IDC's Worldwide Accelerated Computing Platforms and Systems Taxonomy, 2024

IDC's Worldwide Accelerated Computing Platforms and Systems Taxonomy

• Figure: Accelerated Computing Platforms and Systems Primary Segments

Accelerated Computing Platforms and Systems Taxonomy Changes for 2024

Taxonomy Overview

Hierarchical Mapping: Accelerated Computing Platforms and Systems Primary/Secondary Markets

Definitions

• Accelerator Characteristics

• Figure: Accelerator Characteristics

• Accelerator Design Type, Architecture, and Form Factor

• Design Type

• Architecture

• Graphics Processing Unit (GPU)

• Field-Programmable Gate Array (FPGA)

• Application-Specific Integrated Circuit

• Hybrid/Composite

• Form Factor

• Discrete

• Integrated Onboard

• Integrated on Chip (Embedded)

• Open Compute Project Open Domain-Specific Architecture

• Computational Characteristics

• Memory

• Synchronous Graphics Random-Access Memory Double Data Rate

• High-Bandwidth Memory

• Embedded Dynamic Random-Access Memory

• Synchronous Dynamic Random-Access Memory

• Resistive Random-Access Memory

• Core Count

• Floating-Point Performance

• Dynamic Performance and Precision

• Computational Performance

• Comparing Accelerator Performance

• Interconnect and Access Software

• Accelerator/Platform Interconnect

• Peripheral Component Interconnect Express

• Computer Express Link

• Other Emerging Interconnect Standards

• Access Software

• Accelerator-Specific SDKs and APIs

• Open Standards–Based SDKs and APIs — Open Computing Language

• Other SDKs and APIs

• Virtualization and Partitioning

• Power and Cooling

• Thermal Design Power

• Cooling

• Computing Platform Characteristics

• Figure: Computing Platform Characteristics

• Form Factors, Attach Rate, and Attach Type

• Form Factors

• Small Form Factor Racks and Non-Rack Form Factors

• Medium-Sized Rack-Based Form Factors

• Large Standalone Rack-Scale Form Factors

• Attach Rate

• Attach Type

• Computing System Characteristics

• Figure: Computing System Characteristics

• System Type and Convergence

• System Type

• Convergence

• Scaling and Networking

• Scaling

• Figure: Scaling and the Russian Doll Effect

• Level 1 (Intra-Silicon)

• Level 2 (Intraplatform)

• Level 3 (Intrasystem)

• Level 4 (Intersystem)

• Networking

• Procurement, Installation, and Deployment

• Figure: Procurement, Installation, and Deployment

• Shipment Type

• Factory Installed

• Reseller or SI Installed

• End User Installed

• Buyer Type

• Hyperscalers

• Service Provider

• Non-Service Provider

• Deployment Location

• Deployment Type

• Dedicated (Private) Cloud

• Shared (Public) Cloud

• Noncloud

• Edge

• Use Cases, Workloads, and Offload Type

• Figure: Use Case to Accelerator Type Mapping

• Use Cases

• Industry-Specific Use Cases

• Artificial Intelligence

• Modeling and Simulation (aka High-Performance Computing)

• Data Analytics Use Cases

• Other Application Use Cases

• Industry-Agnostic Use Cases

• Datacenter Hyper-Customization

• IT Infrastructure

• Workloads

• Offload Type

• (General) Application-Offload

• (Embedded) Function-Offload

• Use Case to Server Workloads Mapping

• Figure: Apportioning Accelerated Compute from Use Cases to Workloads

Related Markets

Learn More

• Related Research

• Synopsis