The GeForce RTX 5060 Ti will be a “premium” mid-range graphics card, a very important position as it will target fill the gap that will remain between GeForce RTX 5060 and GeForce RTX 5070, i.e. between mid-range and high-range entry. The GeForce RTX 4060 Ti is one of the best-selling graphics cards of this generation, so you can get an idea of ​​the importance of this new model.

Those who can’t or don’t want to pay the price of a GeForce RTX 5070, but also don’t want to buy a GeForce RTX 5060 because this one is too small for them, they will be able to turn to the GeForce RTX 5060 Ti. We still don’t have specific information about the specs that this graphics card will have, but I know many of you are looking forward to it, so I want to share with you estimate What I did with this new model.

Not only will I discuss with you its possible specifications, always keeping in mind that as I said, this is a realistic estimate, but we will also see comparative power and raw power dataas well as possible consumption values.

GeForce RTX 5060 Ti specifications

AI generated image

There are two big options that this graphics card uses core GB206the same one that will theoretically mount the GeForce RTX 5060, or that it uses core GB204which is the one that will power the GeForce RTX 5070. Both options make sense as you could use the GB206 in the full version with more shaders or the GB204 in the trimmed down version to differentiate yourself from the GeForce RTX 5070.

Seeing the results that NVIDIA has achieved with the GeForce RTX 4060 Ti, I am inclined to think that it is most likely that they will use shortened version of GB204, maintaining an L2 cache and a 192-bit memory bus. This would be ideal, but there could also be a reduction of both elements and the use of a 128-bit bus. With that in mind, here’s my ideal GeForce RTX 5060 Ti estimate:

• Graphics core GB204 made on TSMC’s 4nm node.
• 48 SM units.
• 6,144 shaders at speeds between 2.8 GHz and 3 GHz.
• 192 texturing units.
• 64 raster units.
• 48 cores of the fourth generation RT.
• 192 fifth generation tensor kernels.
• 192-bit bus.
• 12 GB of GDDR7 memory at 28 Gbps (672 GB/s bandwidth).
• 48 MB L2 cache.

GeForce RTX 5060 Ti vs GeForce RTX 4060 Ti

If NVIDIA uses this configuration, the GeForce RTX 5060 Ti will have it a much larger number of shaders than the GeForce RTX 4060 Ti, as the former would have 6,144 shaders and the latter has 4,352 shaders. Obviously, one will also have more texturing units, more rasterization units, and more RT kernels and tensor kernels.

Other important differences that would significantly affect performance is the memory bus, which would increase from 128 bits to 192 bitsand increasing the L2 cache from 32 MB to 48 MB. All of this would make the GeForce RTX 5060 Ti much more powerful than the GeForce RTX 4060 Ti, and it can run optimally at 1440p.

In terms of consumption, the GeForce RTX 5060 Ti will be more or less at the same TGP level as the GeForce RTX 4060 Ti. I don’t expect a significant increase except for the node change and architectural improvements that NVIDIA will introduce with Blackwell. For this reason, I’m inclined to think that this graphics card will have a TGP between 160 and 185 watts.

The GeForce RTX 4060 Ti is a graphics card that, due to the configuration of the memory bus and L2 cache, gives a maximum in 1080p, although it is true that it is capable of running games in 1440p. We would find ourselves face to face with the GeForce RTX 5060 Ti a much better 1440p ready modelif the specifications we have seen in this article are met.

Blackwell architecture: new tensor and RT cores

Blackwell Architecture will be an evolution of Ada Lovelace, who will keep the classic divisionand in shaders, tensor cores, and RT cores. The optical flow accelerator will also reappear and the L2 cache will have an important part of the space reserved at the silicon level.

