Memory and ICs costly and/or unavailable: What’s to be done?
Global tensions have and are continuing to threaten supply of memory integrated circuits (ICs) and integrated circuits generally. This is at best cause price hikes and at worst unavailability.
This is driving designers to innovate by removing hardware and supplanting it by firmware and software. This removes the need for some of the hardware in the first place. The astute product developer must be cognisant and responsive to this. This all translates into the message below:
Products must have connectivity such that requirements for memory can be lifted into the cloud (AI), paying the attention that’s needed for security. Designs must be understood analytically at the edge (AI) such that storage requirements can be removed via machine learning.
Avoiding need for memory chips via cloud (big data): Pros and Cons
We can, in theory, remove nearly all the memory needs of an Edge device by making it store what it needs to in the cloud. This of course has the requirement that the Edge device has appropriate connectivity, but that is pretty much par for the course these days.
This will remove the need for much of the memory on PCBA but at the cost of latency and greater susceptibility to security issues.
Removing the need for memory chips at source (PCBA): Pros and Cons
Memory requirements have traditionally accumulated on PCBA ICs because it is easier and more reliable to test scenarios with a product and produce look up tables to allow performance under different circumstances to be interpolated and extrapolated. This requires on-chip memory.
The need for on-chip IC memory can be overcome by better understanding, in terms of transfer functions, the process through machine learning.
This allows for the removal of memory ICs and faster performance at the cost of the development time to fully scope the science and mathematics surrounding an electronic product design.
Thinking ASIC: Reducing IC requirement by consolidating through ASIC design
If a product development team is game, a shift in design thinking can assist with removing the need for ICs. ASIC (application specific integrated circuits) used to be the domain of big organisations only, e.g., Apple. This has now changed and lower volume reasonably priced development is now possible. This has the scope to massively reduce IC requirements. However, a shift in design philosophy is necessary to make this happen. Please feel free to read our articles on ASIC development.
How is this integrated circuit supply challenge to be overcome?
EM’s answer is as above: connectivity to allow for cloud storage; machine learning; and, if the stomach for it exists, a shift to ASIC design. If this article is of interest or you wish to consider work with the EM, please contact us.