📚 调研资料全索引

📑 顶层报告

• 📄 report

📁 decode-memory-bottleneck/

• 📝 google_arxiv_designspace.txt
• 📝 google_memory_semantic_pd.txt
• 📝 google_mori_blog_exact.txt
• 📝 google_pd_kvpool_sizing.txt
• 📝 google_sglang_decode_memory.txt
• 🐍 google_vllm_blog_mori_url.txt
• 🐍 google_vllm_decode_hbm.txt
• 📑 report
• 📝 sglang_23022_shm_pd.txt
• 📝 sglang_24523_pd_input_cap.txt
• 📝 sglang_30010_decode_deadlock.txt
• 📝 sglang_31252_hicache.txt
• 📝 sglang_discussions.txt
• 📝 sglang_issue_search1.txt
• 🐍 vllm_43470_bottleneck_expansion.txt
• 🐍 vllm_46107_morio_rfc.txt
• 🐍 vllm_blog_disagg_tag.txt
• 🐍 vllm_issue_search1.txt

📁 industry-quotes/

• 📝 references.md

📁 longcontext-qual/

• 📝 arxiv_3dls.txt
• 📝 arxiv_cxl_1m.txt
• 📝 arxiv_dualpool.txt
• 📝 arxiv_hetero.txt
• 📝 arxiv_lynx.txt
• 📝 arxiv_mooncake.txt
• 📝 arxiv_p4a.txt
• 📝 arxiv_q1.txt
• 📝 arxiv_q10.txt
• 📝 arxiv_q11.txt
• 📝 arxiv_q12.txt
• 📝 arxiv_q13.txt
• 📝 arxiv_q14.txt
• 📝 arxiv_q2.txt
• 📝 arxiv_q3.txt
• 📝 arxiv_q8.txt
• 📝 arxiv_q9.txt
• 📝 arxiv_q_hicache.txt
• 📝 arxiv_q_laps.txt
• 📝 arxiv_q_mooncake.txt
• 📝 arxiv_q_router.txt
• 📝 arxiv_sac.txt
• 📝 google_q4.txt
• 📝 google_q5.txt
• 📝 google_q6.txt
• 📝 google_q7.txt
• 📝 mooncake_html.txt
• 📝 mooncake_html2.txt
• 📑 report_longcontext_pd_challenge
• 📝 sglang_issues.txt
• 📝 sglang_pd_docs.txt
• 📝 sglang_pp_longctx.txt
• 📝 sglang_pp_pd_pr.txt
• 🐍 vllm_issue_11286.txt

📁 vllm-decode-memory/

• 📝 arch_overview.md
• 📝 cache_config.py
• 📝 config.py
• 📝 configuration.md
• 📝 conserving_memory.md
• 📝 disagg_prefill.md
• 📝 engine_args.md
• 📝 gpu_model_runner.py
• 📝 gpu_worker.py
• 📝 hybrid_kv_cache_manager.md
• 📑 index
• 📝 kv_cache_manager.md
• 📝 kv_offloading_usage.md
• 📝 nixl_kv_cache_lease.md
• 📝 nixl_kv_push_connector.md
• 📝 online_serving_readme.md
• 📝 optimization.md
• 📝 paged_attention.md
• 📝 prefix_caching.md
• 📝 serving_opt.md
• 🐍 vllm_profiler_layerwise_profile.py
• 🐍 vllm_utils_mem_constants.py
• 🐍 vllm_utils_mem_utils.py
• 🐍 vllm_v1_core_block_pool.py
• 🐍 vllm_v1_core_kv_cache_manager.py
• 🐍 vllm_v1_core_kv_cache_utils.py