Inter-BMC Integration for HPC Platform Management

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

High Performance Computing (HPC) platforms face challenges in the design, deployment, management, and integration of Baseboard Management Controllers (BMCs), as conventional BMCs are not designed to work together, leading to inefficiencies in management tasks, telemetry data collection, and power throttling across hardware accelerators.

Innovation Solution

The integration of multiple BMCs within HPC platforms using a high-speed Out-of-Band (OOB) management link, enabling distributed management tasks, intelligent license management, and power throttling across hardware accelerators, with one BMC acting as a bridge to manage and monitor multiple accelerators.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If multiple conventional BMCs are used in HPC platforms, then management coverage is improved, but management efficiency deteriorates due to inability to work together

Engineering Contradiction:
Improvemanagement coverageVSAvoidmanagement efficiency
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent implements a hierarchical BMC architecture where a root BMC merges control functions across multiple accelerator BMCs. The root BMC aggregates telemetry data from multiple accelerators, coordinates firmware updates, and manages power throttling policies across the entire HPC platform, enabling unified management while maintaining individual BMC functionality for local control.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The root BMC serves multiple functions including: (1) centralized telemetry collection from all accelerators, (2) coordinated firmware update distribution, (3) platform-wide power throttling management, and (4) high-availability takeover capability. This multi-functional design allows a single BMC to manage the entire HPC platform efficiently.

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Reliability

If distributed BMC architecture is implemented, then system availability is improved, but device complexity increases

Engineering Contradiction:
Improvesystem availabilityVSAvoidBMC architecture complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The BMC system is segmented into a root BMC and multiple accelerator BMCs, each with specific responsibilities. The root BMC handles platform-wide coordination and high-availability management, while accelerator BMCs focus on local accelerator control. This segmentation distributes complexity across multiple components with defined roles.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The root BMC acts as an intermediary between the host processor and multiple accelerator BMCs. It receives management commands from the host, translates them into coordinated actions across multiple accelerators, and aggregates telemetry data from all accelerators for centralized monitoring, simplifying the overall architecture.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Speed

If direct PCIe access is used from BMC to accelerators, then access speed is improved, but management control is reduced

Engineering Contradiction:
Improveaccess speedVSAvoidmanagement control
Core Design Contradiction:
SpeedVSEase of operation

Solution Approach 1:

Each accelerator BMC maintains direct PCIe access to its local accelerator for high-speed telemetry collection and control. The root BMC maintains direct PCIe access to the host processor for rapid command reception. This localized direct access preserves speed while the hierarchical coordination layer maintains centralized management control.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent introduces a hierarchical dimension to the PCIe access architecture. While individual BMCs maintain direct PCIe paths to their respective accelerators for speed, the root BMC adds a coordination layer that manages access patterns, prioritizes telemetry collection, and coordinates firmware updates across the entire system.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Data Source

PatentUS11657013B2Inter-baseboard management controller (BMC) integration for high performance computing platforms
Publication Date: 2023.05.23 DELL PROD LP
  • US11657013B2 patent drawing
  • US11657013B2 patent drawing
  • US11657013B2 patent drawing

AI summary

Embodiments of systems and methods for inter-Baseboard Management Controller (BMC) integration for High Performance Computing (HPC) platforms are described. In some embodiments, an HPC platform may include a first BMC coupled to a host processor; and a hardware accelerator comprising: (a) one or more managed subsystems coupled to the host processor, and (b) a second BMC coupled to the one or more managed subsystems and decoupled from the host processor, wherein the first BMC and the second BMC are coupled to each other via a high-speed, Out-of-Band (OOB) management link.