The EOMA68 Libre Computer Developer Wants To Tackle A Quad-Core RISC-V Libre SoC Design

Stemming from the recent proposal about a libre GPU using a RISC-V chip running a Rust-based software renderer like a software-based Vulkan implementation, the developer appears to be ready to take on designing a quad-core RISC-V libre SoC that he believes can be competitive for mobile devices.

Luke Kenneth Casson Leighton, the developer most commonly known for his multi-year effort on the EOMA68 libre computer cards/laptop that still is in the works, wants to design a quad-core 64-bit RISC-V SoC that is a fully open-source/libre design.

He wants to pursue a "Libre RISC-V SoC" design that consumes significantly lower power consumption than other (ARM-based) mobile chips that he believes can lead to a 40% reduction in power consumption, reduce development costs for others getting involved, and reduced royalties would mean a reduced selling price for the hardware.

The efforts are noble but it will be very tough to achieve with the limited manpower and resources. Let's keep in mind that even SiFive's rather sizable efforts for RISC-V with their initial Freedom U540 SoC on the HiFive Unleashed board isn't all that fast (much slower than recent ARM SoCs) but rather pricey at $999 USD.

There is also the EOMA68 libre efforts to keep in mind that after that successful crowdfunding effort of $230k+ USD, the boards still aren't shipping years after its original design going back to the KDE Vivaldi tablet days and two years after that crowdfunding campaign. It's looking like their goal to ship in 2018 will be missed. The latest update on the EOMA68 front is they have placed an order for the PCBs. But due to the time and costs involved, they might end up sending out the PCB unassembled and leave it up to the users. "It’s not hard: it just needs to be done very carefully," from the project update earlier this month. The EOMA68 effort is still based on an Allwinner A20 SoC that uses two Cortex-A7 cores.

So while I applaud the initiative to create a fully libre quad-core RISC-V SoC, unless some notable backers step up behind the project, I'm having a hard time seeing it come to a fruition in today's environment but I would love to be proven wrong. Creating a quad-core RISC-V SoC for general computing purposes will be hard enough but for it to be successful for running a Vulkan (or OpenGL) software-based solution to the extent of handling "3D mobile games" will be even more challenging, especially if going by our past trials of LLVMpipe on multi-core high-end x86_64 CPUs.

More details on the libre RISC-V proposal can be read at CrowdSupply.com.