ref(upstream): FULL TREE — Deep-Spark xllm (1470) + ds_vllm csrc/models (703)

Replaces cherry-picked upstream_ref with complete source trees.

xllm/ — Iluvatar official C++ inference engine (15MB, 1470 files)
  Complete: kernels → layers → models → runtime → scheduler → api
  Excluded: .git, binary images, third_party submodule checkouts

ds_vllm/ — Iluvatar official vllm fork (8MB, 703 files)
  Included: csrc/ (ALL CUDA kernels), fused_moe/, qwen3_5 model, _custom_ops
  Excluded: tests, benchmarks, docs, examples (not needed for reference)

Critical call chains now fully traceable:
  MoE: moe_topk_softmax_kernels.cuh → ixformer.h → fused_moe.cpp → layer
  GDN: qwen3_gated_delta_net_base.cpp → qwen3_5_gated_delta_net.cpp
  Attention: ixformer.h → xllm_paged_attention → attention.cpp
This commit is contained in:
EX Engine
2026-08-10 02:53:54 +00:00
parent 9e4fb3712f
commit 002f9879b2
2179 changed files with 494021 additions and 79 deletions

View File

@@ -0,0 +1,20 @@
# Prefix Cache 优化
## 功能介绍
xLLM支持prefix_cache匹配。prefix_cache基于mermer_hash使用lru淘汰策略提供更极致的匹配效率同时提高prefix_cache命中率。
同时对prefix_cache进行了优化支持continuous_scheduler、chunked_scheduler和zero_evict_scheduler在prefill之后即更新
prefix_cache提高匹配时效性同时对于chunked_scheduler支持多阶段chunked_prefill匹配减少计算量并尽可能减少kv_cache占用。
## 使用方式
prefix_cache已在xLLM实现并向外暴露gflag参数控制功能的开关。
- 开启zero_evict策略并设置max_decode_token_per_sequence。
```
--enable_prefix_cache=true
```
## 性能效果
开启prefix_cache之后在Qwen3-8B模型上限制TPOT50msE2E时延 **下降10%**
!!! warning "注意"
暂不支持PD分离调度器