The RTX 5080 uses a complete BG203-400 core, which is different from the previous RTX 4080. Its GB203 architecture consists of 7 GPCs, including 42 groups of TPCs, 84 groups of SM units, 10752 CUDA cores, and 336 5th generation Tensor Core volume cores. It is surprising that the core area of GB203 and AD103 is 378 square millimeters, and the number of transistors has been reduced from 45.9 billion to 45.6 billion, but the number of CUDA cores has increased by 512.
In terms of performance, RTX 5080 has excellent performance, and its performance is close to RTX 4090D. In terms of power consumption, it subverts the impression of people’s high power consumption of Blackwell GPU, lower than RTX 4080. Its video memory uses GDDR7 with a frequency of 30GHz, with 256bit width, a total bandwidth of 960GB/s, an increase of 34%compared to the RTX 4080, close to 1008GB/s of RTX 3090 TI. Compared with previous products, RTX 5090D improved its performance by only 30% of specifications, and was hailed as the most efficient GPU in ten years.
The previous generation RTX 4090 blessed the number of 2.7 times the number of transistors and 1.4 times the operating frequency. Compared with the RTX 3090 TI, it improved by 50%. The excellent performance of RTX 5080 in many aspects such as performance, power consumption and memory bandwidth makes it very competitive in the graphics card market. It is expected to become the new favorite of consumers and gamers. In the future, we will continue to pay attention to its performance in the market Essence
Copyright must be reprinted without permission…Read more by Team Passionategeekz