Nvidia VERA CPU

 



NVIDIA Vera CPU is a next-generation data-center central processing unit developed by NVIDIA, specifically engineered to support large-scale AI, high-performance computing (HPC), analytics, and cloud workloads with very high bandwidth, energy efficiency, and tight integration with GPUs. It forms one of the core components of NVIDIA’s Vera Rubin platform, which is designed as a rack-scale AI supercomputing architecture.

Overview and Positioning

Purpose:

  • Designed to serve as the CPU foundation in modern AI datacenters and AI “factories,” where traditional CPUs often become bottlenecks when paired with large fleets of AI accelerators.

  • Optimized for data movement, memory throughput, and orchestration to keep GPUs fully utilized during AI training and inference workloads.

  • Can also operate as a standalone high-performance platform for cloud services, analytics, HPC, and enterprise workloads independent of GPUs.

Integration:

  • A key part of NVIDIA’s Vera Rubin NVL72 and NVL144 system families, where Vera CPUs work tightly with Rubin GPUs, BlueField DPUs, ConnectX SuperNICs, and NVLink interconnects to deliver scalable AI compute at rack scale.

Architecture and Key Features

Core Design:

  • 88 NVIDIA-designed Olympus CPU cores with 176 simultaneous threads enabled via Spatial Multithreading (a new multithreading approach where hardware resources are partitioned per thread rather than time-sliced).

  • Based on Arm® architecture (Armv9.2 compatibility) but with NVIDIA’s custom core microarchitecture optimized for throughput and efficiency.

Memory and Bandwidth:

  • Supports up to 1.5 TB of LPDDR5X system memory with extremely high bandwidth (up to 1.2 TB/s), critical for memory-intensive AI pipelines, data preparation, and cache operations.

  • Second-generation NVIDIA Scalable Coherency Fabric (SCF) ensures uniform, high-bandwidth access across all CPU cores and the memory subsystem for predictable performance.

High-Speed Interconnects:

  • NVLink-C2C (Chip-to-Chip) interface with about 1.8 TB/s coherent bandwidth connects Vera CPUs to Rubin GPUs, enabling unified memory semantics and efficient data sharing between CPU and GPU.

Advanced Capabilities:

  • Confidential computing support provides hardware-anchored security features protecting sensitive data and code across CPU and GPU domains.

  • Full support for AI-relevant precisions (including FP8) and optimizations tailored to modern AI workflows.

Role in AI and Datacenter Workloads

AI Factories:
In large AI deployments, the Vera CPU is designed to orchestrate GPU workloads by managing data movement, instruction dispatch, cache and KV-store operations, and memory hierarchy, ensuring that GPUs spend more time doing useful work rather than waiting on the CPU.

Standalone CPU Tasks:
Even independent of GPUs, Vera’s architecture targets demanding HPC, analytics, and cloud workloads where predictable throughput, large memory capacity, and high bandwidth are essential.

Comparison to Prior NVIDIA CPUs:
Compared to earlier-generation NVIDIA CPUs (such as Grace), Vera delivers significantly higher memory bandwidth and capacity, improved coherency fabric performance, and tighter GPU coupling, all tailored for next-gen AI and data-heavy workloads.

Industry Context

NVIDIA unveiled the Vera CPU as part of the Vera Rubin next-generation AI computing platform at CES 2026. The complete platform targets commercial deployment in the second half of 2026, claiming significant performance and cost-efficiency gains over the prior Blackwell- and Grace-based systems. 

Comments

Post a Comment

Popular posts from this blog

Nvidia BlueField DPUs

  What Are BlueField DPUs? NVIDIA BlueField DPUs (Data Processing Units) are specialized processors designed to offload, accelerate, and isolate data‑center infrastructure tasks such as networking, storage, and security . Think of them as a third pillar of computing alongside CPUs and GPUs—purpose‑built to handle the heavy lifting of data‑center operations so CPUs can focus on applications. 🔍 Key Characteristics Hardware accelerators for networking, storage, and security tasks Programmable and designed for software‑defined infrastructure High‑performance networking with Mellanox heritage (acquired by NVIDIA) Offload infrastructure workloads like packet processing, encryption, compression, firewalls, and intrusion detection Improve efficiency and free CPU resources for application workloads 🏢 Where Are BlueField DPUs Used? BlueField DPUs are deployed across modern, high‑performance environments where latency, bandwidth, and security are critical. 🌐 1. AI Data Centers Pow...
Read more

Slurm : Job , Step , Task

  In Slurm, job , step , and task describe different layers of work. The terminology can be confusing, so a clean breakdown helps. 🧩 Slurm Concepts: Job vs Task (and Job Step) 1. Job A job is the top‑level unit you submit to Slurm using sbatch , srun , or salloc . Represents the entire workload you want Slurm to run. Has resource requests: nodes, CPUs, memory, time limit, etc. Can contain one or more job steps . Think of a job as the container . 2. Job Step A job step is a subdivision of a job, created with srun inside a job allocation. Each step can run a different program or phase. Steps share the job’s allocated resources. Steps can run sequentially or in parallel. Example: preprocessing → simulation → postprocessing. 3. Task A task is the smallest unit: typically one process (often one MPI rank). Created by srun or by Slurm when launching a job step. If you request --ntasks=8 , Slurm launches 8 tasks . Each task may have 1 or more CPU threads ( --cpus-per-task ). Thin...
Read more

Nvidia Rubin

NVIDIA names its major GPU architectures after scientists (e.g., Kepler, Pascal, Volta, Ampere, Hopper). Rubin follows this convention and is named after Vera Rubin , the astronomer. Rubin = architecture / generation Actual GPUs will have different product names built on the Rubin architecture Analogy with earlier generations To make this concrete: Hopper → Architecture Products: H100, H200 Blackwell → Architecture Products: B100, B200 Rubin → Architecture Products: R100, R200, etc. (exact names not finalized publicly) What “Rubin” represents technically Rubin refers to an entire compute platform , not just a chip: New GPU microarchitecture New tensor core generation Likely paired with next-gen CPUs, networking (NVLink), and memory Targeted primarily at AI training/inference and HPC NVIDIA has also indicated that Rubin will come in multiple configurations (e.g., standard Rubin and Rubin Ultra), similar to Hopper vs. Hopper Superchips...
Read more