Finally, the last RTX 40 Super series graphics card is here, focusing on the 4K high-brush gaming experience, and the initial price is 1499 yuan lower than the RTX 4080, which is 8099 yuan. If you've recently had an upgrade or installation, you need a budget and want to balance everything in one step. So today's NVIDIA GeForce RTX 4080 Super review will definitely help you.
Before the evaluation, let's introduce the test platform as usual, and the specific configuration is as follows:
This time, all models of the public version card have been changed to black livery, which can be said to be the biggest change in appearance. It is precisely because of this difference in detail that the new version of the NVIDIA GeForce RTX 4080 Super looks even more handsome.
Unboxing the public version of the card itself is also an exciting thing, the packaging is very exquisite, and when it is placed on the desktop like the picture above, it is not an exaggeration to say that it is an "ornament" full of metal technology.
The front of the NVIDIA GeForce RTX 4080 Super still has an X-shaped design, and there is a large fan that takes up almost half of the area, and the RTX 4080 Super model text on the left side is engraved in the shade and in the shade respectively, which is quite exquisite. However, the size has become a 35 slot card, so the volume is relatively large, and the weight has increased a lot, even when installing the machine, it is still recommended that you put on the graphics card bracket.
The heat sink fins and fan on the right side are all blackened, which is very weighty to hold in your hand and is also conducive to more efficient heat dissipation.
Similarly, there is an identical fan on the left side of the back, which creates a vertical air duct for better removing heat from inside the graphics card.
The side of the NVIDIA GeForce RTX 4080 Super graphics card is also blackened, which is visually very uniform, and a little concave arc is slightly done in the middle to enhance the texture. The power supply interface adopts 12+4pin design, which makes the power supply more stable.
The interface bezel is dark gray and equipped with 3*DP14A interface and 1*HDMI21 interface, up to 8K60Hz output.
In terms of accessories, the NVIDIA GeForce RTX 4080 Super is equipped with a 16pin to 3*8pin adapter cable, which is more compatible with power supplies.
The GeForce RTX 4080 Super is also built on the AD103 core, but the core specs are pretty scary. For example, it is equipped with 10,240 CUDA cores, which is 5% higher than the RTX 4080's 9,728 CUDA cores2%。It also boasts 52 Shader Core TFLOPS units and 121 RT Core TFLOPs, and the graphics gaming performance has been upgraded again. In addition, its 320 Tensor Cores and 836 AI Tops will provide powerful AI computing power and DLSS performance.
In terms of memory specifications, the GeForce RTX 4080 Super still has 16GB GDDR6X, which is the same as the RTX 4080. However, 64MB of L2 cache allows the GeForce RTX 4080 Super to get the most out of its gaming performance at 4K. In terms of GPU frequency, the base clock is 2295MHz and the boost clock is 2550MHz, which is 40MHz and 80MHz higher than the RTX 4080, respectively. Judging from the graphics card specifications and official positioning, GeForce RTX 4080 Super is a high-end graphics card positioned for 4K gaming and powerful AI computing power.
Of course, the GeForce RTX 4080 Super is built on the latest NVIDIA ADA architecture, built on TSMC 4N NVIDIA custom technology, so it delivers up to 2x the performance-per-watt. Streaming multiprocessors deliver up to 2x the throughput of the previous generation. DLSS 3 of the third-generation RT Cores, which delivers up to 2x faster ray tracing performance, and Tensor Cores' new FP8 engine with up to 132 petaflops of Tensor processing performance with up to 4x faster DLSS.
Before testing performance, let's test the donut oven to see how it performs in terms of power consumption. After 20 minutes of baking, the GPU power consumption is kept at about 320W, the core temperature is about 69, and the video memory temperature is 70, which is still relatively balanced, and the GPU frequency at this time can come between 2325MHz-2450MHz, and the performance release is still guaranteed.
After the roast, we ran a 3D Mark Time Spy stress test on the GeForce RTX 4080 Super to see if the graphics card performance release was smooth and durable under continuous high loads, and the GeForce RTX 4080 Super came back with a score of 996% results, and the consistent performance output is outstanding.
Next up is the GPU Theory test, where the GeForce RTX 4080 Super scored 14,401 points in the 3DMark TimeSpy Extreme DX12 test, up 67 points from the RTX 3080's 8,602 pointsAround 4%, which is also better than the previous generation RTX 3090.
In the 3DMark Fire Strike Extreme DX11 test, the GeForce RTX 4080 Super got a graphics card score of 34,894, an improvement of about 71 over the RTX 3080's score of 20,3822%, compared to the previous generation RTX 3090 Ti's 27074 points, there is also 288% lift.
In the 3DMark Fire Strike Ultra DX11 test, the GeForce RTX 4080 Super achieved a graphics card score of 17,395, which is about 67 percent higher than the RTX 3080's score of 10,3658%, which is also ahead of the previous generation RTX 3090 Ti's 13855 points.
