01
从 Profiling 到 Simulation：推理性能分析的证据链方法
ai-systems / profiling
profilingsimulationllm-inferenceperformance-analysis
+1
02
MoE 推理：Expert 并行、显存与调度机制
ai-systems / llm-inference
moeexpert-parallelisminference
03
Prefill Trace：Worker 供给、DSA/MLA 与 Chunked Prefill
ai-systems / llm-inference
llm-inferenceprefilltracemla
+2
04
GDN 与 Chunked Prefill：为什么 prepare_chunk_indices 会出现在 trace 里
ai-systems / llm-inference
llm-inferencegdnqwen3nextchunked-prefill
+3
05
DeepSeek MLA：低秩 KV Cache 与推理效率
ai-systems / llm-inference
llm-inferenceattentionkv-cachedeepseek
+1
06
Chunked Prefill 深入分析：调度、Chunk Size 与 Attention 形状
ai-systems / llm-inference
llm-inferencechunked-prefillschedulingprefill
+3
07
Token Flow 与 Hidden State：从 Attention 到 LM Head
ai-systems / llm-inference
llm-inferencetransformerhidden-stateattention
+2
08
DeepSeek-V3 Technical Report：中英对照解读
ai-systems / llm-inference
LLMDeepSeekMoEFP8
+2
09
DSpark 与 MTP：DeepSeek 投机解码调研
ai-systems / llm-inference
llm-inferencespeculative-decodingdeepseekdspark
+1
10
Causal Attention：为什么 KV hit 后 Attention 按 1 - h² 缩放
ai-systems / llm-inference
llm-inferenceattentionkv-cachesimulator
+1
11
KV Cache Hit Ratio 修正模型：从直觉到统一公式
ai-systems / llm-inference
llm-inferencekv-cachesimulatorprefill
+1
12
模拟器建模指南：显存与吞吐公式
ai-systems / llm-inference
simulatormemory-modelinginference
13
FT vs VLLM vs SGLang 推理框架对比摘要
ai-systems / profiling
profilinginferencertp-llmvllm
+4
14
LLM 推理系统全栈综合分析
ai-systems / llm-inference
llm-inferencesynthesissystemsoptimization
15
Agentic Infra：LLM 推理性能优化与 GPU 利用率提升
ai-systems / llm-inference
llm-inferencegpu-optimizationprofilingawp
+5
16
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
17
02. Reasoning Model、Agent 与长任务
ai-systems / reasoning
AILLMReasoningAgent
18
03. RAG、Memory、Fine-tuning 与 Distillation
ai-systems / reasoning
AILLMRAGFine-Tuning
+1
19
01. 什么是 AI 推理
ai-systems / reasoning
AILLMReasoning
20
AWP 六维 Breakdown 框架与能力体系摘要
ai-systems / profiling
awpgpu-profilingbreakdowngpu-efficiency
+2
21
LLM 推理性能优化与 GPU 利用率提升摘要
ai-systems / profiling
llm-inferencegpu-optimizationprofilingawp
+3
22
CUDA Agent
ai-systems / gpu-computing
GPUCUDARLLLM
+1
23
KV Cache：推理性能的命根子
ai-systems / llm-inference
LLMInferenceKV CachePagedAttention
+2
24
Compute-bound vs Memory-bound：推理的两大瓶颈
ai-systems / llm-inference
LLMInferencePerformanceGPU
+3
25
量化：INT8 / INT4 / FP8 到底在干嘛
ai-systems / llm-inference
LLMInferenceQuantizationGPTQ
+4
26
投机解码：突破 decode 一次只出一个 token 的限制
ai-systems / llm-inference
LLMInferenceSpeculative DecodingEAGLE
+2
27
批处理与调度：推理服务的灵魂
ai-systems / llm-inference
LLMInferenceBatchingScheduling
+3
28
推理引擎架构：vLLM / TensorRT-LLM / SGLang
ai-systems / llm-inference
LLMInferencevLLMTensorRT-LLM
+3