-
Notifications
You must be signed in to change notification settings - Fork 5
All issues
Issue creation is restricted in this repository
Issues
is:issue state:open
is:issue state:open
Search results
[RFC] Extend the native vLLM connector with CPU/CXL L1 and SSD L2 caching
enhancementNew feature or requestNew feature or requestStatus: Open.#87 In xcena-dev/maru;feat(maru_vllm): place the layer-overlap load with single_layer_kv_transfer instead of PyTorch indexing
enhancementNew feature or requestNew feature or requestStatus: Open.#86 In xcena-dev/maru;perf(load): CXL→GPU load path does not scale with in-flight requests — TTFT ≈ N × 176 ms when decode is short
enhancementNew feature or requestNew feature or requestStatus: Open.#79 In xcena-dev/maru;- Status: Open.#74 In xcena-dev/maru;
[Bug] deferred load + write-behind 조합에서 요청 취소 시 EngineCore 사망
bugSomething isn't workingSomething isn't workingStatus: Open.#73 In xcena-dev/maru;- Status: Open.#68 In xcena-dev/maru;
- Status: Open.#59 In xcena-dev/maru;
- Status: Open.#52 In xcena-dev/maru;
- Status: Open.#51 In xcena-dev/maru;
- Status: Open.#49 In xcena-dev/maru;
maru_sglang: chunk_size should be dynamically derived from per-token KV size
enhancementNew feature or requestNew feature or requestStatus: Open.#34 In xcena-dev/maru;[Feature] Implement eviction and capacity reclaim on CXL pool exhaustion
high-priorityHigh Priority itemHigh Priority itemStatus: Open.#4 In xcena-dev/maru;