Nvidia's Vera CPU, intended for standalone data centers and Vera Rubin NVL72 servers, utilizes proprietary Olympus cores. These cores are designed for high per-core performance and aggregate throughput, differing from the Grace CPU which used Arm's Neoverse V2 cores. The Olympus microarchitecture includes a 10-wide decoder, a neural branch predictor, a deep out-of-order execution engine, six vector execution units, four load pipelines, and two store pipelines, supported by large private caches.
At Supercomputing 2026 (SC26), Nvidia stated that Vera's performance can improve by nearly 39% in specific computationally intensive workloads through architecture-specific software optimizations. This indicates that current software may not fully leverage Vera's extensive execution resources without adaptation. The CPU prioritizes single-thread performance over core count, featuring 88 high-performance cores with spatial multithreading (176 threads).
Nvidia positions Vera primarily for agentic AI and other AI-related workloads, though its Olympus microarchitecture was initially conceived for data processing applications. Key attributes like high per-core performance, high memory bandwidth, and low memory access latency are beneficial for these target applications. The design reflects a trade-off to maximize instructions per clock (IPC) rather than the sheer number of cores per socket.
✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →
One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.
One email a day. Unsubscribe in one click, any time.
Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.
▶ Play today's briefNew every morning, and the back catalogue is archived by date.
Nvidia's Vera CPU, featuring custom Olympus cores, can achieve up to a 39% performance increase in computationally intensive workloads through architecture-specific software optimizations. While Vera offers high single-thread performance and is designed for agentic AI and data processing, it requires software adaptation to fully utilize its resources, unlike its predecessor Grace which used off-the-shelf Arm cores.