A month after the botched release of the first two GeForce Ampere RTX 3000 Series GPUs, Nvidia is still scrambling to send out the long list of pre-ordered GPUs that they failed to have stock for. Meanwhile, their chief competitor in the GPU market, AMD, has announced their new series of RDNA 2 based Radeon graphics cards – the RX 6800, RX 6800 XT and RX 6900 XT; aiming to rival Nvidia’s new lineup, and perhaps even surpass it.
Though AMD has never truly come close to outperforming Nvidia consistently in the past, this new lot of Navi 21 GPUs is certainly different than their past attempts. Firstly, they feature a new cooling solution, with a tri-fan layout that is more akin to there competition, rather than their previous generations of GPUs. Secondly, they have a new Infinity Cache, which is based on their Ryzen CPUs’ Level 3 Cache, allowing a 256-bit memory bus to deliver more than twice the bandwidth of a 384-bit bus, while spending less energy to do so. Together with AMD’s new and improved Compute Units (CU), which have redesigned data paths, pervasive fine-grain clock gating, and a new pipeline rebalancing, AMD has also implemented a new generation of Anti-Lag, which, when in unison with Radeon Boost, can help reduce end-to-end input latency by a considerable amount.
On the other hand, Nvidia also has high-performance technologies that are, arguably, a step ahead of AMD. These features include: Nvidia Reflex, second-generation ray tracing, third-generation tensor cores, and, of course, DLSS 2.0. Nvidia’s Reflex feature, firstly, has an added advantage – over AMD’s Anti-Lag – in that it can be paired with a monitor-hardware component called the Latency Analyzer, which calculates the full amount of end-to-end input lag of a system; something that was only possible with the use thousands of dollars’ worth of dedicated hardware equipment. With the Latency Analyzer, the monitor can display end-to-end input lag to the user in real-time. Currently, the Latency Analyzer is only available in 1080p 360 Hz monitors; though it may become available for more resolutions/framerates later down the road. For e-Sport professional players, or hardcore competitive gaming enthusiasts, nothing beats the combination of Nvidia Reflex together with a 360 Hz monitor, as this combination can drop overall input lag by up to two thirds, when compared to a standard 144 Hz monitor. Next, we have Nvidia’s Tensor cores which, paired with DLSS 2.0, offer real-time resolution upscaling; with graphic fidelity and framerate performance improving along with it. Together with the second-generation of ray tracing cores that Nvidia offers (versus AMD’s first-generation RT cores) Nvidia appears to have a sizeable advantage in regards to overall technological advances and features… Except for one thing: AMD’s Smart Access Memory.
Because AMD is both a CPU and GPU manufacturer, they’ve created an architecture that makes use of the synergy between their components, for an added level of performance. If one pairs a Zen 3 Ryzen 5000 Series CPU, with a Radeon RX 6000 Series GPU, then these two hardware components can share an expanded data channel, effectively removing the GPU memory bottleneck and increasing effective framerate performance. It must be stated that AMD has an additional advantage in regards to their technological features for gaming application, due to the monopoly they hold created in the gaming console industry, as both the Playstation 5 and Xbox Series X use AMD’s CPUs and GPUs within. That means that game developers are far more likely to prefer optimizing their title based on the RDNA 2 architecture and Smart Access Memory, rather than any technology that Nvidia has to offer. It is due to this exclusivity in console hardware, that the performance boost that we see from these RDNA 2 GPUs may very well increase as time passes by.
When discussing which is the best value for performance GPU, that Nvidia and AMD have to offer, two graphics cards come to mind: The GeForce RTX 3080 and the Radeon RX 6800 XT. But, between these two models, which one has the advantage? Can AMD’s GPU surpass the processing power of the RTX 3080, and overthrow Nvidia’s long held position as the best performance GPU manufacturer? To find out, let’s compare these two graphics cards in terms of specifications (memory, rendering power/performance, power and efficiency), benchmarks, and pricing, and see who comes out on top.
RX 6800 XT vs RTX 3080: Specifications
|RX 6800 XT||RTX 3080|
|Interface||PCIe 4||PCIe 4|
|GPU Cores||4608 Shading Units||8704 CUDA cores|
|Ray Tracing Cores||72||68|
|Base clock||TBC||1440 MHz|
|Boost clock||2250 MHz||1710 MHz|
|Memory||16GB GDDR6||10GB GDDR6X|
|Memory speed||16 Gbps||19 Gbps|
There is a distinct trade-off when it comes to the memory provided by two GPUs; one that comes in the form of the quality versus quantity argument. The Radeon RX 6800 XT has 60% more VRAM than the GeForce RTX 3080, but the latter has GDDR6X memory, which has faster bandwidth speeds than the GDDR6 that the RX 6800 XT uses.
Currently, there are not many (if any) games that will make use of more than 10 GB of VRAM, but as the next-generation progresses, gaming at 4K may become more memory demanding, and that is where the GeForce RTX 3080 may run into issues. Additionally, video editing programs like Adobe Premiere have certain functions (like the Mercury Playback engine that can run in GPU mode) where the more the VRAM a GPU has, the better it can handle the task.
