Many laptop manufacturers have started exploring ARM-based processors. The logic behind using ARM-based chips is the battery-efficiency it provides. Especially if the laptop is searching for a cellular network, like the coveted 5G, ARM hybrid architecture will save battery life when it can. Intel and AMD, as of now, cannot do that same since they do not employ a hybrid architecture.
Intel is changing that with the much-speculated Lakefield chips (Big-Bigger), which will have a hybrid design. The hybrid design would see a productive and efficient processor/cores paired together.
Lakefield processors will harness maximum efficiency by requiring the operating system and applications to be aware of the architecture and target threads to the correct processor cores. Hence, intensive work would utilize productive processor cores. Similarly, processes that require lower intensity would switch to the efficient processor cores.
AMD’s patent registered on USPTO (United States Patent and Trademark Office) is similar in principle to ARM hybrid architecture (big.LITTLE) and Intel’s Lakefield Chips. However, it is worth noting that AMD filed the patent two years ago and is in no way copying Intel.
Enthusiast @Underfox has also spotted that patent and already taken to Twitter.
Patent: Instruction subset implementation for low power operation – AMD
Basically, AMD BIG.little.
This patent still in the adjustment process.
— Underfox (@Underfox3) August 8, 2020
The patent mentions the hybrid technique as “a heterogeneous processor system.” This system theoretically speeds the transfer of threads between high-performance cores and smaller low-performance cores. The patent also mentions a new instruction set architecture that will consume less power when possible. The availability of an economical “lower-feature second processor” benefits AMD mobile processors by helping increase laptop battery life.
With Zen architecture, AMD has already started making its name when it comes to performance. It’s safe to assume, AMD chips have shaken Intel to become more competitive. But the reducing power consumption of the high-performing AMD processors is not easy. Hence, the use of separate low power consuming cores through a hybrid system seems more plausible.
The patent will implement two subsets of instructions.
- The first subset for the larger high-feature processing cores optimized for higher performance
- The second subset for the smaller low-feature processing cores optimized for low power consumption and lower performance than the first processor.
The patent also outlines a method where the cores will use a shared memory location to speed the transfer of threads between the two types of processor cores. Like with Lakefield chips, intensive work would utilize productive cores, and the efficient cores would be for processes that require lower intensity. In case cores are not utilized, it would shut off to increase the power efficiency.
The difference between Intel’s technique and AMD’s, as mentioned in the patent, is the implementation. Intel requires the OS and apps to implement the hybrid architecture, whereas AMD does not. AMD begins execution at the hardware level by allowing the processor to do the work. The processor sorts which cores to utilize depending on the load.
AMD’s patent also mentions that the cluster of cores could be CPUs, GPUs, or DSPs. However, patents usually take a long time to make it to the production phase. Sometimes they may not even make it there.