foxhunt

Files

jgrusewski 789dc86589 perf(cuda): Wave 2 — multi-stream overlap, pure-CUDA forward kernels for DQN/PPO/supervised

Multi-stream pipeline (evaluate() closure path):
- Secondary env_stream with CudaEvent sync allows env_step to overlap
  with gather/forward of the next iteration on H100's 132 SMs

Pure-CUDA DQN forward (evaluate_dqn):
- backtest_forward_kernel.cu: warp-cooperative dueling Q forward via NVRTC
- Eliminates candle per-op dispatch overhead, enables future CUDA Graph capture
- evaluate_baseline.rs: auto-detects dueling network and uses pure-CUDA path

Pure-CUDA PPO forward (evaluate_ppo):
- backtest_forward_ppo_kernel.cu: actor MLP → softmax → 45→5 exposure collapse → argmax
- One thread per window, bypasses candle entirely

CUDA supervised signal→action (evaluate_supervised):
- backtest_forward_supervised_kernel.cu: threshold bucketing kernel
- Candle still used for model forward (TFT/Mamba/etc), but argmax is GPU-native

Shared helpers: launch_gather, launch_env_step_on, launch_metrics_and_download
deduplicate code across all four evaluation paths.

914 tests pass, 0 clippy errors.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

2026-03-11 22:05:01 +01:00

baseline_common

fix(ml): wire spread_cost_bps into RL training — commission + spread = total tx cost

2026-03-01 22:10:43 +01:00

cuda_test.rs

chore: clean up examples, update ML binaries and risk tests

2026-02-27 01:33:18 +01:00

download_baseline.rs

fix(ml): remove dead parquet path, enable GPU features by default

2026-03-10 16:54:45 +01:00

evaluate_baseline.rs

perf(cuda): Wave 2 — multi-stream overlap, pure-CUDA forward kernels for DQN/PPO/supervised