NVIDIA and AMD may not be the only graphics giants that are cooking up multi-chip-module (MCM) GPU designs for its next-generation graphics cards. As hinted by a new patent application from the company spotted by Twitter user Underfox, Intel also appears to be toying around with the idea of making multiple GPU modules work within a single package as part of a new architecture that could enable greater performance, reduced manufacturing costs, and other benefits. The title of the patent application is “Position-Based Rendering Apparatus and Method for Multi-Die/GPU Graphics Processing.”
United States Application US20210272349 (FPO)
- Position-based rendering apparatus and method for multi-die/GPU graphics processing.
- A method comprising: dividing an image frame into a plurality of tiles; assigning a non-overlapping subset of the tiles to each of a plurality of graphics processors [et al.].
- A graphics processing system comprising: a plurality of graphics processors, each graphics processor assigned a non-overlapping subset of a plurality of tiles of an image frame, wherein each of the graphics processors is integrated on a separate semiconductor die and comprises a graphics pipeline for rendering the assigned subset of tiles [et al.].
Intel Patent Confirms Work On Multi-Chip-Module GPUs (Tom’s Hardware)
- […] the company frames the MCM as a whole approach as a required step to guide silicon designers away from manufacturability, scalability, and power delivery problems that arise from increasing die sizes in the eternal search for performance.
- It […] seems that Intel is looking at integrating tile-based checkerboard rendering (a feature used in today’s GPUs) alongside distributed vertex position calculation (out of the initial frame pass).
- […] it allows for designs that even include multiple graphics processors working in tandem or just sections of graphics processors.