"Agents don't operate on GPUs alone. They need CPUs in order to do their work, whether we're training agentic models or serving them, GPUs today actually call out to CPUs in order to do the tool calling, SQL queries and the compilation of code. This sandbox execution is a critical part of both training and deploying agents across data centers."
"Those CPUs need to be fast to avoid becoming a bottleneck. That requires a new kind of AI-optimized CPU which balances per core frequency, density, and power efficiency."
"Vera is Nvidia's latest CPU and brings several notable improvements, including 88 custom Olympus Arm cores, support for simultaneous multithreading, a much wider memory bus, and faster chip-to-chip interconnects."
Nvidia introduced a new liquid-cooled rack system featuring 256 custom Vera CPUs, designed to address the computational needs of AI agents and reinforcement learning that cannot run on GPUs alone. Agents require CPUs for tool calling, SQL queries, and code compilation during both training and deployment. Vera represents Nvidia's latest CPU advancement, featuring 88 custom Olympus Arm cores, simultaneous multithreading support, wider memory bus, and faster chip-to-chip interconnects. The system aims to eliminate CPU bottlenecks through AI-optimized design balancing per-core frequency, density, and power efficiency. Nvidia plans to position Vera as an alternative to x86 processors from Intel and AMD, claiming 3x more memory bandwidth and 1.5x performance per core compared to contemporary competitors.
Read at Theregister
Unable to calculate read time
Collection
[
|
...
]