jgrusewski 6e49d4a2be docs: H100 epoch optimization spec (37s → <5s) + phase profiling ...

Spec: 3-phase plan to reduce DQN training epoch from 37s to <5s on H100.
- Phase 1: eliminate 300 CPU roundtrips in PER sampling (GPU Philox RNG)
- Phase 2: increase kernel occupancy (batch_size 512+, cuBLAS profiling)
- Phase 3: pipeline overlap (dual-stream experience/training)

Profiling instrumentation added to training loop (init/experience/training/validation breakdown).

RTX 3050 baseline: training=97.2%, experience=2.3% — PER sampling
with CPU RNG + DtoH sync is the primary bottleneck.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

2026-03-21 10:23:17 +01:00

..

2026-03-10-zero-cpu-hotpath-design.md

docs: add zero-CPU DQN training hot path design spec

2026-03-10 15:49:40 +01:00

2026-03-11-dqn-factored-actions-design.md

feat(dqn): enable Branching DQN with 45 factored actions (5×3×3)

2026-03-11 22:00:13 +01:00

2026-03-13-gpu-test-workflow-design.md

docs: rev 2 GPU test workflow spec — fix all 4 blockers

2026-03-13 11:10:01 +01:00

2026-03-14-kanidm-sso-design.md

docs: add Kanidm SSO + NetBird mesh design spec and implementation plan

2026-03-14 19:41:40 +01:00

2026-03-15-fused-cuda-training-design.md

refactor(ml): delete mixed_precision module — BF16 unconditional on CUDA

2026-03-16 16:11:48 +01:00

2026-03-21-h100-epoch-optimization-design.md

docs: H100 epoch optimization spec (37s → <5s) + phase profiling

2026-03-21 10:23:17 +01:00