Dataflow Gaskets with Circular Buffers for On-Chip Data Movement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data movement techniques in electronic systems, such as standard bussing and point-to-point bussing, face inefficiencies due to varying data traffic characteristics and resource overhead, leading to potential bottlenecks and increased power consumption.
Innovation Solution
The deployment of dataflow gaskets with circular buffers that facilitate efficient on-chip data movement among circuit blocks by providing low-latency data access and handling memory allocation and buffer management internally, thus reducing software development efforts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If standard bussing is used for interconnecting circuit blocks, then overall throughput is improved, but the bussing can stall during high traffic periods
Solution Approach 1:
The patent segments the data movement infrastructure into multiple independent dataflow gaskets, each capable of handling specific data streams. This segmentation allows parallel data movement paths that prevent stalling, as each gasket can process data independently without blocking the entire bus.
Solution Approach 2:
The patent introduces dataflow gaskets as intermediary components between circuit blocks and the memory system. These gaskets act as mediators that buffer and manage data traffic, decoupling the direct bus connections and preventing stalls by absorbing traffic spikes through their circular buffer mechanisms.
2Adaptability or versatility
If point-to-point bussing is used to connect each compute block to every other compute block, then any arbitrary data traffic is supported, but unnecessary area and power overhead is incurred
Solution Approach 1:
The patent implements universal dataflow gaskets that can handle multiple data traffic types and patterns through a single infrastructure. Each gasket provides multi-functional capabilities including buffering, routing, and data management, eliminating the need for dedicated point-to-point connections for every possible communication scenario.
Solution Approach 2:
The patent uses configurable circular buffers with variable sizes and adjustable memory allocation strategies. These parameter changes allow the same gasket infrastructure to adapt to different data traffic patterns and requirements, providing flexibility without requiring separate dedicated connections for each scenario.
3Speed
If tightly coupled memories are used for low latency access, then data access speed is improved, but memory allocation and buffer management complexity increases
Solution Approach 1:
The patent implements self-managing dataflow gaskets that automatically handle their own memory allocation and buffer management. The gaskets include internal logic that autonomously allocates memory resources, manages circular buffer pointers, and coordinates data access, eliminating the need for external software intervention and reducing overall system complexity.
Solution Approach 2:
The patent introduces internal management logic within each dataflow gasket as an intermediary between the memory hardware and the circuit blocks. This internal mediator handles the complexity of memory allocation and buffer management, shielding the user from these details while providing simple, high-speed data access through the tightly coupled memory interface.
Data Source
AI summary
Apparatus and methods for facilitating data movement among circuit blocks are disclosed. In certain embodiments, dataflow gaskets with circular buffers are deployed in any number or arrangement to achieve efficient on-chip data movement among different circuit blocks of the die. Each dataflow gasket can be attached to a corresponding circuit block using tightly coupled memories to provide low latency and fast access to incoming and outgoing data streams. Furthermore, memory allocation and buffer management can be handled by the internal logic in the dataflow gasket to reduce or eliminate software development efforts. For example, the dataflow gasket can use circular buffers to allow the circuit block to access the dataflow gasket's memories without needing to understand the internal memory addressing of the dataflow gasket.