GeForce RTX 5080 will be one of the graphics cards stronger next generation NVIDIA and will only be under GeForce RTX 5090. It will be intended for play in 4K resolution, and push today’s most demanding games in maximum quality while maintaining a good level of fluidity. Performance and price will be a step higher than the GeForce RTX 5070.

So far, no information has been leaked to suggest that NVIDIA might use a different architecture in its GeForce RTX 50, so at the moment everything points to this new generation will be based on Blackwell. According to NVIDIA, this architecture improves FP64 performance by 30% compared to Hopper, is 30 times faster in science and simulation tasks, and offers significant power improvements compared to the previous generation.

All these data focus on performance values ​​oriented towards the professional sector and it is not possible to extrapolate them directly on the GeForce RTX 50, but I can share with you a general estimate of what we can expect from the GeForce RTX 5080 based on all the information I have found so far and my knowledge and experience in the technology sector.

Possible specifications of GeForce RTX 5080

It will be a high-end graphics card that will occupy one of the most important positions in its range, as it will be aimed at those who want to enjoy high performance but cannot afford to pay the price of the GeForce RTX 5090. It will use core GB203, probably in a slightly trimmed version to make room for the future GeForce RTX 5080 SUPER.

• Graphics core GB203 made on TSMC’s 4nm node.
• 110 SM units.
• 14,080 shaders at speeds between 2.8 GHz and 3.2 GHz.
• 444 texturing units.
• 128 raster units.
• 110 fourth generation RT cores.
• 444 fifth generation tensor cores.
• 256-bit bus.
• 16 GB or 24 GB of GDDR7 memory at 28 Gb/s (672 GB/s bandwidth).
• 64 MB of L2 cache.

If we compare it with the GeForce RTX 4080, we see important improvements, among which the increase in the number of shaders stands out, which it goes from 9,728 to 14,080and the texturing and rasterization units as well as the tensor core and RT core are also increased. The L2 cache would remain at 64 MB, which is enough for the GeForce RTX 5080 to run comfortably at 4K resolution.

Please note that NVIDIA may use cropped version GB203 cores, so it’s possible that the GeForce RTX 5080 has a lower shader count, although even in the worst case I don’t think it will go below 12,000, as that would be too far from GeForce performance. RTX 4090 and it might not be able to clearly surpass it, which would obviously be unbearable.

If we’re going to talk about consumption, I’m pretty sure NVIDIA will put this graphics card in place between 320 and 350 watts TGPreasonable values ​​for the level of performance we can expect, and very similar to what we saw with the GeForce RTX 4080. This would be good news for those considering an upgrade, as we would not need a very powerful resource.

Graphics core, architecture and possible raw performance

As I said at the beginning of the article, the GeForce RTX 5080 will almost certainly use the Blackwell architecture. Its basis will be the GB203 graphics core, which should adopt a monolithic core design with all GPU components. integrated into the same silicon chip, including L2 cache. This means lower latency compared to chiplet designs and better performance.

It is rumored that NVIDIA could use TSMC’s 3nm node, but the information does not agree and at this point we cannot rule out that the GeForce RTX 5080 will end up using a 4nm node. Both options are open, so keep that in mind. In any case, we can also expect the 4nm node significant improvement in efficiency.

Blackwell will bring new shaders with improved performance per clock, fourth generation RT cores with new features aa higher performance in key operations such as computing ray intersections and triangles, and will come with fifth-generation tensor cores that will also offer superior AI performance.

We can also expect the integration of 64 MB of L2 cache, which will help reduce the number of GPU calls to the graphics memory to load the data and things it needs, which significantly improves performance because GPU-L2 cache communication is much faster. Finally, we will have a of a new generation optical flow acceleratorwhich could significantly improve NVIDIA’s image generation technology, or even serve as a key component for the development of new technologies applied to games.

I know you’re dying for me to talk about the possible power of the GeForce RTX 5080, and I’m not going to keep you waiting. A configuration with 14,080 shaders and a peak frequency of 3,200 MHz would have maximum performance 90.11 TFLOPs in FP32. With this character it barely beats the 82.58 TFLOPs offered by the GeForce RTX 4090although we must bear in mind that the differences in the games could be much more striking.

Power source and what processor recommended for GeForce RTX 5080

The GeForce RTX 5080 will be more powerful than the GeForce RTX 4090, although the difference may not be huge. My intuition tells me that the former will be 10% to 18% more powerful than the latter, so the processor that offers optimal results with the GeForce RTX 4090 It will do the same with the GeForce RTX 5080.

The minimum recommended level for GeForce RTX 5080 movement will be at a Ryzen 5 7600X or Intel Core i5-13600Kand the ideal level will be Ryzen 7 7800X3D or Intel Core i7-13700K (or KF). Keep in mind that this recommendation changes depending on the resolution used, because if we play in 4K, the differences between the mentioned processors will be smaller, while if we play in 1440p or lower, the differences will be much more pronounced.

However, be clear that it is not worth buying a GeForce RTX 5080 for gaming in 1080p, because at this resolution, even with the Ryzen 7 7800X3D, we will have a very significant bottleneck in most current and future games. This graphics card will only make sense at 1440p and 2160p.

The required power supply will be practically identical to the recommended power supply for the GeForce RTX 4080. This means that with quality power supply with a power of 750 watts We won’t have problems if we have, for example, Ryzen 5 7600X or Ryzen 7 7800X3D. However, if we have an Intel Core i7-13700K or an Intel Core i9-13900K, we should install an 850W power supply, since both processors have a higher consumption.

Release date and possible price

GeForce RTX 5080 will be launched at the end of this year. The introduction of this will take place together with the GeForce RTX 5090, probably in the month of September, and both are said to arrive just a few weeks later. In theory, the GeForce RTX 5080 is likely to arrive first in the month of Octoberand shortly after, its older sister GeForce RTX 5090 will arrive.

Its retail price will be similar to, or slightly higher than, the GeForce RTX 4080 SUPER at launch. With that in mind, I think the GeForce RTX 5080 is most likely to hit the market price between 1,100 and 1,200 euros. Keep in mind that it may end up being priced higher due to the cost increases that will come with both the jump to the 4nm node and the use of GDDR7 memory.

AI generated images.