Open Compute Project Courses

High Performance Programmable NIC Architectures for Multi-tenant Networks

Discover how PANIC architecture revolutionizes SmartNIC performance in multi-tenant networks through innovative hardware blocks, switching interconnects, and resource isolation mechanisms.

• YouTube
• 19 minutes
• On-Demand
• Free Video

Using Modularity to Enable Hardware Reuse Across AI Platforms in a Rapidly Evolving Ecosystem

Explore modular architectures and design strategies for flexible GPU platform deployment, addressing data center challenges in AI/ML applications and hardware reusability.

• YouTube
• 15 minutes
• On-Demand
• Free Video

OCP NIC 3.0 PCIe Gen 6 Support with Next Generation SI and Thermal Test Fixtures

Explore PCIe Gen 6 enhancements for OCPNIC 3.0, covering bi-furcation, manageability, test integration, and next-generation SI/thermal test fixtures for improved connectivity and testing capabilities.

• YouTube
• 20 minutes
• On-Demand
• Free Video

Deep Dive into CMS Composable Memory Architecture with Kubernetes Fabric-Attached Memory Orchestration

Explore advanced memory orchestration architecture and APIs for data center management, focusing on fabric-attached memory solutions and Kubernetes integration for modern workload demands.

• YouTube
• 15 minutes
• On-Demand
• Free Video

Enabling Composable Scalable Memory for AI Inference with CXL Switch

Discover how CXL 2.0 switch technology enables scalable memory systems for AI inference workloads, exploring architecture, configurations, and performance improvements for large language models.

• YouTube
• 18 minutes
• On-Demand
• Free Video

DIMSUM: Diving into Multihost Shared and Unified Memory

Explore modern AI's data challenges and how multi-host shared memory devices can revolutionize datacenter architecture, reduce costs, and optimize memory usage for large-scale AI workloads.

• YouTube
• 19 minutes
• On-Demand
• Free Video

Enabling Chiplet Interoperability and Breakthroughs in AI

Explore how AMBA CHI C2C protocol enables seamless chiplet connectivity and acceleration for AI systems, focusing on Arm and NVIDIA's collaborative developments in chip-to-chip communication.

• YouTube
• 15 minutes
• On-Demand
• Free Video

Dating and Marrying: An AI Chiplets Perspective

Explore the challenges and solutions in AI chiplet development, focusing on IBM's research into disaggregation, interoperability, and validation methodologies for domain-specific applications.

• YouTube
• 18 minutes
• On-Demand
• Free Video

Platform Root of Trust Module Interface Architecture and Design

Explore a modular Platform Root-of-Trust architecture that enables flexible system configuration while addressing technical challenges in interface design and security implementation.

• YouTube
• 18 minutes
• On-Demand
• Free Video

NVIDIA MGX: Modular Reference Architecture for Accelerated Data Centers

Discover how NVIDIA MGX's modular reference architecture enables faster development of accelerated computing systems while reducing costs and improving ROI in data centers.

• YouTube
• 19 minutes
• On-Demand
• Free Video

Power Delivery Challenges for AI Accelerators - Advancements and Development of High Density Solutions

Explore power delivery challenges in AI accelerators and innovative solutions for high-density power modules, focusing on AMD Mi300X's vertical stacking and ecosystem standardization.

• YouTube
• 17 minutes
• On-Demand
• Free Video

Scalable I/O Virtualization Framework (SIOV R2) for Next-Generation Datacenters

Explore the next-generation SIOV R2 framework for I/O virtualization, designed to enhance datacenter capabilities through improved VM management, containerization, and accelerator partitioning.

• YouTube
• 13 minutes
• On-Demand
• Free Video

Vertical Power Delivery for Enhanced xPU and AI Product Performance

Explore how vertical power delivery architecture revolutionizes server performance by optimizing DC/DC converter placement and enabling enhanced computing capabilities for AI and ML applications.

• YouTube
• 19 minutes
• On-Demand
• Free Video

Realizing Composable Infrastructure Through DPUs - Building Modern Data Centers

Explore how Data Processing Units (DPUs) enable composable infrastructure in datacenters, focusing on hardware separation, resource allocation, and orchestration mechanisms for efficient workload management.

• YouTube
• 18 minutes
• On-Demand
• Free Video

Beyond CPUs and GPUs - Extending Confidential Computing to DPUs for Enhanced AI Security

Explore how confidential computing extends beyond CPUs and GPUs to DPUs, examining privacy-preserving AI implementations and hardware-based security solutions for third-party infrastructure.

• YouTube
• 15 minutes
• On-Demand
• Free Video