Multi-Axial Buffer Circuitry for Vector Data Transfer
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data processing systems face inefficiencies in transferring data structures between memory and register banks, particularly in vector operations, requiring a significant number of accesses to memory and register banks to gather and scatter data elements, which hinders performance.
Innovation Solution
The implementation of a multi-axial buffer circuitry with an array of storage elements organized in sets and groups, allowing data elements to be stored and retrieved based on identifiers, enabling efficient construction and storage of vectors in parallel, reducing memory and register bank accesses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data structures are transferred directly between memory and register bank using conventional techniques, then data element gathering can be performed, but a significant number of memory and register bank accesses are required which reduces performance
Solution Approach 1:
A multi-axial buffer is introduced as an intermediary component between memory and the register bank. This buffer contains an array of storage elements organized in N sets along a first axis and VL groups along a second axis, enabling efficient reorganization of data structures. The buffer receives data structures from memory, stores them in appropriate groups based on identifiers, and then outputs the reorganized data to the register bank, significantly reducing the total number of accesses required.
Solution Approach 2:
The buffer employs a two-dimensional array structure with N sets along the first axis and VL groups along the second axis. This multi-axial organization allows data structures to be stored and retrieved along different dimensions, enabling parallel construction of multiple vectors simultaneously. The dimensional reorganization transforms the data access pattern from sequential to parallel, improving transfer efficiency.
2Reliability
If multiple data structures are accessed to form vector operands, then complete vector data can be assembled, but the number of access operations increases which affects energy efficiency
Solution Approach 1:
The buffer merges multiple data structure accesses into a single unified buffer operation. By storing multiple data structures in the buffer's array structure and then retrieving them in the desired vector format, the system combines what would otherwise be multiple separate memory-to-register transfers into more efficient buffered operations, reducing overall energy consumption.
Solution Approach 2:
Data structures are preliminarily stored in the buffer in groups based on their identifiers before the actual vector construction is needed. This preliminary organization allows subsequent vector assembly operations to proceed efficiently without requiring additional memory access operations, thereby reducing energy consumption during the critical vector formation phase.
Data Source
AI summary
An apparatus and method are provided for transferring a plurality of data structures from memory into one or more vectors of data elements stored in a register bank. The apparatus has first interface circuitry to receive data structures retrieved from memory, where each data structure has an associated identifier and comprises N data elements. Multi-axial buffer circuitry is provided having an array of storage elements, where along a first axis the array is organized as N sets of storage elements, each set containing a plurality VL of storage elements, and where along a second axis the array is organized as groups of N storage elements, with each group containing a storage element from each of the N sets. Access control circuitry then stores the N data elements of a received data structure in one of the groups selected in dependence on the associated identifier. Responsive to an indication that all required data structures have been stored in the multi-axial buffer circuitry, second interface circuitry then outputs the data elements stored in one or more of the sets of storage elements as one or more corresponding vectors of data elements for storage in a register bank, each vector containing VL data elements. Such an approach can significantly increase the performance of handling such load operations, and give rise to potential energy savings.