NVIDIA GeForce RTX 4070 GPU Family Specifications

We reviewed the NVIDIA GeForce RTX 4070 Founders Edition video card, offering a more affordable RTX 40 Series GPU geared for 1440p high framerate gaming. Its gaming performance was very fast, surpassing the previous generation at the same price point by a good margin. It does so with great power efficiency, offering a cool running well built video card with exceptional savings on power demands. It offers more features than the previous generation, but the price point is elevated in this category compared to the previous generation, raising the price segment for this tier of GPU.

GPU Family Basics
GPU Manufacturer	NVIDIA
GPU Generation	RTX 40 Series
GPU Architecture Name	Ada Lovelace
GPU Die Code	AD104
Default TGP	201 W

GPU Family Memory
Default Memory Bandwidth	504 GHz
Default Memory Size	12 GB
Memory Bus Width	192 bit
Memory Type	GDDR6X

GPU Family Chip Details
Transistors	35.8
Die Size	294 mm^2
CUDA Cores	5888
RT Cores	46
ROPs	64
TMUs	184

GPU Family Clock Speeds
Default GPU Base Clock	1920 MHz
Default GPU Boost Clock	2475 MHz
Default Memory Clock	21 GHz

The GeForce RTX 4070 has 4 Graphics Processing Clusters, 23 Texture Processing Clusters, 46 Streaming Multiprocessors, and 5,888 CUDA Cores. It has 184 4th Generation Tensor Cores and 46 3rd Generation RT Cores as well as 184 Texture Units and 64 ROPs. The base clock is 1920MHz and the boost clock will be 2475MHz on the Founders Edition. Memory consists of 12GB of GDDR6X running at 21GHz on a 192-bit memory bus providing 504GB/s of memory bandwidth. For comparison, the GeForce RTX 3070 Ti and 3070 had 8GB of VRAM, therefore 4070 does provide an upgrade over the previous generation.