Modular GPU Chiplet Architecture for Client and Server Scalability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing graphics processing units (GPUs) are not scalable across different market segments such as clients and servers, requiring separate core architectures and software structures, which limits flexibility and efficiency in data processing.
Innovation Solution
A modular and scalable system-on-a-chip (SoC) architecture for GPUs, allowing selection of modules based on target product classes and market segments, enabling packaging into various end products.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If separate core architectures are designed for client and server GPUs, then each segment can be optimized for its specific use case, but the overall system complexity increases and scalability across market segments is limited
Solution Approach 1:
The patent implements a unified GPU core architecture that can function across both client and server market segments. The same core design supports different workload types (graphics, compute, AI) and can be configured for different performance levels, eliminating the need for separate architectures while maintaining optimization for various use cases through software configuration and resource allocation
2Adaptability or versatility
If different software structures are used for client and server segments, then each can be optimized for its specific requirements, but the difficulty of detection and measurement increases and compatibility issues arise
Solution Approach 1:
The patent employs a unified software structure that can adapt to both client and server requirements. The same software layer handles different workload types and can be configured for specific optimizations without requiring separate software stacks, reducing compatibility issues while maintaining segment-specific performance
3Adaptability or versatility
If a unified modular architecture is implemented across all market segments, then scalability and flexibility improve, but the manufacturing precision requirements increase to ensure consistent performance
Solution Approach 1:
The patent divides the GPU into modular components that can be independently manufactured and then assembled. This segmentation allows each module to be produced with standardized precision requirements, and the modular assembly process ensures consistent performance across different market segments while maintaining flexibility in configuration
Data Source
AI summary
One embodiment provides a graphics processor including an active base die including a fabric interconnect and a chiplet including a switched fabric, wherein the chiplet couples with the active base die via an array of interconnect structures, the array of interconnect structures couple the fabric interconnect with the switched fabric, and the chiplet includes a first modular interconnect configured to couple a block of graphics processing resources to the switched fabric and a second modular interconnect configured to couple a memory subsystem with the switched fabric and the block of graphics processing resources, the memory interconnect including a set of memory controllers and a set of physical interfaces.