NVIDIA could launch two GeForce RTX 4050s with different specifications

The GeForce RTX 4050 will be the last of the GeForce RTX 40 series graphics cards to hit the market. We still don’t have an exact date, but knowing that the GeForce RTX 4060 will hit the market in July, it’s very likely that it will come out eventually. between the months of August and Septemberunless NVIDIA decides to delay it.

At the technical level of GeForce RTX 4050 keeps all keys and all backups we saw in the Ada Lovelace architecture, which means that NVIDIA will not cut corners, even if it will be the cheapest graphics card of its new generation. This is exactly one of the things that has generated the most interest and could give it the most value since then having DLSS 2 and DLSS 3 at this level would be awesome.

Thanks to the leaks that have been produced in recent months, we are pretty clear what we can expect from the GeForce RTX 4050but recently new information came out that changed a little bit the idea we already had about the said graphics card, and the truth is that it is quite interesting, so I want to share it with you.

Initially, everything indicated that NVIDIA was going to launch only one version of the GeForce RTX 4050 and that it would be configured with 6 GB of graphics memory on a 96-bit bus. This would leave us with more limited bandwidth and buffering than the GeForce RTX 3050, but to compensate, the former could have 12 MB L2 cachethat is, it would have 10 MB more L2 than the other.

According to new information, NVIDIA could launch a version of the GeForce RTX 4050 that it will have 8 GB of graphics memory and a 128-bit bus. The rest of its specs would remain unchanged, at least in theory, but with this increase in buffer memory and bandwidth, this version would offer superior performance.

Possible specifications of GeForce RTX 4050 6 GB

• AD107 graphics core in 5nm from TSMC.
• 2,560 shaders.
• GPU at 2505MHz-2640MHz, normal and turbo mode.
• 80 texturing units.
• 80 tensor cores.
• 32 raster units.
• 20 RT cores.
• DLSS 3 support.
• 96-bit bus.
• 6 GB GDDR6 18 GHz graphics memory (bandwidth 216 GB/s).
• 13.51 TFLOPs of performance in FP32.
• TGP: 90 watts. You will need an additional 6-pin power connector.

Possible specifications of GeForce RTX 4050 8 GB

• AD107 graphics core in 5nm from TSMC.
• 2,560 shaders.
• GPU at 2505MHz-2640MHz, normal and turbo mode.
• 80 texturing units.
• 80 tensor cores.
• 32 raster units.
• 20 RT cores.
• DLSS 3 support.
• 128-bit bus.
• 8 GB 18 GHz GDDR6 graphics memory (bandwidth 288 GB/s).
• 13.51 TFLOPs of performance in FP32.
• TGP: 90 watts. You will need an additional 6-pin power connector.

Estimated performance and price of both versions

The version with 6GB of graphics memory and a 96-bit bus should perform slightly better in rasterization than the GeForce RTX 2060, but it would still fall short of the 12GB GeForce RTX 3060. It will be the graphics card designed for 1080p playbackand thanks to DLSS 3 it reaches average FPS typical of much more powerful graphics cards.

A GeForce RTX 4050 with 8GB and 128bit bus should be more or less atGeForce RTX 2060 Super level, and thus quite close to GeForce RTX 3060But I don’t think I’ll ever match it. Like the previous one, it will move like a fish in water working with 1080p resolution and with DLSS 3 there will be a difference

In terms of efficiency, both models will represent a significant advance. We don’t have official figures yet, but since the GeForce RTX 4060 will be in the 110 watt range, the GeForce RTX 4050 will most likely have TGP between 80 and 90 watts, which makes it an ideal model for low-energy sources.

In terms of price, the 6GB version should be around $199, and the version with 8GB of graphics memory will be around $249. After converting to Euros and applying taxes, both are most likely to hit the Spanish market with an approximate price of 219 euros and 249 euros respectively.