Nvidia has clarified that the GeForce RTX 4070 Super does feature a 48MB L2 cache. The AD104 GPU used in the graphics card supports this cache size, enhancing its performance capabilities. The GeForce RTX 4070 Super utilizes the AD104 graphics processor, comprising 7,168 CUDA cores and a 192-bit memory interface. Although it doesn't fully utilize the AD104 GPU like the RTX 4070 Ti, it benefits from the full 48MB L2 cache, improving memory bandwidth.
The initial miscommunication regarding the cache size remains unclear. Nvidia's official product description now accurately states the 48 MB L2 cache and a memory subsystem with a peak bandwidth of 504 GB/s. In terms of compute performance, the GeForce RTX 4070 Super is comparable to the RTX 4070 Ti, offering 35.5 FP32 TFLOPS against the Ti's 40.09 FP32 TFLOPS. With a similar memory configuration, the RTX 4070 Super's performance approaches that of the Ti model, with potential for further enhancement through overclocking.