已下载 PDF 与标签目录
整理日期：2026-10-22。PapersPDF 当前已下载 146 份 PDF；按相同 PDF、arXiv／DOI 和核对后的正式标题合并后，对应 121 篇论文、20 项教材与技术资料。
来源：本机 ~/Desktop/PapersPDF/metadata/catalog.json 与 metadata/zotero-items.json，分类名称与当前 Zotero 集合核对。主题标签保留原集合层级，Zotero 标签保留原始文本；仅文件名的条目按 PDF 首页核对标题。
“阅读清单”链接指向所属主题中的阅读条目，阅读状态以该条目为准。PDF 链接通过 Zotero 打开本机原文件；多个 PDF 版本共用一个阅读入口。教材、课程和技术文章归入 Foundations。
| 主题 | PDF 文件 | 论文 | 教材与技术资料 |
|---|---|---|---|
| Attention | 25 | 23 | 1 |
| 推理与 KV Cache | 19 | 19 | 0 |
| 分布式训练与 MoE | 12 | 12 | 0 |
| GPU 编译与性能 | 53 | 49 | 1 |
| 模型架构与推理能力 | 11 | 11 | 0 |
| 计算基础 | 8 | 7 | 0 |
| 教材与技术参考 | 18 | 0 | 18 |
Attention
| 论文／资料 | 标签 | 阅读入口 | 原文 |
|---|---|---|---|
| A Survey of Efficient Attention Methods: Hardware-efficient, Sparse, Compact, and Linear Attention | Attention / Attention 综述 | 论文清单 | PDF / Zotero |
| Big Bird: Transformers for Longer Sequences | Attention / 稀疏 Attention；Zotero：Computer Science - Computation and Language；Computer Science - Machine Learning；Statistics - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| Block Sparse Attention with Log-Linear Complexity | Attention / 稀疏 Attention；Zotero：Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Models | Attention / 线性 Attention 与记忆 | 论文清单 | PDF / Zotero |
| DoSA: Dots Sparse Attention | Attention / 稀疏 Attention；Zotero：Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| Efficient Context Scaling with LongCat ZigZag Attention | Attention / 稀疏 Attention；Zotero：Computer Science - Artificial Intelligence；Computer Science - Computation and Language | 论文清单 | Paper / PDF / Zotero |
| Fast Weight Attention for Continual Learning | Attention / 线性 Attention 与记忆；Zotero：Computer Science - Computation and Language；Computer Science - Machine Learning；Statistics - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning | Attention / 高效 Attention 内核；Zotero：Computer Science - Machine Learning | 论文清单 | Paper / PDF / PDF2 / Zotero |
| FlashAttention-3: Fast and Accurate Attention with Asynchrony and Low-precision | Attention / 高效 Attention 内核；Zotero：Computer Science - Artificial Intelligence；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| FlashAttention-4: Algorithm and Kernel Pipelining Co-Design for Asymmetric Hardware Scaling | Attention / 高效 Attention 内核；Zotero：Computer Science - Computation and Language | 论文清单 | Paper / PDF / Zotero |
| FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness | Attention / 高效 Attention 内核；Zotero：Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| FlashDecoding++: Faster Large Language Model Inference on GPUs | Attention / 高效 Attention 内核；Zotero：Computer Science - Computation and Language；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| FSA: An Alternative Efficient Implementation of Native Sparse Attention Kernel | Attention / 稀疏 Attention；Zotero：Computer Science - Distributed, Parallel, and Cluster Computing；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| Gated Linear Attention Transformers with Hardware-Efficient Training | Attention / 线性 Attention 与记忆 | 论文清单 | PDF / Zotero |
| IndexCache: Accelerating Sparse Attention via Cross-Layer Index Reuse | Attention / 稀疏 Attention；Zotero：Computer Science - Computation and Language；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| Kimi Linear: An Expressive, Efficient Attention Architecture | Attention / 线性 Attention 与记忆；Zotero：Computer Science - Computation and Language；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| Memory Attention | Attention / 线性 Attention 与记忆；推理服务与 KV Cache；Zotero：Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| MiniMax Sparse Attention | Attention / 稀疏 Attention；Zotero：Computer Science - Artificial Intelligence | 论文清单 | Paper / PDF / Zotero |
