📦 WP-071 — perf(data): packed + uint8 batch transport¶
Research log · Phase 6 — Evaluation, v0.1.0
🔗 Depends on WP-014 · Needed by none
WP-071 — packed uint8 batch transport¶
The transport format and its error bound are A33. What the register does not carry is the shape of the win: flattening the ragged list[Targets] into eight dense tensors takes the DataLoader IPC hop from roughly 600 pickled segments per batch to 9, with bytes down 4x from the uint8 quantization. Segment count, not byte count, is what makes a containerized shm ceiling bite.