Nvidia GeForce RTX 4090 Massive Leak: Fast GDDR7 Memory Almost Confirmed

Here is a fresh update regarding what probably could be the next flagship GPU from NVIDIA -- RTX 4090. According to the YouTube channel named Moore's Law Is Dead, the RTX 4090 GPU with the AD102 chip could be the first mainstream GPU to use GDDR7 memory, which should both improve the performance and energy efficiency.

NVIDIA GeForce RTX 4090 Specifications (AD102)

According to the leak, the RTX 4090 GPU will use GDDR7 video memory. The GPU is said to have 18432 CUDA cores, which is the highest number of CUDA cores that we have seen on any graphics card. On top of that, the GPU is also said to offer around 80 to 100 percent rasterization increase when compared to the RTX 3090.

Even when it comes to ray-tracing performance, the RTX 4090 should offer at least double the performance as of the RTX 3090. Lastly, the flagship RTX 4090 graphics card is said to have an on-air TDP of around 450-600W, which makes this GPU a one power-hungry graphics card. it is also said that the same AD102 chip will be used on the RTX 4080, which will have slightly lower CUDA cores.

NVIDIA GeForce RTX 4070, RTX 4060 Ti (AD103)

The RTX 4070 and the RTX 4060 Ti are likely to use the AD103 chip and are expected to have 10752 CUDA cores. It is said to offer 256-bit GDDR memory. However, as of now, there is no confirmation if it uses GDDR7 or the current generation GDDR6x.

In terms of performance, the RTX 4070 is said to be around 10 to 30 percent faster than the current RTX 3090, which is a huge leap in performance, considering the RTX 4070 will be a mid-tier GPU. Lastly, even the ray-tracing performance is also said to be at least twice as good as the most powerful Ampere graphics card.

NVIDIA GeForce RTX 4060 (AD104)

The NVIDIA GeForce RTX 4060 is said to offer 7680 CUDA cores and will use 192-bit GDDR video memory. In terms of performance, the RTX 4060 is said to be as powerful as the RTX 3090 Ti, which is currently the most powerful graphics card from NVIDIA and the same also applies for the ray-tracing performance.

NVIDIA GeForce RTX 4050 Ti (AD106)

The NVIDIA RTX 4050 Ti is said to use the AD106 chip and is likely to offer 4608 CUDA cores. This GPU is said to match the performance of the RTX 2080 Ti and the RTX 2080 Super and is said to use a 128-bit GDDR video memory. Again, there is no confirmation if this GPU will use the high-end GDDR7 video memory.

NVIDIA GeForce RTX 4050 or Below (AD107)

The AD107 is said to be a chip for the entry-level RTX 4000 series of GPU and is likely to be used on the RTX 4050. This GPU will offer 3072 CUDA cores and is said to match the performance of the TX 3060. Even in terms of pricing, the RTX 4050 could be the most affordable next-generation GPU from NVIDIA.