| MoBA: Mixture of Block Attention for Long-Context LLMs | Attention / 稀疏 Attention；Zotero：Computer Science - Artificial Intelligence；Computer Science - Computation and Language；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention | Attention / 稀疏 Attention；Zotero：Computer Science - Artificial Intelligence；Computer Science - Computation and Language；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| PageWeaver: KV-Guided Query Unions for Sparse Attention | Attention / 稀疏 Attention；Zotero：Computer Science - Distributed, Parallel, and Cluster Computing；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference | Attention / 稀疏 Attention；Zotero：Computer Science - Computation and Language；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| SpargeAttention: Accurate and Training-free Sparse Attention Accelerating Any Model Inference | Attention / 稀疏 Attention；Zotero：Computer Science - Artificial Intelligence；Computer Science - Computer Vision and Pattern Recognition；Computer Science - Machine Learning；Computer Science - Performance | 论文清单 | Paper / PDF / Zotero |
| ThunderMLA: FlashMLA, Faster and Fused-er! · Hazy Research | Attention / 高效 Attention 内核 | 教材／技术资料清单 | PDF / Zotero |
推理与 KV Cache
| 论文／资料 | 标签 | 阅读入口 | 原文 |
|---|---|---|---|
| ByteTransformer: A High-Performance Transformer Boosted for Variable-Length Inputs | 推理服务与 KV Cache | 论文清单 | Paper / PDF / Zotero |
| DFlash: Block Diffusion for Flash Speculative Decoding | 推理服务与 KV Cache；Zotero：Computer Science - Computation and Language | 论文清单 | Paper / PDF / Zotero |
| EAGLE-2: Faster Inference of Language Models with Dynamic Draft Trees | 推理服务与 KV Cache；Zotero：Computer Science - Computation and Language；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| EAGLE-3: Scaling up Inference Acceleration of Large Language Models via Training-Time Test | 推理服务与 KV Cache；Zotero：Computer Science - Computation and Language | 论文清单 | Paper / PDF / Zotero |
| EAGLE: Speculative Sampling Requires Rethinking Feature Uncertainty | 推理服务与 KV Cache；Zotero：Computer Science - Computation and Language；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| Efficient Memory Management for Large Language Model Serving with PagedAttention | 推理服务与 KV Cache | 论文清单 | Paper / PDF / Zotero |
| Efficient Streaming Language Models with Attention Sinks | 推理服务与 KV Cache | 论文清单 | PDF / Zotero |
| Fast Inference from Transformers via Speculative Decoding | 推理服务与 KV Cache；Zotero：Computer Science - Computation and Language；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| FlashInfer: Efficient and Customizable Attention Engine for LLM Inference Serving | 推理服务与 KV Cache | 论文清单 | PDF / Zotero |
| HiCache: A Plug-in Scaled-Hermite Upgrade for Taylor-Style Cache-then-Forecast Diffusion Acceleration | 推理服务与 KV Cache；Zotero：Computer Science - Computer Vision and Pattern Recognition | 论文清单 | Paper / PDF / Zotero |
| HiSparse: Scaling Sparse-Attention Decoding with Hierarchical KV Cache Management | 推理服务与 KV Cache；Zotero：Distributed, Parallel, and Cluster Computing (cs.DC)；FOS: Computer and information sciences | 论文清单 | Paper / PDF / Zotero |
| Jenga: Effective Memory Management for Serving LLM with Heterogeneity | 推理服务与 KV Cache；Zotero：Computer Science - Distributed, Parallel, and Cluster Computing | 论文清单 | Paper / PDF / Zotero |
| Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads | 推理服务与 KV Cache；Zotero：Computer Science - Computation and Language；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| Mooncake: Trading More Storage for Less Computation — A KVCache-centric Architecture for Serving LLM Chatbot | 推理服务与 KV Cache | 论文清单 | PDF / Zotero |
