Managed Node Composition Across Compute Sleds for Workload Scale-Up

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In data centers, workloads often underutilize or overutilize available processors, leading to resource wastage and inefficiencies in cloud services due to incorrect resource allocation, resulting in financial and time losses.

Innovation Solution

The implementation of a system that dynamically composes managed nodes with multiple processors from multiple compute sleds, allowing for cooperative execution of workloads across physically dispersed processors, utilizing an orchestrator server and accelerator to manage coherence data and bandwidth for optimal resource utilization.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If multiple processors are physically located on the same compute device, then communication between processors is fast through shared memory and local bus, but the workload may not utilize all available processors leading to resource underutilization

Engineering Contradiction:

Improveprocessor communication speedVSAvoidresource utilization

Core Design Contradiction:

SpeedVSProductivity

Solution Approach 1:

The system segments the compute resources by physically separating processors onto different compute sleds while maintaining logical coherence through coherence logic units. Each compute sled operates as an independent segment with its own processors and memory, yet they function together as a unified managed node through the coherence protocol.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The coherence logic unit acts as an intermediary between processors on different compute sleds, managing coherence data and maintaining consistency across distributed memory systems. This intermediary enables fast communication comparable to local bus speeds while allowing access to remote processors and memory.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Productivity

If more processors are allocated to a workload, then concurrent execution capability increases, but resources may be overallocated leading to wastage and financial loss

Engineering Contradiction:

Improveconcurrent execution capabilityVSAvoidresource wastage

Core Design Contradiction:

ProductivityVSLoss of energy

Solution Approach 1:

The system dynamically allocates and deallocates processors to workloads based on real-time demands. The orchestrator server continuously monitors workload requirements and adjusts the composition of managed nodes, adding or removing processors from compute sleds as needed to match actual concurrent execution requirements without overallocation.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

Processors on compute sleds can be dynamically assigned to different workloads and managed nodes as needs change. The same physical processor resource can serve multiple different workload types and purposes, maximizing utilization and preventing wastage through flexible, multi-functional resource allocation.

Inventive Principle:

Principle #6Universality (Multi-functionality)

3Adaptability or versatility

If processors are distributed across multiple compute sleds, then resource flexibility and scalability increase, but system complexity increases due to coordination requirements

Engineering Contradiction:

Improveresource allocation flexibilityVSAvoidsystem coordination complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system maintains homogeneity in the coherence protocol and interface design across all compute sleds. Each compute sled implements the same coherence logic unit architecture and communication protocols, simplifying coordination by providing uniform, predictable behavior across distributed resources despite physical separation.

Inventive Principle:

Principle #33Homogeneity

4Productivity

If processors on different compute sleds communicate as if local, then workload execution efficiency improves, but bandwidth requirements and coordination overhead increase

Engineering Contradiction:

Improveworkload execution efficiencyVSAvoidbandwidth consumption

Core Design Contradiction:

ProductivityVSUse of energy by moving object

Solution Approach 1:

The coherence logic units establish coherence protocols and data consistency mechanisms in advance before workload execution begins. By pre-configuring the coherence relationships and memory consistency models, the system enables processors to access remote memory and resources efficiently without incurring excessive coordination overhead during actual computation.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS11977923B2Cloud-based scale-up system composition

Publication Date: 2024.05.07 INTEL CORP

AI summary

Technologies for composing a managed node with multiple processors on multiple compute sleds to cooperatively execute a workload include a memory, one or more processors connected to the memory, and an accelerator. The accelerator further includes a coherence logic unit that is configured to receive a node configuration request to execute a workload. The node configuration request identifies the compute sled and a second compute sled to be included in a managed node. The coherence logic unit is further configured to modify a portion of local working data associated with the workload on the compute sled in the memory with the one or more processors of the compute sled, determine coherence data indicative of the modification made by the one or more processors of the compute sled to the local working data in the memory, and send the coherence data to the second compute sled of the managed node.