01
Engram：TP 分片与同机 DP 共享的源码研究
ai-systems / llm-inference
llm-inferencevllmarchitectureprofiling
+1
02
P/D 分离：从 KV 交接到路由、背压与 SLO
ai-systems / llm-inference
llm-inferenceprefilldecodekv-cache
+2
03
分层 KV Cache：从 GPU 命中到外部存储复用
ai-systems / llm-inference
llm-inferencekv-cacheprefillnetworks
+1
04
Engram 预取与 NUMA 放置：从共享页到关键路径
ai-systems / llm-inference
llm-inferencevllmarchitectureprofiling
+1
05
GLM-5.2 量化执行图：W4A8、BMM 与 KV8
ai-systems / llm-inference
llm-inferencequantizationfp4fp8
+2
06
量化方法与评测：从 PTQ/QAT 到可复现实验
ai-systems / llm-inference
llm-inferencequantizationprofilingresearch
07
从 Profiling 到 Simulation：推理性能分析的证据链方法
ai-systems / profiling
profilingsimulationllm-inferenceperformance-analysis
+1
08
GDN 与 Chunked Prefill：为什么 prepare_chunk_indices 会出现在 trace 里
ai-systems / llm-inference
llm-inferencegdnqwen3nextchunked-prefill
+3
09
AWP Profiling API
toolbox
profilingapigpucpu
+2
10
Agentic Infra：LLM 推理性能优化与 GPU 利用率提升
ai-systems / llm-inference
llm-inferencegpu-optimizationprofilingawp
+4
11
Agentic AWP：规模化 Profiling 驱动的 GPU 效率 Breakdown 与能力体系
ai-systems / profiling
awpgpu-profilingbreakdowngpu-efficiency
+3
12
GPU Trace 时间分解与通信计算重叠分析
ai-systems / profiling
GPUProfilingPerformanceDistributed Training
13
Cprof C++ Profiling 核心技术
ai-systems / profiling
profilingperfDWARFeBPF
+2
14
HTA 算法原理与实现
ai-systems / profiling
profilingpytorchgpudistributed-training
+2
15
Critical Path of AI Trace
ai-systems / profiling
AITraceCritical PathGPU
+1