"The biggest generational leap ever", so began the CEO of NVIDIA when from his kitchen he started talking about the new GeForce RTX 30 series GPUs and, rattling off only the technical characteristics of the graphics cards, it is really difficult to blame him. The GPUs announced about ten days ago integrate the best you can currently find in terms of hardware but also software optimization and artificial intelligence. All this with an extremely attractive list price for the Founder's Edition versions.
NVIDIA is said to have literally killed the market; certainly the second-hand one, which has seen an incredible drop in the valuations of the old 20 Series. Try today to sell an "old" 2080 and you will see the absurd price proposals you will receive; but this is another story, what we want to do today is to verify and quantify this generational leap. For this reason we took the new RTX 3080 and compared it with our 2080 Ti, both in Founder's version. What came out of it? Let's find out together.
- ANATOMY OF THE BOARD
- RTX 3080 VS RTX 2080 TI ON PAPER
- PERFORMANCE: WHAT A JUMP!
- TEMPERATURES AND CONSUMPTIONS
- BROADCAST, REFLEX AND STUDIO
- FINAL CONSIDERATIONS
Before going to analyze the technologies that are inside the new NVIDIA GeForce RTX 3080 and to compare it with the older, but still very valid, RTX 2080 Ti, I want to talk a little bit about how the card looks aesthetically and what we find inside the sales package.
The packaging of this new RTX 3080 is extremely sober, as we are used to seeing when we are dealing with Founders version GPUs. Lifting the lid we immediately find our object of desire, imposing in its black and gunmetal gray livery. WONDERFUL! There are no other adjectives to describe it. Mine is a purely subjective opinion, of course, but it certainly cannot be denied that the design of these cards is incredibly accurate and also, as far as possible, absolutely clean and sober.
The main novelty in the construction is given by the unibody skeleton with an X-shaped central joint that acts as a housing for what we could define as the actual body of the board. The latter, except for the fans and for the plate on which the PCB is placed, is in effect a single large lamellar heatsink useful for transporting as much heat as possible outside the board.
The two fans are positioned to create two flows of air that enter and exit parallel from the two sides of the board in what is referred to as the new Dual-Axial Flow Through system. The result should be that of greater efficiency in terms of dissipation and, consequently, the achievement of temperatures in any case always within the norm. On paper, the new ventilation system can manage up to about 90W more than the previous solution.
In terms of dimensions, it does not differ much from the RTX 2080 Ti. The new 3080 remains a two-slot solution and is only a few millimeters longer when compared to the older model. The choice of adopting a new 12-pin connector is certainly particular, much more compact than the previous 2×8. At the moment we do not yet find on the market PSUs that use this connector but in the package there is an adapter that allows you to split the 12 Pin into two classic 8 connectors.
We close this description by talking about connectivity. The short outer side of this RTX 3080 houses an HDMI 2.1 output and three Display Port 1.4a. As already highlighted in the news of the official launch, thanks to the HDMI 2.1 and the support for the Codec AV1 these NVIDIA 30 series cards allow you to manage the 8K HDR resolution with a single cable, taking advantage of the greater bandwith and the optimization provided by the Codec.
Considering the GeForce RTX 3090 in another market segment – both because of NVIDIA's claims that it is the successor of the TITAN and because of the price – GeForce RTX 3080 presents itself on the consumer market as the card to beat in this 2020. numbers provided by NVIDIA at the presentation stage leave little room for doubt: double performance compared to GeForce RTX 2080 and up to 30-40% higher than GeForce RTX 2080 Ti. It is right to remember this aspect because, at least on paper, the most correct comparison might seem that between RTX 3080 and RTX 2080 Super.
Our choice is therefore the result of the data declared by the black / green company and of the fact that NVIDIA with the Ampere architecture has managed to drastically optimize the processing units of the GPU, at the same time improving efficiency (up to 1 , 9 times), obviously helped by the new production process. All this basically leads to a leap in performance such that this comparison still risks being unequal and in clear favor of the new 3080. Without forgetting that the RTX 2080 Ti remains the top of the range of the 20 series and RTX 3080 currently occupies the same position if let's look at series 30.
The NVIDIA GeForce RTX 30 series and, in this case, our GeForce RTX 3080 represent NVIDIA's second generation RTX GPU. At the base of these chips, built with Samsung 8nm production process (custom design), there is the new Ampere architecture, able to offer, from specific sheet, up to double the performance compared to the previous Turing.
When NVIDIA talks about doubled performance, it doesn't go very far from reality, quite the opposite. Ampere GPUs feature a completely revised Streaming MultiProcessor (SM) unit that is capable of delivering twice the throughput of Turing in FP32 operations (128 vs 64 operations per clock). But not only that, we must also take into account the optimization of the first level Cache, which on Ampere increases by 33%, doubling the available bandwidth (over 210GB / s).
