NVIDIA’s official information about NVIDIA’s GeForce RTX 40 series, which many gamers waited for, was released on September 21, 22, at NVIDIA GTC ‘GeForce Beyond: GTC Special Broadcasting’. The broadcast was played by Jensen Hwang, the founder and CEO of NVIDIA.
The RTX 4090 and RTX 4080 were first released in the RTX 40 series, which adopted the new ADA LOVELACE architecture. The RTX 4090 is expected to boast twice as much as the RTX 3090TI and will be sold on October 12. The RTX 4080, which will be divided into VRAM 16 GB and 12 GB, will provide twice as fast as the RTX 3080ti and will be available from November. NVIDIA’s official price was set at $1599, $1199, and $899, respectively. (2.23 million won, 1.67 million won, 1.2 million won, 22.09.21 exchange rate)
The RTX 4090 supports 24 GB of G6X VRAM and is equipped with 16,384 Cuba Core, 1,321 Tensor-TFLOPs, 191 Pre-flops and 83 Shader-TFLOPs. If you use DLSS 3, SER, etc., a new ADA architecture, you can experience up to four times more improved performance in friendly titles such as optimized cyberpunk 2077 and portal with RTX.
The RTX 4080 (16 GB) supports 9.728 UDA Core, 780 TensorFlow, 113 Pre-flops, 49 Shader-Tflop and GDDR6X memory. The RTX 4080 (12 GB) boasts performance beyond the current RTX 3090TI, which boasts the best performance with 7,680 UDA Core, 639 Tensor-TFLOPs, 92 Pre-flops, 40 Shader-Tflop and GDDR6X memory.
NVIDIA unveiled the Founders Edition (Founders Edition (FE) of the RTX 40 series, and informed the collaboration with the non-reference graphics card manufacturers ASUS, Colorful, Ga inward, Galaxy, Gigabyte, In nod, MSI, Plait, ANY and ZO TAC. It also worked with a finished product PC partner for leading systems with ACER, Alienware, ASUS, DELL, HP, Lenovo and MSI.
The new NVIDIA ADA LOVELACE architecture can expect high performance improvement in terms of performance, efficiency and functionality. It is based on customized TSMC 4 N processes with up to 76 billion transistors (28 billion in the last generation).
The ADA provides more than twice the same power compared to the AMPERE architecture adopted in the RTX 30 series, and as the power increases, it provides high scalability and the possibility of overclocking. The GPU architecture is highly parallelized and is most efficient when running a similar workload at the same time. However, high-end ray tracing calculates millions of rays that collide with types of various materials and shapes throughout the scene, exuding the shader and handling inefficient work load.
NVIDIA’s new SER (Shader Execution Reordering) technology has been able to dynamically reconstruct an inefficient workload to improve shader performance up to 2 times and up to 25%in the game. NVIDIA has made a real-time racing in real-time video game. The third generation ray tracing cores of the ADA provides a doubled ray-triangular cross throughput, increasing PRE-FLOP performance by up to 2.8 times.
NVIDIA TENSOR Core accelerates NVIDIA DLSS available in 216 release games and apps. The new 4th generation Tensor core of ADA uses a new 8-bit floating point (FP8) tensor engine, which increases the process in RTX 4090 to up to 5 times and 1.32 Tensorpeta Flops. Gamers can experience NVIDIA’s new DLSS 3 and enjoy a comfortable frame speed.