Sızdırılan Özellikler, Duyurulmayan NVIDIA GeForce RTX 5070 SUPER'in Ayrıntılarını Ortaya Çıkarıyor

What’s Been Uncovered

According to reputable insider Kopite7kimi, details about NVIDIA’s upcoming graphics card—the GeForce RTX 5070 SUPER—have surfaced online. The new GPU is said to feature 6,400 CUDA cores, offering a slight bump over the standard RTX 5070, which comes with 6,144 cores.

A major improvement lies in the memory department: the RTX 5070 SUPER is expected to be equipped with 18 GB of GDDR7 VRAM connected via a 192-bit memory bus. This is a significant step up from the 12 GB GDDR7 found in the base model. Such an upgrade addresses the growing demand for higher memory capacity, especially in modern games at 1440p resolution with ray tracing enabled—where 12 GB can sometimes fall short.

Power and Architecture

The total graphics power (TGP) of the RTX 5070 SUPER is rumored to be around 280W, which is 72 watts more than the base RTX 5070. This suggests a modest performance uplift, likely aimed at delivering more consistent framerates under heavier workloads.

As with the rest of the RTX 50 series, the 5070 SUPER will be built on NVIDIA’s latest Blackwell architecture, promising improvements in efficiency and AI-driven features.

Pricing and Positioning

The RTX 5070 SUPER is expected to hit the market at around $600, placing it firmly in the mid-range segment. If the leaked specs prove accurate, this GPU could serve as NVIDIA’s response to previous criticism over insufficient VRAM in their graphics cards. With increased memory and better performance, the 5070 SUPER could become an appealing choice for gamers seeking a future-proof option without entering high-end pricing territory.