In the 3DMark Portal Royal ray tracing test, the GeForce RTX 4080 Super scored 18,439 points, a 64-point improvement over the RTX 3080's 11,231 points1%, the performance of ray tracing games has been significantly improved.
Speaking of gaming, we have to mention that the NVIDIA platform already has more than 500 games and applications that support RTX technology, and as DLSS 3 technology becomes more mature and popular, the number of games that support DLSS 3 is also rising.
For example, 14 new games with DLSS 3 technology have been added this year, including Dragon Dogma 2, Horizon Forbidden West, Dragon Among Us 8, Half-Life 2 RTX and other well-known games. In addition, Alan Wake 2, Cyberpunk 2077: Shadows of the Past, Call of Duty: Modern Warfare 3, etc., which have been released, also support DLSS 3 or DLSS 3 for the first time5 of the technology. At the same time, the current mainstream popular competitive games all support NVIDIA Reflex technology, which can greatly improve the gaming experience for both low-end and esports players.
In the next playtest session, we will test 6 games, including 2** competitive games, 2 traditional raster games, and 2 DLSS 3 support 35 ray-traced games, all games tested in 4k resolution to verify its performance in 4k 4k ray-traced games.
First up are two ** competitive games, in CS2 we used the default High quality option, played a round at 4K, and recorded an average frame rate with Frame View, and the GeForce RTX 4080 Super was able to average 305 frames per second at 4K.
In addition, by turning on the NVIDIA Reflex option in the game's settings to On + Enhance, the average latency of the game is only 9ms, even at 4K, which can be said to have a very small impact on players.
The GeForce RTX 4080 Super has an average frame rate of 350 fps at 4K resolution and the highest image quality, which is also fully capable of meeting the needs of high-scoring and high-brush e-sports.
As for latency, Rainbow Six Siege also supports NVIDIA Reflex technology, and if you select the option with Vukan while the game is running, you will find the Reflex On + enhanced option in the game, and then grab it through Frame when it is turned on, and the average latency at 4K resolution is only 102ms, the latency performance is quite good.
Then there are 2 traditional raster games, in the classic narrative adventure game Red Dead Redemption 2, the highest image quality is selected, and the average frame rate can reach an average of 99 fps at 4K resolution, compared to 95 fps on the RTX 4080, which is an improvement of 42%。
We also tested Ubisoft's latest title, Assassin's Creed: Mirage, and ran the game's own performance benchmarks at 4K resolution, bringing the average frame rate to 100 fps on the GeForce RTX 4080 Super and 92 fps on the RTX 4080, an improvement of about 86%。From the test results, it is not difficult to see that in traditional raster games, the RTX 4080 Super has a relatively limited improvement over the RTX 4080.
So how does the GeForce RTX 4080 Super perform in ray tracing games? IT Home has chosen two models that support DLSS 3 35 ray tracing games, one of which is the very well optimized Forza Horizon 5, select the extreme quality preset in the game, and turn on DLSS 3 + DLSS 2 quality file, the NVIDIA Reflex option will also be turned on by default, we choose Reflex On+Boost as usual, and run the game's built-in performance benchmarks.
Entering the game, choose 4K resolution with extreme image quality, the average frame rate can reach 149 frames, and the average latency of the game is only 164ms。The RTX 4080 averages 132 frames, an improvement of about 128%。
So what if we only turn on ray tracing, don't turn on DLSS, and turn off Reflex Low Latency? You can see that in Forza Horizon 5, the average frame rate for 4K ray tracing still reaches 125 fps without DLSS, but without Reflex Low Latency technology, the average latency of the game soars to 241ms。
Finally, "Cyberpunk: 2077", which has very high requirements for graphics card performance, this time we tested it 3 times, the first time we chose super ray tracing image quality, turned on DLSS 2 quality file + DLSS 3 frame generation, and ran the built-in benchmark performance test at 4K resolution, at this time the average frame rate can reach 83 frames, which is not bad.
The second time, we chose ultra-fast ray tracing quality, and turned on DLSS 2 quality + DLSS 3 frame generation + DLSS 35 Ray reconstruction, at this time, the GeForce RTX 4080 Super averages exactly 60 fps at 4K resolution, which is considered to hold the bottom line of 4K 60 fps.
Of course, if you turn off DLSS completely and only turn on ray tracing, the average frame rate of a game in 4K will only be 29 fps, after all, not all games can have the optimization level of Forza 5, and the load on performance of image quality is also different. Through comparison, it is not difficult to find that DLSS technology is definitely the most practical technology for gamers.
Turn off light reconstruction.
Turn on Ray Reconstruction.
Then you can't turn on DLSS 35 What is the impact on the game's image quality? If you look at these two pictures, the reflection of the slippery road after rain in the same scene in the game gives us the answer, and the clarity of the reflection has improved significantly after turning on the light reconstruction, and there are many fewer artifacts that were common in the previous picture, which is obvious DLSS 35 It can be said to be the lubricant of high-grade ray tracing, which can also produce the best frame rate performance while retaining the highest picture effect.
