coorl-lost-cities

112 Commits 3 Branches 0 Tags

2 Commits

Include renames

Author	SHA1	Message	Date
coolguyandClaude Opus 4.7	4014e49168	Profile GPU forward to evaluate batched traversal inference Measured DeepCFRMLP forward at bs={1,4,16,64,256,1024} on RTX 3090. Per-state cost drops 232× from bs=1 (80 µs) to bs=256 (0.34 µs) while per-call latency stays near 90 µs through bs=256. Policy-call supply from a real run is ~368 states per traversal and ~200k per iteration, well above the bs=64–256 plateau, so batched inference is not supply-limited. GPU forward is not the limiter once batching exists. Verdict: Optimization Priorities #5 (batched traversal inference) is worth pursuing. End-to-end gain will still be bounded by encoding and worker-GPU coordination overhead. - scripts/profile_gpu_forward.py: standalone profiling script - docs/performance.md: new "GPU forward profiling for batched traversal" experiment section with table, supply estimate, and verdict Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>	2026-05-07 18:35:28 +09:00
coolguy	12b59211b2	Python 포맷팅과 pre-commit 설정 맥락: - Python 코드만 대상으로 Ruff 기반 format/lint와 pre-commit hook을 도입한다. - Cython game.pyx는 Ruff 대상에서 제외해 포맷터 충돌을 피한다. 변경: - ruff와 pre-commit dev dependency 및 Ruff 설정을 추가했다. - pre-commit config와 Python format/check 스크립트를 추가했다. - Ruff format/check --fix 결과로 Python import 정렬과 포맷을 적용했다. 확인: - uv sync --extra gui - uv run pre-commit run --all-files - scripts/check-python.sh - uv run lost-cities-classic	2026-05-06 19:59:07 +09:00

Author

SHA1

Message

Date

coolguyandClaude Opus 4.7

4014e49168

Profile GPU forward to evaluate batched traversal inference

Measured DeepCFRMLP forward at bs={1,4,16,64,256,1024} on RTX 3090.
Per-state cost drops 232× from bs=1 (80 µs) to bs=256 (0.34 µs) while
per-call latency stays near 90 µs through bs=256. Policy-call supply
from a real run is ~368 states per traversal and ~200k per iteration,
well above the bs=64–256 plateau, so batched inference is not
supply-limited. GPU forward is not the limiter once batching exists.

Verdict: Optimization Priorities #5 (batched traversal inference) is
worth pursuing. End-to-end gain will still be bounded by encoding and
worker-GPU coordination overhead.

- scripts/profile_gpu_forward.py: standalone profiling script
- docs/performance.md: new "GPU forward profiling for batched traversal"
  experiment section with table, supply estimate, and verdict

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

2026-05-07 18:35:28 +09:00

coolguy

12b59211b2

Python 포맷팅과 pre-commit 설정

맥락:
- Python 코드만 대상으로 Ruff 기반 format/lint와 pre-commit hook을 도입한다.
- Cython game.pyx는 Ruff 대상에서 제외해 포맷터 충돌을 피한다.

변경:
- ruff와 pre-commit dev dependency 및 Ruff 설정을 추가했다.
- pre-commit config와 Python format/check 스크립트를 추가했다.
- Ruff format/check --fix 결과로 Python import 정렬과 포맷을 적용했다.

확인:
- uv sync --extra gui
- uv run pre-commit run --all-files
- scripts/check-python.sh
- uv run lost-cities-classic

2026-05-06 19:59:07 +09:00