GTC conference released the highly anticipated RTX 40 series desktop-grade high-performance independent graphics cards. The author learned that the characteristics of RTX 40 series graphics cards are doubled in performance and doubled in ray tracing capabilities.
RTX 40 series is Nvidia's third generation RTX graphics card. It adopts the new Ada-Lovelace architecture and uses TSMC's customized 4N process technology . GPU has 76 billion transistors inserted, 70% more than the previous generation Ampere.
RTX 40 series graphics card SM stream processor has added the shader reordering execution capability, which can reschedule tasks in real time, and the ray tracing speed is increased by 2-3 times. On the RTX 4090 graphics card, it can output 90TFLOPS computing power, and its performance is twice as high as the previous generation.
On the RTX 40 series graphics card, Nvidia introduced DLSS3 from the Ada architecture, which automatically fills the frame while improving the resolution. According to official description, DLSS3 super-resolution technology uses low-resolution content as input and uses AI technology to output high-resolution frames, thereby enhancing graphics card performance. DLSS samples multiple images with lower resolution and uses previous frame motion data and feedback to reconstruct native quality images.
DLSS3 can process the current frame and the previous frame at the same time. The optical flow accelerator provides the neural network with motion direction and speed information. Combining the motion vector of graphics and pixels, input neural network to generate intermediate frames.
DLSS3 generates a brand new frame without involving graphics pipeline processing, which can improve performance by up to 4 times compared to simple rendering. The author learned that DLSS has been used in more than 200 games and applications, including popular games such as "Cyberpunk: 2077" and "Black Myth: Goku".
RTX 40 series graphics card conventional hardware parameters:
RTX 4090 uses AD-102GPU core, has 16384 CUDA cores and 24GB GDDR6X video memory, and the default TDP power consumption is 450W.
RTX 4080 16G has 9728 CUDA cores, and the default TDP power consumption is 320W.
RTX 4080 12G has 7680 CUDA cores, and the default TDP power consumption is 285W.
From the perspective of power consumption, Nvidia's replacement to TSMC's 4N process has significantly improved efficiency and improved performance at the same level.
RTX 4090 has a larger chip through SER optimization. In addition, the chip acceleration frequency has been increased from 1.7GHz to 2.52GHz, and the overall performance has doubled compared to 3090Ti, and the ray tracing performance has been improved by 4 times. Huang Renxun said that Ada's performance of power consumption is twice that of Ampere.
About the pricing of graphics card :
RTX 4090 public version official suggested retail price of 1599 US dollars (equivalent to RMB about 11268 yuan).
RTX 4080 16G public version official suggested retail price US$1,199 (equivalent to RMB about 8,450 yuan).
RTX 4080 12G public version official suggested retail price of US$899 (equivalent to RMB approximately RMB 6335). These three graphics cards will be available on October 12.
From the official pricing point of view, the price of RTX 4090 has almost not increased, and the price of RTX 4080 has increased by $500 compared to the previous generation 3080. For domestic users, public graphics cards are basically unavailable unless they are through some special channels. The author believes that Nvidia should not launch more affordable graphics cards this year because the inventory of the 30 Series has not been cleared yet. On the surface, this is a new graphics card and new technology launch conference, but in fact, this is a live broadcast of goods carefully prepared by Nvidia for 30 series graphics cards. After reading the parameters and prices, will you still choose an RTX 40 series graphics card?
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-