Next-gen GPU’s from AMD to contain a stacked accelerator die

More From Author

See more articles

Best Induction Cooktop India as of May 2025

Best Induction Cooktop India: Induction cooktops are becoming increasingly popular due to the hassle-free cooking experience they...

Myntra Upcoming Sales 2025: Your Fashion Calendar for Maximum...

Myntra Upcoming Sales 2025 In the ever-evolving world of fashion e-commerce, Myntra continues to be India's go-to destination...

Dimensity 6020 vs Snapdragon 695: Mid-Range Chipset Battle

Dimensity 6020 vs Snapdragon 695: Qualcomm Snapdragon 695 5G (SD695) is a fast mid-range ARM-based SoC found...

MCM technology is just the beginning for AMD’s next-generation RDNA GPUs, which are becoming more technically advanced with each iteration. According to a patent discovered by Coreteks, AMD addresses the integration of stacked accelerator dies on board next-gen GPUs.

AMD’s GPU MCM solution already uses cutting-edge technology, and next-generation RDNA GPUs with 3D Infinity Cache technology in a chipset-based architecture are rumoured. According to the latest rumours, APD, or Accelerated Processor Die, may be coming to next-generation RDNA GPUs. Consider it a die built into the primary GPU (perhaps a stack chipset) that does machine learning operations.

The APD die is described as a memory and machine learning accelerator die that comprises a memory, machine learning accelerators, memory interconnects, inter-die interconnects, and controllers in the two patent schematics. The memory on the APD die can be used as a cache for the APD core die or directly by machine learning accelerator operations like matrix multiplication.

The unit guides a set of machine learning arithmetic logic units to complete a set of machine learning tasks via one or more inter-die interconnects once a request to execute a shader task on the APD core die is received.

These specialist AI/ML cores could be AMD’s response to NVIDIA’s Tensor cores, which power NVIDIA’s DLSS gaming package and also help with DNN and Machine Learning activities in HPC. As companies tap in more performance by delegating some activities to these GPU-assisting accelerators, such specialized cores will be the main component of next-gen GPUs such as RDNA 3 and beyond.

With being stated, patents like these don’t just appear out of nowhere. This one was released on December 2nd, and AMD’s flagship RDNA 3 GPU is already supposed to be taped out. It’s very plausible that if APD ends up being a stacked chipset, it will be readily incorporated when RDNA 3 is mass-produced, or that it will end up with RDNA 4 or something else. If it can assist enhance performance, it’s an intriguing technology that we’d like to see included in our gaming GPUs.

also read:

Intel CPUs report the highest share for the entire year 2021, and it’s all due to Alder Lake

Source

LEAVE A REPLY

Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.

━ Related News

Featured

━ Latest News

Featured