AWS unveils core-packed Graviton4 and beefier Trainium accelerators for AI

On Tuesday Amazon Web Services has unveiled its next-gen Graviton4 CPUs and Trainium2 AI accelerators at its Re:Invent shindig, which it claims will deliver a healthy boost in performance and efficiency in machine learning. Amazon showed off its latest custom-built Arm-compatible processor, unsurprisingly dubbed Graviton4. …

[HuBo]

Let's palaver under the caoutchouc

I like the name "Project Ceiba". Apt for Amazon as a pre-Columbian Mesoamerican mythical tree where deities might live, a connector of the under-, over- and earthly worlds. And for Nvidia (AI/LLM), the African palaver tree (arbre à palabres, grand fromager), with kapok-covered soft fruits. The caoutchouc (Ficus elastica) could clearly also work for some "Elastic Fabric Adapter" aspects of this project.

Cool name!

[NeilPost]

Cost?

“These chips will be made available in bundles of 16 as part of Amazon's EC2 Trn2 instances. However, for larger workloads, this can be scaled out to up to 100,000 accelerators connected using Amazon's EC2 UltraClusters interconnect for a peak performance of 65 exaFLOPS.”

Do you maybe want to translate that into $dollars/Hr, as once you press the loud pedal, your bank account starts draining very quickly.