Several sources have leaked a good dose of information about the GeForce RTX 5080, and this has led me to do a complete review of the special article that I published at the time about this graphics card so that you have all updated information.
The positioning of the GeForce RTX 5080 has not changed, it will be a high-end graphics card that will be placed below the GeForce RTX 5090, but we have an interesting new detail that comes from Chinese sources, and that is that NVIDIA will launch two versions of this, one with 16 GB of graphic memory and another with 24 GB of graphic memory.
The 16 GB model will be the original model and the first to hit the market, and the 24 GB model will arrive later, probably as an intergenerational renewal accompanied by the tagline. «Ti» or «SUPER».
Imagen: Chiphell.
GeForce RTX 5080 specifications
The latest information have drastically reduced the total shader count of this graphics card, and has taken it to a level that would place it just a little above the GeForce RTX 4080. However, thanks to the improvements that the Blackell architecture will introduce, and the significant increase that we expect In the clock frequencies, the difference between the two in terms of performance is going to be quite large.
- GB203 graphics core manufactured on TSMC’s 4nm node.
- 84 SM units.
- 10,752 shaders at speeds between 2.8 GHz and 3.1 GHz.
- 336 units of textured.
- 112 raster units.
- 84 fourth generation RT cores.
- 336 fifth generation tensor cores.
- Bus de 256 bits.
- 16 GB of GDDR7 memory at 28 Gbps (896 GB/s).
- 64 MB L2 cache.
- 400 watt TGP.
GeForce RTX 5080 frente a GeForce RTX 4080
- AD103 GPU manufactured in TSMC’s 5nm process.
- 76 SM units.
- 9,728 shaders at 2,205 MHz-2,505 MHz, normal and turbo mode.
- 304 units of textured.
- 112 rasterized units.
- 304 fourth generation tensor cores.
- 76 third generation RT cores.
- Bus de 256 bits.
- 16 GB of GDDR6X memory at 22.4 Gbps (716.8 GB/s).
- 320 watt TGP.
With these specifications, the GeForce RTX 5080 would have only a small increase in the total number of shaders compared to the GeForce RTX 4080, since it has 9,728 shaders. The difference between the two would be 1,024 shaders, and as we can see when comparing the specifications of both, there would also be an increase in the rest of the key components of the GPU, such as the texture and raster units, the tensor cores and the RT cores.
GeForce RTX 5080
PG144/147-SKU45
GB203-400-A1
10752FP32
256-bit GDDR7 16G
400W— kopite7kimi (@kopite7kimi) September 26, 2024
The increase in working frequencies would be spectacular, since we are talking about up to 600 MHz more in turbo modeand the bandwidth would also increase thanks to the use of GDDR7 memory, although this amount will remain the same as the previous generation, 16 GB, a figure that is still more than enough to play with all the guarantees in 4K.
The GeForce RTX 5080 is much more powerful, but also will have a higher consumption. The latest rumors speak of 400 watts of TGP, if this is confirmed the difference with the current model would be 80 watts, a reasonable increase taking into account the differences that there will be between both graphics cards at the power level.
New model with 24 GB of GDDR7
As I said at the beginning of the article, Chinese sources assure that NVIDIA is going to launch a second model of the GeForce RTX 5080 that will have 24 GB of graphics memory. This will be an intergenerational renewal as was the GeForce RTX 4080 SUPERand therefore should also come with a more powerful GPU.
This means that it should have a greater number of active SM units, which would translate into more shaders, texturing and raster units, tensor cores and RT cores. No specific details have been leaked yet, but this second model exceeds 12,000 shaders and has a 384-bit memory bus.
The difference in terms of power compared to the 16 GB model It will depend on the improvement it brings at the GPU levelmore than the increase in the amount of graphics memory, because as I said, currently having more than 16 GB of graphics memory does not make a difference in performance in games, and it will not do so in the short or medium term.
Possible performance and raw power
If the GeForce RTX 5080 arrives with 10,752 shaders at a maximum of 3 GHz, its power will be 64,51 TFLOPs en FP32, a figure that would exceed 48,74 TFLOPs which reaches the GeForce RTX 4080, but would be behind the GeForce RTX 4090, which has a raw power of 82.58 TFLOPs in FP32.