Additionally, though the memory bus of the Radeon RX 6800 XT is smaller than the GeForce RTX 3080’s 320-bit memory bus, with the use of the 128 MBs of AMD’s Infinity Cache allows the Radeon RX 6800 XT to close the gap. All in all, it is quite unfortunate that the GeForce RTX 3080 comes with so little VRAM for its $699 price tag – and its GDDR6X VRAM type does not save it from this drawback.
Rendering Power & Performance
The Radeon RX 6800 XT comes equipped with 4,608 shading units, 288 TMUs, 128 ROPs, 72 RT cores, and 72 CUs. This renders it capable of a theoretical performance of 41.47 TFLOPs, 20.74 TFLOPs, and 1.296 TFLOPs in FP16, FP32, and FP64 performance respectively; allowing for a pixel rate of 288 GP/s and texture rate of 648 GT/s.
Conversely, the GeForce RTX 3080 comes equipped with 8,704 shading units (or CUDA Cores), 272 TMUs, 96 ROPs, 68 RT cores, and 272 Tensor cores. This gives it a theoretical performance of 29.77 TFLOPs (FP16 and FP32) and 0.4651 TFLOPs (FP64). It has a pixel rate of 164.2 GP/s, and texture rate of 465.1 GT/s.
Though the Radeon RX 6800 XT does lack heavily in regards to its available shading units, it has more TMUs, ROPs, Ray Tracing cores, and a higher rendering power than its Nvidia counterpart – with the only exception being its FP32 performance. These specification advantages are also made apparent in the benchmarks comparisons released by AMD.
In these tests, the Radeon RX 6800 XT outperforms the GeForce RTX 3080 in every game tested at a 1440p resolution, and only trails (in framerate) at 4K in four out of the ten titles that AMD benchmarked: The Division 2, Doom Eternal, Resident Evil 3, and Shadow of the Tomb Raider – all differences being very slim. Where the Radeon RX 6800 XT outshines its competition even further, is in the FPS/$ studies. In 20 such analyses that AMD calculated, the Radeon RX 6800 XT comes out on top in 18; trailing to the Radeon RX 6800 for Battlefield V at 1440p, and the GeForce RTX 3080 for Division 2 at 4K. Given that the Radeon RX 6800 XT outperforms Nvidia’s $1,500 flagship GeForce RTX 3090 (in both 4K and 1440p resolutions) in Battlefield V, Borderlands 3, Forza Horizon, and Gears V, the fact that it is so dominant in the FPS/$ category comes as no surprise.
Power & Efficiency
If AMD’s benchmarks are to be trusted, and the Radeon RX 6800 XT does actually outperform the GeForce RTX 3080 by the margin displayed, then the fact that the Radeon RX 6800 XT actually has a less power demand than the RTX 3080 really is quite remarkable. Of course, the 20-Watt differential may not seem like much, but the less power required by the GPU, the less heat it will have to dissipate.
The fact that AMD’s Zen 3 Ryzen 5000 Series CPUs are also more energy efficient than their Intel competition, thanks in part to their 7 nm processing node, really makes the AMD CPU plus GPU combination one that is incredibly power efficient.
Prior to judging on the superiority of either GPU, it is best to wait for a wider range of benchmarks that can prove what the marketing team of AMD claims.Lest we forget, Nvidia’s bold statement that the GeForce RTX 3090 would be the first 8K gaming GPU – on that was utterly debunked by benchmarks ran by Gamer’s Nexus, who displayed the GeForce RTX 3090’s inability to handle the task, even at 30 FPS.
If, however, the benchmarks in this case are true, it does appear that the Radeon RX 6800 XT is the superior GPU for its price. It has 60% more VRAM storage, better overall performance, better energy efficiency, and it sells for 50 dollars cheaper. Of course, when speaking on performance, it must be stated that in the aforementioned benchmarks, the Radeon RX 6800 XT was boosted by AMD’s Smart Access Memory feature. This means that these referenced benchmarks only hold true for builds that incorporate a Ryzen 5000 Series CPU. If you do not plan on building a new PC, and would rather just upgrade your GPU and keep the CPU you are already using, then there is an argument to be made for which GPU purchase is a smarter choice.
Without Smart Access Memory, AMD’s benchmarks (4K resolution) from their RDNA 2 announcement, show the GeForce RTX 3080 leading in five of the ten gaming titles they tested. However, even with even performance across the two GPUs, the Radeon RX 6800 XT does cost 50 less, while having more VRAM storage, and technologies that game developers will surely invest the time to optimize in order to reach the maximum graphical potential of their games.
All in all, it is always best to hold back any excitement for the release of new hardware, at least until their performance is proven by more than one sources. For now, it does seem that AMD holds the best price to performance-optimized GPU in the market: The Radeon RX 6800 XT.