AMD has detailed its latest data center GPU, the MI325X.
The new accelerator is built on the CDNA 3 architecture and features 256GB of HBM3E. The company said that it supports 6TBps and offers 1.8x more capacity and 1.3x more bandwidth than Nvidia's H200 GPU.
It also has 1.3x the peak theoretical FP16 and FP8 compute performance of the H200, the company said.
The first production shipments are expected in Q4 2024, and will be more broadly available from server makers such as Dell Technologies, Eviden, Gigabyte, Hewlett Packard Enterprise, Lenovo, Supermicro, and others starting in Q1 2025.
It will soon be followed by the Instinct MI350 in the second half of 2025. That GPU will sport some 288GB of HBM3E. The company expects to offer a 35x improvement in inference performance compared to CDNA 3 chips.
At the same AMD event, the company announced the Pensando Salina DPU and the Pensando Pollara 400, the industry’s first Ultra Ethernet Consortium (UEC) ready AI NIC.
“AMD continues to deliver on our roadmap, offering customers the performance they need and the choice they want, to bring AI infrastructure, at scale, to market faster,” said Forrest Norrod, EVP and GM, Data Center Solutions Business Group at AMD.
“With the new AMD Instinct accelerators, Epyc processors, and AMD Pensando networking engines, the continued growth of our open software ecosystem, and the ability to tie this all together into optimized AI infrastructure, AMD underscores the critical expertise to build and deploy world-class AI solutions."
