Hi, I am interested in developing a system on a chip optimized for a ultra-portable solar power source. A variety of low-light power managers exist for microcontrollers, although integrating a microprocessor with low power RAM is still a challenge.

I would also be interested in creating a website that integrates a list from Efabless cores listed at:
https://platform.efabless.com/projects/public with OpenCores. Their Slack channel had expressed the rationale for developing that page was inspired in part on Open Cores (see attachment). Since it seems like the goals of both organizations have a lot in common, it could help avoid redundancies in searching for cores, etc.

Additionally, I think it would be good to create a list of cores of which patents have expired, such as 386 and 486. I do not know the status of these, but know that manufacturers in Taiwan still produce them: https://www.vortex86.com/products
https://www.rdc.com.tw/
https://en.wikipedia.org/wiki/List_of_x86_manufacturers

Whether they use a BSD license, I don't know, but I wonder if it is based on an open core.

I also did find an opencore for the 486SX:
https://opencores.org/projects/ao486
https://github.com/alfikpl/ao486

https://www.tinytransistors.net/2020/11/21/intel-486sx/

I used an 486SX on a hand-me down Compaq from a friend in the late 90s- 486SX @33MHZ- more than capable of running Windows 95 on 4MB of RAM and even burning a CD at 1X (taking over an hour). But to modernize this processor, MMC support at the very least would be needed. That I do not know how to do. Ideally this processor could be produced on a 22nm FD-SOI node at Global Foundries:
https://www.cmc.ca/globalfoundries-22fdx-fdsoi-22-nm/

It utilizes some of the lowest leakage processes for making energy-harvesting processors possible.
By comparison, an Ambiq Apollo3 runs at 48MHz and uses 384KB of RAM on TSMC 28nm ULL
https://www.tsmc.com/english/dedicatedFoundry/technology/logic/l_28nm ULP
They also have 22nm ULL, with 12nm in development/testing. Skywater had 90nm FD-SOI, although production is delayed, and may not be available for some time. It was announced in 2022, with MIT's Lincoln Lab offering both radiation-hard and standard FD-SOI. Porting a 486 to run a userspace OS could allow a portable typewriter, internet browser, IRC, VNC, VOIP portable laptop on solar power: https://hackaday.io/project/177716-the-femtotx-motherboard-standard Word processing, Browsing without Javascript, applications that do not require a heavy processor such as a Core i3 or even a Pentium, and relatively low-transistor count for manufacturing at a foundry such as Skywater (https://timeline.intel.com/1989/meet-the-i486 (had 1.2million transistors, compared the 386's 275,000)
https://hackaday.io/project/177716/gallery#c45ab168cc08515a2fa4fd703dbc50ee

Maybe I am being a bit optimistic, but is a 1.2 million transistor chip something that Skywater could produce for an open-source project?

The memory the controllers may also need to be fabricated on a lower process node (90nm 65nm, etc) because I do not know how much power consumption it would use, since the TDP needs of the laptop estimated to be around 5 milliwatts. This TDP is an arbitrary number that I selected to ease the solar-powerable startup time of a low lumen environment (indoor), and is not intended to be representative of typical use cases. It takes an extremely conservative estimate of the amount of lumens that might be available on a very cloudy winter day with just a small window as the energy source. Therefore it is by no means a typical system on a chip. This chip design considers energy as a first order design principle. Thus, many of the conventional chip design processes/expectations cannot be assumed for integrating multiple cores, since the power consumption of each module would need to be electrically compatible- whether that is a 1.8V power source or 1.2V would need to be determined by the designer, but likely a 3.3V or higher would cause an efficiency loss.

Ideally this system on a chip can be powered by a hybrid capacitor, known as a Lithium-Ion-Capacitor. Sodium-Ion Capacitors are also fast chargers, but are not commercialized yet. A small 30F lithum ion capacitor can have 6x the energy density of a typical ELDC, and can store 11mAh. Larger 250F ones can store up to 90mAh. Thus the entire design of the platform anticipates a low power display that proportionally uses around 3-5mW, and allowing an 8-10hr run time on a 1-3milliwatt processor.

To help increase interest in the platform, selecting a codebase such as x86 would not require as much portability, since much software has already been written for it. Going with a RISC-V or ARMv3 processor might consume less energy, but it's not clear whether there would be a lot of software support or porting interest to these lighter weight platforms. Thus I present this project to the OpenCore community. If there was a general interest in a solar powered chip, which platform would you be most interested in developing towards?

Sincerely and thanks for reading,

Giovanni