terms of applications, things are looking less favourable for AMD. In the case of Premiere, this is mainly due to the fact that Nvidia cards with CUDA cores are optimised for the programme.
AMD or Nvidia? In terms of FPS alone, Nvidia still has the better cards. However, AMD offers a good overall package in combination with a CPU from the Ryzen 5000 series. If you are purely interested in playing classic rasterisation games, I recommend the Radeon card in combination with a Ryzen 5000 CPU. If you want ray tracing and use applications such as Blender or Premiere, I recommend Nvidia.
The release price of 720 francs is fair for the card and the performance. Unfortunately, the reference model is hard to come by and custom cards are also in short supply. What's more, the prices are higher than expected due to the shortage. This is all very annoying and tarnishes the impression of what is actually a good card. It will probably take a little longer before things settle down. That's why I can't recommend buying the card at the moment. At the moment, it makes sense to wait and keep an eye on the price development.
Alternatively, digitec will be giving away the option to buy other reference card models that have subsequently come into stock in a few days' time. Our colleague Yannick Cejka will keep you up to date. Follow him as author.
Kevin Hofer
Senior Editor
kevin.hofer@digitecgalaxus.chFrom big data to big brother, Cyborgs to Sci-Fi. All aspects of technology and society fascinate me.
Computing
Follow topics and stay updated on your areas of interest
Gaming
Follow topics and stay updated on your areas of interest
31 comments
later