AMD recently published a patent for distributing the load from the display to multiple GPU chips. The game scene is broken into individual blocks and spread on wooden boards to improve the use of shading in games. A two-tier foil container is used for this.
AMD publishes a patent to implement GPU chiplets to make better use of shader technology
A new patent released by AMD provides further insight into the company’s plans for next-generation GPU and CPU technology in the years to come. At the end of June it was announced that 54 patent applications had been submitted for publication. It is not known which of the more than fifty published patents will be used in AMD’s plans. The applications discussed in the patents illustrate the company’s approach in the following years.
An app, noted by community member @ETI1120 on the website Computer base patent number US20220207827, discusses critical image data in two stages to efficiently route a large amount of display from the GPU across many chips. This CPU was first filed with the US Patent Office late last year.
When image data is rasterized on the GPU by default, the shader unit, also known as the ALU, performs the similar task and assigns a color name to individual pixels. In contrast, the textured polygons found in the selected pixel in a given game scene are mapped directly to the pixel. After all, the formulated task retains atypical principles and differs only in other textures located in different pixels. This method is called SIMD or Single Instruction – Multiple Data.
For most current games, shaders aren’t the only task that the GPU has spawned. Instead, many post-processing elements are added after the initial shading. Actions that the GPU will add include preventing anti-aliasing, vignetting, and blocking in the gaming environment. However, ray tracing occurs together with shading, creating a new calculation method.
When we talk about the GPU that drives the graphics in today’s games, the load generated by the computer increases exponentially to thousands of compute units.
For games on GPUs, this computing load is ideally several thousand compute units. This differs from processors in that applications must be specially written to add more cores. The CPU scheduler creates this action and breaks the work from the GPU into more understandable tasks that are processed by processing units, also known as binning. The image from the game is presented, and then divided into separate blocks containing a certain number of pixels. The block is computed by a graphics processor subunit where it is synchronized and generated. After this process, the pixels to be counted are combined into one block until the graphics card subunit is finally used. Considerations are made for shading computing power, memory bandwidth, and cache sizes.
AMD states in the patent that partitioning and bonding requires extensive and complete data connectivity between all elements of the GPU, which poses a problem. Datalinks not included in the template have high latency, slowing down the process.
CPUs have made this transition to chiplets effortless as they can send the job across multiple cores, making them very amenable to chiplets. GPUs don’t offer the same flexibility, making them comparable to a dual-core preprocessor.
AMD recognizes the need and tries to provide answers to these problems by changing the rasterization pipeline and sending tasks between multiple GPUs, similar to CPUs. This requires advanced binning technology offered by the company Binning Binning, also known as Binning Binning.
In superassembly, the division is processed into two separate phases, rather than processing directly into pixel-by-pixel blocks. The first step is to calculate the equation, take a 3D environment and create a 2D image from the original. The stage is called the vertex shader and it is completed before rasterization, and the process is very small in the first chip of the GPU. Once the game scene is complete, it begins to fade and evolve into jagged boxes and processing in a single GPU chip. After that, routine tasks such as puncturing and post-processing can begin.
It is not known when AMD intends to use this new process or if it will be approved. However, it does give us a glimpse into the future of more efficient GPU processing.
News Sources: Computer Base, and the Free Patents Online