Your next MacBook Pro powered by a future version of Apple’s in-house-designed M1 processor could be even faster and last longer on a charge if Apple’s new hybrid memory patent, which combines high-density, low-bandwidth memory with low-density, high-bandwidth memory, becomes a reality. Rather than share memory between the CPU and GPU on Apple’s current system on a chip (SoC) design — which has its own set of limitations — Apple proposes in its patent, filed with the U.S. Patent and Trademark Office, that the use of a hybrid system would be more efficient and deliver additional performance.
“Providing a memory system with two types of DRAM (e.g., one high-density and one low-latency, high-bandwidth) may permit a highly energy-efficient operation, which may make the memory system suitable for portable devices and other devices where energy efficiency and performance per unit of energy expended are key attributes,” the company said in its filing.
This would differ from the unified memory architecture, or UMA, that Apple currently employs on its ARM-based processors, as the CPU and GPU would also need to share memory capacity and bandwidth. This in turn could have a material effect on performance, according to Tom’s Hardware.
Conversely, using a hybrid approach, as Apple is proposing, would mitigate the need to use large amounts of costly high bandwidth memory. Apple’s patent combines DDR memory with HBM memory. Apple’s design was likely envisioned for portables, like the MacBook Air and MacBook Pro, as the company detailed that the DRAMs are to be soldered onto the logic board.
“With two types of DRAM forming the memory system, one of which may be optimized for bandwidth and the other of which may be optimized for capacity, the goals of bandwidth increase and capacity increase may both be realized, in some embodiments,” Apple elaborated. “Additionally, energy efficiency may be managed in the high-bandwidth portion of the memory. The portion of the memory that is optimized for capacity may have a lower-bandwidth goal and a relaxed (longer) latency goal, since these goals may be served by the portion that is optimized for bandwidth. Similarly, the portion of the memory that is optimized for bandwidth may have lower area efficiency goals, but latency and energy efficiency improvements may be made.”
The company added that high-bandwidth, low-latency, energy-efficient, and high-memory systems could be achieved in a more cost-effective manner through this hybrid memory architecture. “Particularly, implementing the high-density portion and the high-bandwidth, low-latency portion in separate chips that together form the main memory system may allow for each memory to implement energy-efficiency improvements, which may provide a highly energy-efficient memory solution that is also high performance and high bandwidth,” the company said.
Apple isn’t the only company to work on a hybrid memory architecture. Intel’s Xeon processor use both DDR4 memory and Optane memory to support a hybrid approach, and the next-generation Xeon chipsets are said to support HBM. It’s unclear if or when Apple’s hybrid memory architecture will debut on some future version of the M1 chip — technology companies like Apple often file patents that don’t make it to a final product.
Apple has also been rumored to be working on a future M1 processor version with more cores and enhanced graphics capabilities. Such a chip could find its way into a new Mac Pro.