As we mentioned earlier, the GeForce RTX 4080 Super also has 64MB of L2 cache, 23Gbps of memory bandwidth, and up to 836 AI Tops cores, which provides a strong guarantee for creative production and AI performance in GeForce RTX 4080 Super. Users can also install NVIDIA Studio drivers to accelerate more than 110 of the most popular creative apps, such as Optix, DLSS, and Maxine, as well as the full suite of NVIDIA Studio creative apps, including NVIDIA Omniverse, Broadcast, Canvas, and RTX Remix.
In modeling and rendering software, GeForce RTX 4080 Super's 10,240 CUDA cores provide hardware acceleration for increased productivity. Almost all modeling software optimizes NVIDIA's GPUs for efficiency, stability, and compatibility. For example, in V-Ray, a popular rendering tool, you can take advantage of RTX-accelerated ray tracing to achieve high-performance final frame rendering. In addition, GPUs with AI noise reduction can further accelerate interactive rendering for a smoother work experience.
Take a look at the GeForce RTX 4080 Super in terms of professional productivity. First up is the V-Ray Benchmark, a modeling software that scored 3103 points in the GPU CUDA sub-score.
In the GPU RTX benchmark category, the GeForce RTX 4080 Super got a final score of 4292 points.
In the benchmark test of another modeling software, Blender, the GeForce RTX 4080 Super graphics card scored 4574, 2162, 2264 in three scenes, which is also a significant improvement.
The GeForce RTX 4080 Super comes with 2 NVENC encoders and supports the latest 1 encoding technology, which provides faster encoding and higher quality streaming performance at the same size. The domestic Jianying Professional Edition has achieved acceleration support for the nvidia**1**encoder,We have also tested this,Import a 4K** footage with a duration of 2 minutes and 30 seconds in the Jianying Professional Edition,And then use H.264 encoding and **1 encoding for export.
As you can see, in the case of the same 4k 60 frame specification, the size of the 1 encoded export is 609MB, hThe size of the 264 encoded export is 800MB, and the file size is encoded with **1 compared to hThe ** of the 264 code is nearly a third smaller.
And in terms of image quality,By intercepting the details of the picture of the two** for observation,There is basically no difference between the two encoding methods,Small and high-definition,**1 is bound to become the new favorite in the encoding format。
Then there's RTX Super Resolution technology, which can upscale up to 4K resolution from 1080p to 4K resolution through the GPU's AI calculations, achieving algorithmic clarity. We just have to find the Adjust image settings in the NVIDIA control panel to customize the adjustments, with 4 levels of Superscore.
From left to right: 480p, vsr1, vsr2, vsr3, vsr4
We passed**posted a post on the IT home station B account**, and adjusted the resolution to 480p, and then successively increased the superscore level for screenshot testing, and we can see that the characters under 1-4 files have become clearer.
Finally, we also briefly tested the performance of this card in AIGC with Stable Diffusion, and we used SD XL Base 1 based on the SDXL open source model0_0.9vae.Safetensors, 21 images of girls on the snowy mountains were drawn in the form of Wensheng diagrams, the image resolution was 1024*1024, the sampling method was Euler A, the iteration number was 50, the total number of batches was 10, and it took 1 minute and 55 seconds to complete.
The second time, we turned on Tensorrt Engines and ran it again with all other settings being the same, and this time it only took 1 minute and 1 second, which was nearly half the time.
In addition, even in the case of relatively simple drawing parameter settings in the current evaluation, 8GB video memory usage is the norm, and the instantaneous video memory occupation will soar to more than 10GB 12GB, not to mention higher resolution and more complex parameter settings. If the video memory of this card is only 8GB, then for professional AI creators, it is likely to happen frequently, and once the video memory is not enough, the image generation time will be greatly extended, which will delay things very much. That's why the GeForce RTX 4080 Super's 16GB graphics play a very important role in the AI aspect of the AI aspect of the display, which is directly linked to the efficiency of AI content generation.
Compared with the GeForce RTX 4070 Super we reviewed before, this GeForce RTX 4080 Super is obviously aimed at users who are looking for the ultimate 4K gaming experience, that is, mainstream games can achieve 4K 100fps gaming performance with full-quality special effects, so players don't need to choose between game image quality and special effects, just enjoy the game. Users who are still using RTX 40 series old cards can buy it with their eyes closed if they want a one-step gaming experience. However, if you already have any RTX 40 graphics card, the upgrade cycle can be postponed. In addition, users who have AIGC creation or other AI needs can also consider this card, which is also the most suitable for AI-related work at present.
On the other hand, the public version of GeForce RTX 4080 Super has excellent workmanship and texture, as well as a unique form factor, which is also one of the reasons why many gamers are sought after. Not to mention that this debut** is a full 1400 yuan cheaper than the original RTX 4080. Interestingly, if we compare the previous generation RTX 3080Ti, then you will find that the number of CUDA cores, Tensors Cores, and RT Cores, both of them, are the same, but because of the ADA Lovelace architecture, the performance is much stronger in all aspects, and the power consumption is even 30W lower.