Documentation

Everything you need to start designing chips with AI.

Quickstart

Get from zero to your first AI-synthesised chip in under 10 minutes with our Docker setup.

Read guide →

API Reference

Full REST API documentation for all orchestrator, agent, and model endpoints.

View API →

Architecture

Deep dive into the 5-layer APEX-EDA architecture: infra, EDA tools, AI models, agents, API.

Read docs →

AI Models

HeteroGNN-Place, GraphWISE timing, I-GNN IR-drop, PPO routing agent — training and inference.

Read docs →

Chipset RTL

SystemVerilog source for APEX-X1: compute tiles, UCIe links, HBM4 controller, chiplet mesh.

View RTL →

Datasets

CircuitNet 2.0, ISPD benchmarks, synthetic RTL generation pipeline for model training.

Read docs →

Quickstart: Your First AI Chip

Step 1 — Clone and start
git clone https://github.com/apexsilicon/apex-eda
cd apex-eda
./setup.sh   # pulls Docker images, starts all 8 services, runs smoke test
Step 2 — Describe your module
curl -X POST http://localhost:8300/generate \
  -H "Content-Type: application/json" \
  -d '{
    "description": "4-bit synchronous counter with active-high reset",
    "module_name": "counter",
    "ports": [
      {"name": "clk",   "direction": "input",  "width": 1},
      {"name": "rst",   "direction": "input",  "width": 1},
      {"name": "count", "direction": "output", "width": 4}
    ],
    "pdk": "sky130"
  }'
Step 3 — Run the full flow
curl -X POST http://localhost:8080/flow/rtl-to-gdsii \
  -H "Content-Type: application/json" \
  -d '{
    "rtl_content": "<paste generated verilog here>",
    "top_module": "counter",
    "pdk": "sky130",
    "target_frequency_mhz": 100
  }'

# Returns: { "flow_id": "abc123", "status": "running" }
Step 4 — Poll for results
curl http://localhost:8080/flow/abc123/results

# Returns: {
#   "status": "complete",
#   "area_um2": 312.5,
#   "timing_wns_ps": -8.2,
#   "power_mw": 3.1,
#   "drc_violations": 0,
#   "gdsii_path": "/workspace/abc123/final.gds"
# }
API Endpoints Reference
POST /flow/rtl-to-gdsii          Full pipeline run
POST /flow/synthesise-only       Synthesis only
GET  /flow/{id}/status           Flow status + current stage
GET  /flow/{id}/results          Final metrics + artifacts

POST /agent/design               Autonomous AI design run
GET  /agent/design/{id}          Agent design status
POST /agent/chat                 Conversational interface

POST /api/nl/generate            NL → RTL generation
POST /synthesise                 Direct Yosys synthesis
POST /place                      Direct OpenROAD placement
POST /route                      Direct TritonRoute routing
GET  /health                     Service health