Although this comparison in numbers points to the opposite, the GeForce RTX 5080 would be around 10% more powerful than the GeForce RTX 4090 in games, thanks to the different improvements it will bring at the architectural level. It could also have support for new technologies within the DLSS ecosystem, but this has not yet been confirmed.
The performance difference in ray tracing could be even higher, thanks to the possible improvements that the Blackwell architecture will introduce in the fourth generation RT cores. We will have to wait for NVIDIA to officially present this architecture to clear up any doubts.
If we talk about the model with 24 GB of graphics memory, the performance jump would be greater. We do not have an exact shader count, but we are going to use a realistic estimate to achieve an approximate performance value that is not crazy, such as 12,288 shaders, which is equivalent to 96 active SM units.
A GeForce RTX 5080 SUPER or Ti with 12,288 shaders at 3 GHz would have a raw power of 73.72 TFLOPs at FP32. It would still be insufficient to surpass the GeForce RTX 4090 in raw power, but the former would be more powerful in games. In this case I think that that hypothetical GeForce RTX 5080 SUPER or Ti it would beat the GeForce RTX 4090 in games by 15% or 18% on average.
Power supply, consumption and wiring
If the TGP of 400 watts is confirmed, this would be normalized consumption of the GeForce RTX 5080 working with 100% load. In situations where the GPU load is lower, because the game is not very demanding or because we have NVIDIA DLSS activated and these technologies reduce the GPU load, consumption will be lower.
In order to use a GeForce RTX 5080 with all the guarantees we will need a 750 watt power supply, and ideally it will have a 16-pin power connector so you don’t have to use adapters. With this we will also make the assembly much more aesthetic.
In most cases, this power supply will be enough so that we do not have any type of problem in terms of stability and performance. However, it is very important that you keep in mind that everything will depend on the consumption of the rest of the components of your PC.
If you use processors like Ryzen 7 and below and Core i5 and below You won’t have a problem, but starting with Ryzen 9 and Core i7, CPU consumption may increase, and in these cases it might be advisable to have a slightly more powerful power supply. In these cases, with an 850 watt power supply we will not have problems.
Recommended configuration for a GeForce RTX 5080
For the GeForce RTX 5080 to develop its full potential we will need a powerful PC, although we will not necessarily have to reach a high-end configuration in every sense, which would be recommended with a GeForce RTX 5090.
Starting with the processor, the ideal would be not to go below un Ryzen 5 7600, un Intel Core i5-12600K o un Ryzen 7 57000X3D. Having 32 GB of RAM and a PCIe NVMe SSD capable of reaching at least 3.5 GB/s sequential read speed are two other important recommendations to balance a PC with this graphics card.
We have already seen what power supply we would need, and we now have to talk about it.the recommended resolution. The GeForce RTX 5080 is going to be a graphics card designed to achieve a perfect experience in games under 1440p with ray tracing activated and maximum qualities, and also to play without problems in 4K, so those would be the two ideal target resolutions for this card graph.
With this I do not mean that we will not be able to use it in 1080p, but in said resolution we could find cases in which it will be so sufficient that in the end could suffer from CPU bottleneck. This is something we’ve already seen with the GeForce RTX 4090, and the GeForce RTX 5080 is going to be more powerful than it is, so the story tells itself.
Release date and price
It seems that little by little a certain consensus is being reached between different sources, and that the presentation of the GeForce RTX 5080 could occur in the CES 2025. A presentation does not equate to a release date, so several weeks could pass between that presentation and the actual availability date of said graphics card.
The rumors that NVIDIA could present the GeForce RTX 5080 before the end of the year have not been heard again, so right now has more credibility that possible presentation in 2025. However, be clear that nothing is confirmed yet.
The price of the GeForce RTX 5080 could be similar to that of the GeForce RTX 4080 at launch, but it could also be slightly lower or slightly higher. NVIDIA has been fine-tuning the price of its new graphics cards with considerable increases after each new generationa strategy that could be repeated with the GeForce RTX 50 series.
I think that, at best, the GeForce RTX 5080 could cost 999,99 euros, and that in the worst case could end up around the 1.299,99 euros. As a reference, I remind you that the GeForce RTX 4080 hit the market with a price of 1,469 euros, but the GeForce RTX 4080 SUPER was much more affordable, and was launched with a price of 1,119 euros.
Source: www.muycomputer.com
