Cisco UCS C880A M8 HGX B300 AI Server for AI Workloads

Cisco continues to redefine AI infrastructure with the launch of the UCS C880A M8 Rack Server, now geared up with the ultra-powerful NVIDIA HGX B300 SXM GPUs and the most recent Intel Xeon Sixth-Era (Granite Rapids) CPUs. This marks a convergence of state-of-the-art GPU acceleration, high-throughput CPU compute, and enterprise-grade infrastructure administration, optimized for contemporary AI workloads.

NVIDIA: HGX B300 — Unprecedented AI Efficiency

Drawing on NVIDIA’s messaging across the HGX B300 platform:

• Inference efficiency: As much as 11× increased throughput for fashions like Llama 3.1 405B versus the earlier Hopper era, due to Blackwell Tensor Cores and Transformer Engine optimizations.
• Coaching velocity: As much as 4× sooner for large-scale LLMs like Llama 3.1 405B by way of FP8 precision enhancements, NVLink 5 interconnects (with 1.8 TB/s bandwidth), InfiniBand, and Magnum IO software program.
• SXM kind issue benefit: Excessive-bandwidth socket structure permits superior GPU-to-GPU NVLink connectivity and energy supply with out cabling, minimizing bottlenecks and simplifying scaling.

Intel: Xeon Sixth-Gen CPUs — CPU Energy Meets AI Acceleration

Powered by the newly launched Intel Xeon 6 (6700P/6500P) P-core processors, the UCS C880A M8 delivers:

• As much as 86 cores per socket, doubled reminiscence bandwidth, and built-in AI accelerators—together with Belief Area Extensions (TDX), Superior Matrix Extensions (AMX), Information Streaming Accelerator (DSA), QuickAssist (QAT), and In-Reminiscence Analytics Accelerator (IAA).
• Assist for DDR5-6400 and MRDIMMs, boosting reminiscence throughput additional.
• Optimized for compute-intensive and hybrid AI workloads—from inference pipelines to large-scale coaching nodes.

Cisco: Intersight Administration + AI POD Integration

Cisco’s AI infrastructure goes past uncooked compute:

• The UCS C880A M8 integrates seamlessly with Cisco Intersight, Cisco’s SaaS-based administration platform providing centralized management, visibility, and coverage administration throughout the distributed AI stack.
• It suits inside Cisco AI PODs, modular AI infrastructure options designed for fast deployment at scale. These validated information heart items simplify AI manufacturing facility rollouts whereas guaranteeing interoperability with compute, networking, and safety.

Key Use Instances Enabled by HGX B300 (SXM)

The combination of HGX B300 SXM GPUs and Xeon Sixth-Gen CPUs permits a collection of AI workloads:

1. Actual-Time LLM Inference
   • Run large fashions like Llama 3.1 405B with ultra-low latency and excessive throughput, best for chatbots, brokers, and real-time reasoning.
2. Massive-Scale Mannequin Coaching & Superb-Tuning
   • Benefit from 4× sooner coaching and big inter-GPU bandwidth to coach or fine-tune fashions with tons of of billions of parameters.
3. Excessive-Efficiency AI Pipelines
   • Leverage CPU offload for information preparation and orchestration, alongside GPU acceleration, for ETL, multimodal processing, and inference workflows.
4. AI-Native Information Facilities / AI Factories
   • Construct composable, safe, and scalable AI infrastructure blocks with Cisco AI PODs, prepared for integration in information facilities or on the edge.
5. HPC & Scientific Simulation
   • Run contiguous reminiscence fashions and multi-GPU workloads with enhanced NVLink connectivity for top constancy simulations and analytics.

Abstract Desk

Remaining Ideas

The Cisco UCS C880A M8: HGX B300 with Intel Xeon Sixth-Gen units a brand new benchmark in AI infrastructure. It presents hyperscale-level AI efficiency, rock-solid CPU help, enterprise-grade manageability by way of Intersight, and safe deployments by Cisco Safe AI manufacturing facility with Nvidia and Cisco AI PODs scalable architectures. Whether or not you’re constructing an AI coaching cluster, LLM inference engine, or composable AI infrastructure, this platform is purpose-built for the subsequent frontier of AI.

Uncover the ability of next-gen AI infrastructure—learn the Cisco UCS C880A M8 Information Sheet