Intel Patent Explores New Image Rendering Solution
AMD will launch Radeon RX 7000 series graphics cards this year, equipped with Navi 3x series GPUs based on the RDNA 3 architecture. Among them, Navi 31 and Navi 32 will use MCM multi-chip packaging, and GCD (graphics computing chip) and MCD (multi-cache I/O chip) will use two different process technologies.
A recent Intel patent describes how multiple computing modules work together for image rendering, indicating that Intel GPUs will also use MCM multi-chip packages in the future. Intel, which already uses multi-chip designs in Ponte Vecchio for data centers and supercomputers, is no stranger to MCM-like technologies. In this patent, Intel proposes a GPU image rendering solution that integrates multiple chips in the same unit to address issues such as manufacturing and power consumption, while optimizing for scalability and interconnectivity to provide the best performance.
At present, such image rendering problems are solved by algorithms such as AFR (alternating frame rendering) and SFR (framing rendering). However, Intel is talking about checkerboard rendering that integrates computing modules, as well as distributed computing, in order to have higher computing efficiency on GPUs with multi-chip designs. In the patent document, Intel did not describe the details of the architecture too much, but it can be expected that it is only a matter of time before Intel Arc brand graphics cards are equipped with GPUs in MCM multi-chip packages.
Nvidia researchers also published an article detailing how Nvidia is exploring how to deploy multi-chip designs for future products. With the rise of heterogeneous computing, Nvidia is looking for a way to increase the flexibility of its semiconductor designs to flexibly match various modules depending on the workload, which is where the MCM multi-chip package comes in.
AMD has taken the lead in using MCM multi-chip packaging technology on consumer products, and I believe Intel and Nvidia will eventually choose to do so. In the future, AMD, Intel, and Nvidia will likely choose the same process from TSMC, and every slight advantage in technology can have a huge impact on the final product.
