Nvidia has officially unveiled the first of its RTX 4000 series graphics, the GeForce RTX 4090 and RTX 4080. Both will be powered by the company’s new Ada Lovelace GPU and offer performance that’s more than double that of their 3000 series counterparts.
Starting with the RTX 4090, it will utilize 16,384 CUDA cores – compared to 10,752 on the GeForce RTX 3090 Ti and 10,496 on the RTX 3090 – and come with 24GB of GDDR6X VRAM accessed via a 384-bit memory bus. It will also run at up to a boost clock speed of 2,520MHz. Nvidia claims it will have 2X the performance of the RTX 3090 Ti.
| RTX 4090 | RTX 4080 16GB | RTX 4080 12GB | |
| CUDA cores: | 16,684 | 9,728 | 7,680 |
| Memory/VRAM: | 24GB GDDR6X | 16GB GDDR6X | 12GB GDDR6X |
| Boost clock: | 2,520MHz | 2,505MHz | 2,610MHz |
| Bus width: | 384-bit | 256-bit | 192-bit |
| Power: | 450W | 320W | 285W |
As for the RTX 4080, it will see a significant drop in CUDA cores to 9,728 for the 16GB variant of the card or 7,680 CUDA cores for the 12GB variant. Both versions will also use GDDR6X memory accessed via a 256-bit bus.
The company also revealed the Ada Lovelace GPU that powers these GPUs has a whopping 76 billion transistors and enough processing power to provide 90 TFLOPs of shader processing, 200 TFLOPs of ray tracing processing and 1,400 TFLOPs of tensor core processing.
The RTX 4090 release date will be October and it will cost an eye-watering, though not entirely surprising $1,599 or £1,679 inc VAT in the UK, while the RTX 4080 will arrive in November for $1,199 (£1,269) for the 16GB variant and $899 (£949) for the 12GB version.