AMD Cyan Skillfish DCN 2.01 display engine support
Display Core programming for AMD's Cyan Skillfish APU/GPU, an early RDNA-based chip that ended up most visibly in Sony PS5-derived BC-250 accelerator boards repurposed from surplus cryptocurrency mining hardware. It handles the display pipeline (DCN version 2.01) for that specific silicon rather than for any mainstream Radeon consumer card.
recommendation
Worth keeping but documenting as a niche path, because the Cyan Skillfish hardware was never a mainstream product and today mostly survives as second-hand BC-250 boards bought from ex-mining surplus. Even so, upstream development remains genuinely active — dozens of commits across many authors with changes landing as recently as late 2025, and Phoronix reported continuing enablement work in early 2025 — so removal is not justified, but a comment flagging the limited hardware footprint would help future maintainers.
repository signals
sources
- kernel.org
AMD Display Core is part of the active amdgpu display driver stack, not a separate legacy out-of-tree driver family.
- phoronix.com
As of January 20, 2025, upstream/open-source work was still improving Cyan Skillfish support, and BC-250 boards were mainly appearing on used resale markets.
- elektricm.github.io
BC-250 boards using Cyan Skillfish are described as ex-cryptocurrency mining hardware repurposed from surplus, indicating niche rather than mainstream new deployments.
- tomshardware.com
Cyan Skillfish was introduced as a specialized AMD APU/GPU codename rather than a broad mainstream PC product line.
codex reasoning notes (technical)
In-tree inspection via shell `rg`/`sed` showed `dcn201` is selected for `DCN_VERSION_2_01`, and `dc_resource.c` maps that path to FAMILY_NV device IDs that match `CHIP_CYAN_SKILLFISH` in `amdgpu_drv.c`. User-provided history shows strong recent maintenance (42 substantive commits in 5y, latest 2025-12-08, 26 authors), so this is not dormant. `lei` was unavailable in shell and no `lore-http` MCP server was present, so direct lore timeline/removal-query verification could not be performed; with active in-tree maintenance and no verified removal thread, I did not escalate to `remove`. Web sources were obtained by web search: kernel.org docs for stack context, Phoronix for 2025 ongoing support work and used-market status, BC250 docs for present-day niche surplus deployment evidence, and Tom's Hardware for chipset/product-family context. Conclusion: hardware is niche and likely not sold new in 2025, but upstream support remains live enough that deprecation/removal is not justified; annotate as niche/internal-IP coverage rather than remove.