Gahow Wang 21db2affb4 Trace runner (run_v3_trace.sh) + concurrent mb7 correctness test

2026-05-28 17:28:48 +08:00

..

results

Layerwise under load: overlap benefit survives (bg=16)

2026-05-28 16:30:14 +08:00

cache_aware_proxy.WRITEMODE.py

Proxy write-mode: concurrent prefill+decode dispatch for v3 (EAR_WRITE_MODE=1)

2026-05-28 17:22:18 +08:00

DESIGN.md

Layerwise under load: overlap benefit survives (bg=16)

2026-05-28 16:30:14 +08:00

mb7_layerwise.py

Trace runner (run_v3_trace.sh) + concurrent mb7 correctness test

2026-05-28 17:28:48 +08:00

mooncake_connector.BASE.py

Layerwise KV transfer on Mooncake: PoC + microbench (worktree exploration)

2026-05-28 15:34:43 +08:00

mooncake_connector.LAYERWISE.py

Chunk-safe + concurrent layer-wise connector (per-step incremental shipping)

2026-05-28 17:15:54 +08:00

mooncake_connector.LAYERWISE.v1_singlexfer.py

Chunk-safe + concurrent layer-wise connector (per-step incremental shipping)

2026-05-28 17:15:54 +08:00

run_mb7.sh

Trace runner (run_v3_trace.sh) + concurrent mb7 correctness test

2026-05-28 17:28:48 +08:00

run_v3_trace.sh

Trace runner (run_v3_trace.sh) + concurrent mb7 correctness test

2026-05-28 17:28:48 +08:00