Shown older models of the series – GeForce RTX 3090, GeForce RTX 3080, and GeForce RTX 3070. They are based on the GPU generation Ampere and are made using an 8-nm process technology.
The RTX 3090 boasts 24GB of GDDR6X memory clocked at 19.5GHz. Bus – 384 bits, total bandwidth – 396 GB/s. The company claims that the peak performance of the new product is 36 teraflops. When working with tracing technology, the best indicator improves to 69 teraflops. 3DNews writes that performance in operations using tensor cores reaches 285 teraflops (Tensor-TFLOPS).
NVIDIA also claims that the RTX 3090 will run games at 8K resolution at 60fps. They also say that compared to the Titan RTX, performance at 4K resolution has increased by 50%.
The RTX 3080 is another flagship model that sits down one notch in the product line. It has 10 GB of GDDR6X memory, a 320-bit bus, and a bandwidth of 760 GB/s. The performance of the card with normal rendering is 30 teraflops. The manufacturer claims that overall, the RTX 3070 is twice as powerful as the RTX 2080.
Another graphics card, the RTX 3070, is the lowest-powered graphics card in the lineup. However, the progress in comparison with the similar model of the previous generation is obvious, and considering the price, the model is competitive. It has 8 GB of its own GDDR6 memory and a performance of 20 teraflops. The company claims the RTX 3070 is faster than its previous flagship, the RTX 2080Ti.
The RTX 3080 will be available for purchase on September 17th for $699, the RTX 3070 will be available for purchase in October for $499. The most advanced RTX 3090 will arrive on September 24th for $1,499.