In November last year, AMD officially announced the successor of the 7nm Zen3 architecture - the 5nm Zen4 processor, of which the new generation EPYC codenamed Genoa (Genoa) will upgrade the 96-core 192 thread, which is also a new record for the current x86 processor, and now AMD has shipped the first Batch of Zen4 processors to customers to build the 10 billion supercomputer El Capitan.
El Capitan is one of the new generation of supercomputers in the United States, jointly built by AMD and HPE, will be installed at THE LLNL Livermore National Laboratory in the United States, scheduled for delivery in early 2023, with more than 2 exaflops (10 billion times) of double precision performance, is expected to become the world's fastest supercomputer after delivery.
El Capitan will use AMD's new generation of CPUs and GPU acceleration cards, of which the CPU is the ZEN4 architecture EPYC and the GPU is the Instinct MI200 accelerator card.
According to information released by Livermore National Laboratory, El Capitan supercomputer has begun to install, which means that AMD's Zen4 processor has begun to ship in small quantities for the supercomputing team to test, although AMD has not officially launched the Zen4 processor.
Previously reported, Zen4 codenamed Genoa has up to 96 cores, supports 12-channel DDR5, 128 PCIe 5.0, and is mass-produced and listed in the second half of this year.
However, the 96 core is not the limit of Zen4, and there is a derivative version of Zen4c, the product code name Bergamo, up to 128 cores, launched in the first half of next year.
