About SiliCore

A student-born engineering team in Türkiye building an open microcontroller SoC and accelerators on RISC-V.

Who we are

SiliCore is a student-born engineering team in Türkiye. We design reusable IP — AXI4-Lite peripherals, an LDPC block, DMA, a neural-network accelerator — verify it with modern methods, and prove the results on FPGA and on open PDKs.

Most of the team is in their first year of an Electrical & Electronics Engineering degree, and most of what we know about digital design and verification is self-taught. We mention that not as an excuse but as context: the flows we use are the industry-standard ones, and we hold the work to that standard rather than to a student one.

The competition

Silicore is developed for TEKNOFEST, Türkiye's national technology competition, under the Chip Design Competition — Microcontroller Design track. The project is packaged with the formal deliverables the competition requires: a Design Technical Report (DTR), a demo report, hardware / verification / validation requirement specifications, and a runnable demo package with guides in English and Turkish.

The competition ran in stages — OTR, DTR, Hardening, and Final. We finished 4th of 94 applicant teams; the scores at each stage are on the home page timeline.

How we work

Two habits shape most of our decisions.

The first is incremental bring-up. Rather than integrating everything and debugging a black box, the design advances through numbered demo milestones, each built on a version already known to run on hardware. When something breaks, the search space is one increment wide.

The second is distrusting our own testbenches. We run a second, independent verification pass using reference models nobody on the team wrote, specifically so a misunderstanding cannot get written into both the RTL and the test that checks it. That habit has already caught a real bug our own testbench passed — the story is on the verification page.

Where we are going

The MCU subsystem is tape-out ready and the AI accelerator is on-chip as of demo_v10, so the near-term work is clean documentation and pushing the AI/ML-driven image-processing capability further. Longer-term, we want to make open silicon practical for startups and industry partners — which means the boring parts, documentation and reproducible builds, matter as much as the RTL.

Licensing

The project is released under the Apache License 2.0.