Nvidia’s Blackwell-based GPU will launch in 2024
Nvidia suffered a ransomware attack from a hacker organization at the beginning of this year. After the internal system was invaded, more than 1TB of data was leaked, including drivers, design drawings, and firmware. In the subsequent large number of exposed pictures, GPUs that NVIDIA had not released at the time were mentioned, including the Ada Lovelace, Hopper, and Blackwell architectures.
In the past few months, Nvidia has released the Hopper architecture and Ada Lovelace architecture GPUs, respectively for data centers and game graphics cards. According to the cycle of NVIDIA technology iterations, the next one will be about two years later. The Blackwell architecture succeeds the Hopper architecture, and it is known that there will be two GPUs, GB100 and GB102.
According to Wccftech reports, at the recent Arete Technology Conference, Ian Buck, Vice President of NVIDIA and Chief Director of Accelerated Computing, reiterated that NVIDIA is committed to updating the main GPU architecture every two years and confirmed that the Blackwell architecture GPU will be launched in 2024.
At present, the Hopper architecture GPU is manufactured using TSMC’s customized 4nm process, and it is also the world’s first GPU using HBM3. It is understood that Nvidia will continue its cooperation with TSMC on the next-generation Blackwell architecture GPU, which is expected to be manufactured using a 3nm process.
It is rumored that the name of the Blackwell architecture may come from the late American statistician and mathematician David Harold Blackwell (April 24, 1919 – July 8, 2010). He is one of the eponyms of the Rao–Blackwell theorem. He was the first African American inducted into the National Academy of Sciences, the first black-tenured faculty member at the University of California, Berkeley, and the seventh African American to receive a Ph.D. in Mathematics.
