Claude Code Agent
dc4867c270
feat(snapshot): D→P RDMA link Phase 1 — minimal byte transport
A thin wrapper around mooncake.engine.TransferEngine that does
one-sided RDMA writes between two SnapshotPeer endpoints. Bypasses
SGLang's MooncakeKVManager (which is hard-gated to PREFILL/DECODE
roles via add_transfer_request assertion at conn.py:1563) so the
D→P direction doesn't require invasive role-axis changes upstream.
Smoke test (two subprocess.Popen processes, mlx5_60, 127.0.0.1):
1 KB 9.0 ms (one-time openSegment handshake)
16 KB 0.04 ms 3.5 Gbps
1 MB 0.10 ms 82 Gbps
16 MB 0.58 ms 232 Gbps
64 MB 1.70 ms 316 Gbps (~80% of NDR 400G line rate)
All 5 sizes pass SHA256 verification end-to-end.
Files:
src/agentic_pd_hybrid/snapshot_link.py — SnapshotPeer, SnapshotEndpoint
scripts/snapshot_link_receiver.py — child-process receiver
scripts/smoke_snapshot_link.py — sender + verifier
docs/D_TO_P_PHASE1_LINK_ZH.md — phase 1 acceptance doc
Next: Phase 2 (D-side scheduler commit hook), Phase 3 (P-side prefill
bypass with snapshot KV). See docs/D_TO_P_SYNC_DESIGN_ZH.md §5.
2026-05-13 00:55:55 +08:00
..
2026-05-11 22:28:39 +08:00
2026-04-28 21:10:41 +08:00
2026-05-12 00:10:06 +08:00
2026-04-28 21:10:41 +08:00
2026-04-28 21:10:41 +08:00
2026-04-28 21:10:41 +08:00
2026-04-28 21:10:41 +08:00
2026-04-28 21:10:41 +08:00
2026-04-28 21:10:41 +08:00
2026-05-12 00:21:36 +08:00
2026-05-12 11:45:09 +08:00
2026-05-13 00:55:55 +08:00
2026-04-28 21:10:41 +08:00
2026-05-13 00:55:55 +08:00
2026-05-06 21:29:56 +08:00
2026-05-12 00:21:36 +08:00
2026-05-12 00:49:53 +08:00
2026-05-12 11:45:09 +08:00
2026-04-28 21:10:41 +08:00
2026-04-28 21:10:41 +08:00
2026-04-28 21:10:41 +08:00
2026-04-28 21:10:41 +08:00
2026-04-28 21:10:41 +08:00
2026-04-29 22:29:21 +08:00
2026-04-29 20:04:21 +08:00
2026-04-28 23:40:03 +08:00
2026-05-06 21:29:56 +08:00
2026-05-09 01:18:13 +08:00
2026-05-09 01:18:13 +08:00
2026-05-09 01:18:13 +08:00