The architectural innovations of the Ampere GPUs also concern the RT Core (2nd gen) and Tensor Core (3rd gen) units, both twice as fast as the previous generation ones, present on the Turing GPUs and therefore on the GeForce RTX 20. For these reasons , related to the excellent technical specifications, the GeForce RTX 30 should finally allow you to enjoy the best of ray-tracing, while maintaining high-level performance and good energy efficiency. NVIDIA itself defined these as the first cards capable of delivering good 4K playability with active RTX.
The innovations introduced with the new 30 series do not end there and also in terms of design and construction we have important improvements compared to the previous generation. Starting from the new Dual-Axial dissipation system, Flow Through Thermal Solution which allows to reduce the noise in the face of a dissipation capacity practically doubled. Also of note is the newly created RTX I / O technology that we examined in our in-depth analysis.
But let's come to our comparison. On the one hand, GeForce RTX 3080 which uses an Ampere GA102 (GA102-200) GPU, 8704 Cuda Core chip flanked by 10GB of 19 Gbps GDDR6X memory on 320-bit BUS. On the other, GeForce RTX 2080 Ti, the undisputed queen of the graphics card market until a few weeks ago.
The GPU in this case is a Turing TU102 (TU102-300A), a chip built with TSMC 12nm technology and equipped with 4352 Cuda Core. The amount of memory in this case is in favor of the RTX 2080 Ti which mounts 11GB GDDR6 14Gbps, much slower than the GDDR6X but at the same time equipped with a 352-bit BUS.
And here is also a table that summarizes the technical characteristics:
Two quick premises before moving on to the test charts. In choosing the games to test we wanted to mix recent titles that support RTX and DLSS and games without these technologies but still very popular. All of these were done in 1080p, 1440p, 4K, and where possible 4K with active DLSS and RTX.
- Motherboard: ASUS ROG STRIX Z490 A-Gaming
- CPU: Intel Core i7 10700K
- RAM: 2x8GB DDR4 Corsair Vegeance Pro at 3200MHz
- Storage: 1x Corsair MP510 512GB (O.S.) + 1x Crucial P1 1TB (Data)
- Cooling: NZXT Kraken x53
- PSU: NZXT C850 (80+ GOLD)
- Test Bench: DimasTech Easy V3
CALL OF DUTY MODERN WARFARE: MULTIPLAYER (Shoot House)
The first title to end up under the clutches, or perhaps better to say the fans, of the new RTX 3080 was the well-known shooter from Activision in its competitive version. We decided to test the multiplayer game with the Shoot House map in order to have a scenario as similar as possible between the different tests. The game was set with all picture quality settings at maximum and ran in FullHD, 1440p, 4K and 4K with RTX active. Here are the results.
From CoD to Control, one of the notoriously heaviest titles of the moment. Here, too, the choice was to test all resolutions at the highest quality with regard to graphic details. In addition to standard rasterization, we also performed the test in 4K with RTX and DLSS active, the latter in Balanced mode. In fact, the Quality and Performance settings are also available which render respectively at a higher and lower resolution than the mode we use.
One of the most talked about titles of the year could not be missing, the adventure of the "courier" by Hideo Kojima who in the most excited and animated moments manages to put even the most powerful hardware in difficulty. The scenarios are immense and the work of the GPU is by no means simple. As above we tested the three classic resolutions with 1: 1 rendering and then 4K with DLSS active in its most demanding mode, that is, the one that prefers quality over performance.
RED DEAD REDEMPTION 2
One of my favorite games ever, an action-adventure definitely challenging from a graphical point of view. In this specific case, the game offers an internal benchmark that well simulates the use with different scenes in intensity and complications. The decision was therefore to use the latter to analyze the performance of the two cards.
The DOOM sequel is undoubtedly one of the most popular titles of recent years. Compared to the first version of 1993, the steps forward from the graphic point of view are certainly notable but it is not one of the most demanding games on the market. Even with low-end video cards you can play at very high framerates without too many problems and with the most recent solutions it could be one of the first games with which it will be possible to take advantage of the new FHD 360Hz monitors even at maximum detail.
In 4K with all the details at the highest level, the new game from Square Enix is a real pleasure for the eyes. The gameplay is often broken up by scripted sequences which, however, do not make the work of the GPU easier. In this case, RTX and DLSS are not available and therefore the game was only run in the three standard resolutions: 1080p, 1440p and 4K.
And let's close the circle with one of the most anticipated news when we talk about RTX or Fortnite. In fact, during the launch event of the new NVIDIA graphics cards, the announcement of the RTX version of the very popular multiplayer title by Epic Games arrived. We had the opportunity to preview a special map created precisely to be able to get a taste of what the quality of the images will be once the final version of the game is released. A map that turned out to be much heavier than expected and that in 4K with RTX active seriously put both cards in crisis.
