pccx v003 — LLM line, separate RTL repository

The v003 line continues the LLM track from the v002 line on a dedicated RTL repository. Active code is not yet present in this docs site — this page is a placeholder while the upstream RTL repository stabilises.

Working state

• working repository name: pccx-LLM-v003, public URL TBD

• working staging repository: pccx-LLM-v003-staging (private)

• repositories created in evidence-tracked state; no release branch has stabilised yet

• shared substrate with v002 — same KV260 board, same W4A8 weight × activation ratio, same L2 URAM organisation

• first model focus — Gemma 4 E4B as the foundation, with one architectural novelty layered on top per release point (v003.0, v003.1)

• v002.x phases (sparsity, speculative, EAGLE, SSD, Tree, benchmark) are not repeated here; v003 starts from a different RTL baseline

Status snapshot

Table 1 v003 layer status (placeholder)

Layer	State
RTL	not yet committed in the v003 repository
Driver / runtime	shared API surface with v002; no v003-specific path yet
Verification	inherits the v002 evidence checklist as its release gate
Throughput claim	none — TBD
Timing closure	none — TBD
Bitstream	none — TBD

This page will be expanded as the upstream RTL repository lands.

See also

• Roadmap — release direction across the pccx ecosystem

• pccx v002 Architecture — current active LLM line (KV260)

• pccx vision-v001 — CNN inference track on KV260 — parallel vision track on the same substrate