With a few days until the release of its latest graphics card, tech giant Nvidia confirmed that the upcoming GeForce RTX 4070 SUPER will have 48MB of L2 cache, which is 12MB more than previously stated. This boosted specification means that, for gaming performance, the new Nvidia GPU will receive more of a boost than initially stated.
At the 2024 Consumer Electronics Show (CES), Nvidia pulled back the curtain on the RTX 40 SUPER Series of graphics cards, which are set to launch through the rest of January. With an MSRP of $599, the RTX 4070 SUPER will be joined by the RTX 4070 Ti SUPER and the RTX 4080 SUPER. Though the latter two graphics cards will supersede the non-SUPER variants, the RTX 4070 SUPER will be sold alongside the current RTX 4070 models. As a result of this launch, the RTX 4070 will see a price decrease, down to $549 on average. Whether this will entice PC builders to go for the RTX 4070 or wait a little longer to get their hands on an RTX 4070 SUPER remains to be seen.
With the January 17 launch date in sight, Nvidia clarified one important specification for their incoming GPU. Though it was initially believed to have the same 36MB L2 (level 2) cache as the RTX 4070, the RTX 4070 SUPER will instead see it increased to 48MB. This detail was corrected by Nvidia on its website days before the launch of the RTX 4070 SUPER. While 12MB may not seem like a lot on the surface, the L2 cache of the graphics card works in tandem with the available VRAM and memory bus. Coupled with the same 12GB of GDDR6X video memory, and the same 192-bit memory bandwidth bus as the RTX 4070 series GPUs, the RTX 4070 SUPER will have significant performance gains for just $50 more than its non-SUPER counterpart.
While performance is game-dependent, the additional L2 cache and higher CUDA core count on the RTX 4070 SUPER will put it within a few percent of the outgoing RTX 4070 Ti model in terms of performance. However, the RTX 4070 SUPER will have around a 15% to 20% performance
Read more on gamerant.com