Summing up, it is easy to observe how expectations have been substantially met. In practically every situation there is a significant increase in performance: we are often close to 30% of generation over generation gain, but in some cases we even get to have an increase in the order of 50%. I am absolutely satisfied with this card and with the work done by NVIDIA, at least as regards the performance which is absolutely in line with when declared by the company. Will the same also apply to operating temperatures and consumption? I would say yes, but let's see it together.
In addition to the above tests, we analyzed the behavior of the card even under stress by first using the Furmark software to load the GPU to 100% and then the 3D Mark Time Spy benchmark in the Extreme version to simulate a 4K gaming use. Below you will find the graphs relating to the trend of frequencies, temperatures and consumption on the two cards under analysis.
View in Full Screen
I used the Furmark test to bring the GPU to full load and then to observe how much it consumes to its maximum potential. The test found a maximum value of 320W for the RTX 3080 and 260W for the 2080Ti. Considering the CPU, memories and motherboard I would say that the 750W PSU recommended by NVIDIA is almost the bare minimum; if you have the possibility, aim for an 850 with good efficiency.
View in Full Screen
3D Mark Timespy Extreme is instead the test used to observe the behavior of temperatures. As you can see from the graph, the trend of the latter for the two cards under analysis is very similar and it is essentially the same trend that we have also detected in the real game tests.
With very long sessions and high load, 70 degrees are always exceeded but it is difficult to go beyond 80. This happens on one board as on the other, reflecting the excellent work done by the black-green company in terms of optimizing the 'hardware and improvement of the dissipation system. Basically we have the same temperature but with a card that can easily handle 50W more.
Ok brute power and FPS but the new NVIDIA cards come correlated by a series of tools and technologies that allow the user to make the most of them in different areas. This will not be the place where we will analyze all these other factors, we will probably dedicate some insights to them separately, but it is still right to mention them and explain what they are used for to better understand the work behind a launch of this type. .
Let's start with NVIDIA Broadcast, a new software that accompanies the RTX cards, not only the 30 series, which aims to expand the functions and improve video conferencing and streaming. An optimization software therefore, which is able to improve the audio and video streams of webcams and microphones to bring the transmission experience as close as possible to that of a professional studio. The system intervenes with tools such as noise cancellation or the application of a virtual background to improve video calls or content streaming.
Added to this is Reflex which, more than a software, is a suite of technologies that measures and optimizes the latency of the system to allow the user to obtain a much more reactive gaming experience. There are two fundamental components that will allow you to take advantage of Reflex at 100%: the first element is NVIDIA Reflex Low Latency Mode, a tool that can be activated in the settings of some games that is able, by itself, to reduce latency by up to 50%; to get the maximum experience it will then be necessary to combine the NVIDIA Latency Analyzer with the Low Latency Mode. Inserted inside the new generation monitors, this technology measures the time elapsed between the click and the pixel change and optimizes the reaction speed to further lower this delay.
Latency Analizer will be incorporated into the new 360 Hz gaming monitors equipped with NVIDIA G-Sync, the first of which will arrive in the spring, and will be supported by major gaming peripherals. The technology will arrive on the GameReady drivers in September; the first compatible titles will be, among others, Valorant and Destiny.
And finally, last but not least, the NVIDA Studio program, to which we dedicated an in-depth study some time ago. As well as the previous generation cards, the new Series 30 are compatible with the NVIDIA Studio program and with all the elements that compose it; from dedicated drivers to highly optimized software. The aim is obviously to make the production process smoother and faster by reducing the time gap between idea and realization as much as possible.
And here we are at the moment to draw conclusions and as always I start from the price. The new NVIDIA GeForce RTX 3080 in the Founder's Edition version we tested costs 719 euros which is a significantly lower price than the over 1200 euros that characterized the launch of the 2080 Ti. A price certainly aggressive and that, due to the excellent performance of the new card, puts even those who have recently purchased a top-of-the-range RTX 20 Series in serious difficulty.
Jensen Huang was probably right, we are witnessing the biggest generational leap ever and the gutted numbers here are the tangible testimony. Better performance up to over 50% and temperatures in line with the previous generation even in the face of much higher consumption. In short, this RTX 3080 is a real power monster, a splinter even compared to a card that, technically, would belong to a higher range like the 2080 Ti.
To renew or not to renew? Some would say that this is the problem but in reality, if you have the possibility from an economic point of view, there is no need to even think about it here: the answer is absolutely yes. With these cards, in fact, "maxed" 4K 60 fps is now the standard and 1440p at 144Hz is within the reach of almost everyone.
If we consider the mainstream range of the market (the 3090 is considered by NVIDIA itself to be an enthusiast card) we are certainly facing the best graphics card produced to date and, perhaps for the first time, the price is by no means crazy. Did you get watering? Yes to me!