TODO-list before ISCA

[n0thingNoob]

Summary

Generate as many usable kernel codes as possible, add measurement metrics for comparison with OpenCGRA, implement a case study, and complete the paper (full draft by 2025-11-03; final by 2025-11-17). Focus on experimental data quality and the case study.

Timeline & Milestones

2025-11-03 — Full Draft v1 (complete narrative; preliminary figures/tables OK)

2025-11-17 — Final Submission (camera-ready text, finalized figures/tables, artifact links)

TODO

Benchmarks (CGRA-benchmark)

• Run all testbenches with current toolchain
• Triage ⇒ ✅ usable / ⚠️ minor fix / ❌ unsupported
• (record reason codes: missing op, routing fail, II constraint, timeout, etc.)
• Add promising new testbenches if feasible (with brief justification)

Metrics & Evaluation (vs. OpenCGRA)

• Define metrics: compile success rate, II/throughput, resource/utilization, routing success, compile time (optional: runtime/energy proxies)
• Implement measurement harness; export CSV/JSON; reproducible script
• Generate comparative tables/plots

Case Study

• Specify setup (arch, params, datasets), methodology, and hypotheses
• Run, collect results, analyze, and draft discussion

Writing

• Review previous three sections for clarity/consistency (Next Monday)
• Replace with ISCA template
• Assemble full draft (includes placeholders for figures/tables as needed)
• Integrate results & figures; finalize narrative

[tancheng]

Metrics & Evaluation (vs. OpenCGRA)

This sounds a little bit limited. Shouldn't we compare with other tools? RipTide, CGRA-ME, and other open-source tools? @syifan WDYT?