Work Item Coordinate Decompression Using Valid and Swizzle Masks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The rate at which work item information can be sent across the interface between a computation requesting unit and a computation sequencing unit in a computing system is limited, restricting the scheduling and execution rate of computational instances.
Innovation Solution
Compressing work item coordinate content by creating a work item valid mask and multiple swizzle masks to represent work item coordinates, allowing for efficient data transmission across the interface.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If work item coordinate data is transmitted across the interface between computation requesting unit and computation sequencing unit, then the scheduling and execution of computational instances can proceed, but the transmission rate is limited and restricts the scheduling and execution rate
Solution Approach 1:
The work item coordinate data is segmented into two parts: a compressed representation (work item valid mask and swizzle masks) that is transmitted across the interface, and the full coordinate data that is generated locally at the computation sequencing unit. This segmentation allows only the essential scheduling information to be transmitted, reducing bandwidth requirements while maintaining the ability to schedule and execute computational instances at higher rates
Solution Approach 2:
The computation requesting unit performs preliminary compression of the work item coordinate data before transmission. By pre-computing the work item valid mask and swizzle masks and transmitting them in advance, the system enables the computation sequencing unit to generate full coordinate data locally without requiring high-speed transmission of all coordinate information, thus improving the overall scheduling and execution rate
Data Source
AI summary
Compressed work item coordinate data for work items in a work group across an interface between a computation requesting unit and a computation sequencing unit is decompressed. A work item valid mask indicates valid work items in the work group. A swizzle index is computed for each valid work item in the work group. A first swizzle mask indicates which bits of the swizzle index for each work item correspond to the value of a first coordinate for that work item. A second swizzle mask indicates which bits of the swizzle index for each work item correspond to the value of a second coordinate for that work item. First coordinates for each valid work item in dependence on the first swizzle mask and the swizzle index for that item, and second coordinates for each valid work item in dependence on the second swizzle mask and the swizzle index computed for that item are computed.