We can expect a monolithic GPU-level core design, the use of TSMC’s 4nm node (some rumors talk about a 3nm node) and five big improvements:

• New generation shaders with excellent performance per clock cycle, i.e. at the same frequency.
• The fourth generation of RT cores, specialized in accelerating ray tracing and with higher raw performance in calculating ray intersections and triangles. They may also come with new specific optimizations.
• The fifth generation of tensor cores, specialized in AI and with a higher level of performance when working with matrices.
• A next-generation optical accelerator that could improve the performance and quality of image generation and even enable the introduction of new features in NVIDIA DLSS.
• Improvements to the L2 cache subsystem to improve speed and latency and use of GDDR7 memory to increase bandwidth.

The most interesting will be seen what new technologies NVIDIA is announcing to take advantage of this new architecture. With each new generation we have seen significant improvements in this regard, which is why I am convinced that the Blackwell It will not be an exception. All in all, I dare not reveal anything specific, because NVIDIA could surprise us with anything.

Raw performance and benchmark performance

AI generated image

A GeForce RTX 5060 Ti configured with 6,144 shaders and running at a maximum frequency of 3 GHz would have FP32 performance 36.86 TFLOPsthat would place her slightly above GeForce RTX 4070 SUPERwhich has power 35.48 TFLOPs. However, their gaming performance could be quite similar, because in the end raw performance data in single precision does not reflect actual gaming performance.

Compared to the previous generation, the GeForce RTX 5060 Ti It would be on par with the GeForce RTX 3090 at 1440p, a graphics card that has a performance of 35.58 TFLOPs in FP32. Overall, it should be considered that the GeForce RTX 3090 has more bandwidth and more graphics memory, so it should perform better in 4K games.

If we make a comparison with AMD graphics cards within the current generation the closest equivalent would be the Radeon RX 7900 GREwhich is a direct rival of the GeForce RTX 4070 SUPER and in the previous generation would easily have been placed at the level of the so-called Radeon RX 6950 XT. In terms of efficiency, it would clearly be better than all these equivalents.

Recommended resource and processor

Since we have a fairly plausible estimate of TGP as well as a reliable performance equivalence, it is easy to make recommendations both at the power level and at the processor level for this GeForce RTX 5060 Ti.

Starting with the power supply, we have a graphics card whose maximum TGP would be 180 watts, a very reasonable figure that would be within reach of even a cheap source. WITH 500 watt model We won’t have a problem with the minimum quality, although it will depend on the rest of the equipment components and above all on the processor consumption.

This is important because the Intel Core i9-13900K can consume more than 300 watts, while the Ryzen 5 7600X bordering on 90 watts at full load. The difference in consumption it’s giganticand that’s why I say that we have to take this into account when choosing a source.

As for the recommended processor to power the GeForce RTX 5060 Ti, we should have at least a Ryzen 5 5600and the ideal thing to get the most out of it would be to have a Ryzen 5 7600and Intel Core i5-12600K or equivalent model. The Ryzen 7 5800X3D would also fall into the recommended tier.

Release date and possible price

AI generated image.

The launch of the GeForce RTX 5060 Ti will happen at the same time as the GeForce RTX 5060, or at worst with a slight difference between the two. There is no specific date, but it is most likely that NVIDIA will launch this graphics card between April and May 2025, which means we’re still almost a year away. GeForce RTX 5070, GeForce RTX 5080 and GeForce RTX 5090 will arrive first.

Regarding the price, it is a topic that has been speculated a lot and the truth is that I still have a lot of doubts. I think the use of GDDR7 memory and the jump to the 4nm node will affect the production cost of the GeForce RTX 5060 Tiand that therefore the selling price of this model could end up being higher than that of the GeForce RTX 4060 Ti at the time of launch.

The GeForce RTX 4060 Ti FE hit the market with a price tag 399 dollarsa figure that happened in Spain after the currency was converted and taxes were applied 449 euros. I think the selling price of the GeForce RTX 5060 Ti will be between 449 and 499 euros, including taxes. Some fitters may market slightly cheaper models, but the FE version will be in that range.

AI generated cover image.