| SGLang: Efficient Execution of Structured Language Model Programs | 推理服务与 KV Cache；Zotero：Computer Science - Artificial Intelligence；Computer Science - Programming Languages | 论文清单 | Paper / PDF / Zotero |
| ShadowKV: KV Cache in Shadows for High-Throughput Long-Context LLM Inference | 推理服务与 KV Cache；Zotero：Computer Science - Computation and Language；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| SparseEngine: Sparse-First Inference Engine | 推理服务与 KV Cache；Zotero：Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| Strata: Hierarchical Context Caching for Long Context Language Model Serving | 推理服务与 KV Cache；Zotero：Computer Science - Distributed, Parallel, and Cluster Computing | 论文清单 | Paper / PDF / Zotero |
| TensorCast: The Missing Tensor Management Layer in Large Language Model Infrastructure | 推理服务与 KV Cache；Zotero：Computer Science - Distributed, Parallel, and Cluster Computing | 论文清单 | Paper / PDF / Zotero |
分布式训练与 MoE
| 论文／资料 | 标签 | 阅读入口 | 原文 |
|---|---|---|---|
| Alpa: Automating Inter- and Intra-Operator Parallelism for Distributed Deep Learning | 分布式训练与 MoE；Zotero：Computer Science - Distributed, Parallel, and Cluster Computing；Computer Science - Machine Learning；Computer Science - Programming Languages | 论文清单 | Paper / PDF / Zotero |
| DeepSeek Elastic Compute (DSec): A Sandbox Infrastructure for Effective Agentic Training at Scale | 分布式训练与 MoE；Zotero：Computer Science - Distributed, Parallel, and Cluster Computing | 论文清单 | Paper / PDF / Zotero |
| ECO: An AI-Driven Code Efficiency Optimizer for Warehouse Scale Computers (Operational Systems) | 分布式训练与 MoE / 分布式运行时 | 论文清单 | PDF / Zotero |
| FOLDMOE: Efficient Long Sequence MoE Training via Attention-MoE Pipelining | 分布式训练与 MoE | 论文清单 | PDF / Zotero |
| Insights into DeepSeek-V3: Scaling Challenges and Reflections on Hardware for AI Architectures | 分布式训练与 MoE；Zotero：Computer Science - Artificial Intelligence；Computer Science - Distributed, Parallel, and Cluster Computing；Computer Science - Hardware Architecture | 论文清单 | Paper / PDF / Zotero |
| LoongTrain: Efficient Training of Long-Sequence LLMs with Head-Context Parallelism | 分布式训练与 MoE；Zotero：Computer Science - Distributed, Parallel, and Cluster Computing | 论文清单 | Paper / PDF / Zotero |
| MegaScale: Scaling Large Language Model Training to More Than 10,000 GPUs | 分布式训练与 MoE；Zotero：Computer Science - Distributed, Parallel, and Cluster Computing；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| PithTrain: A Compact and Agent-Native MoE Training System | 分布式训练与 MoE；Zotero：Artificial Intelligence (cs.AI)；Computation and Language (cs.CL)；Distributed, Parallel, and Cluster Computing (cs.DC)；FOS: Computer and information sciences；Machine Learning (cs.LG) | 论文清单 | Paper / PDF / Zotero |
| PockEngine: Sparse and Efficient Fine-tuning in a Pocket | 分布式训练与 MoE / 端侧训练；Zotero：Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| Ray: A Distributed Framework for Emerging AI Applications | 分布式训练与 MoE / 分布式运行时；Zotero：Computer Science - Artificial Intelligence；Computer Science - Distributed, Parallel, and Cluster Computing；Computer Science - Machine Learning；Statistics - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| Seer: Online Context Learning for Fast Synchronous LLM Reinforcement Learning | 分布式训练与 MoE / 强化学习与 Agent 训练；Zotero：Computer Science - Distributed, Parallel, and Cluster Computing；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| UniEP: Unified Expert-Parallel MegaKernel MoE for LLM Training | 分布式训练与 MoE | 论文清单 | Paper / PDF / Zotero |
GPU 编译与性能
| 论文／资料 | 标签 | 阅读入口 | 原文 |
|---|---|---|---|
| AffineGraph: A Dataflow Graph Representation for Effective DNN Hardware-Aware Algorithms | GPU 编译与性能优化 | 论文清单 | PDF / Zotero |
| AMOS: enabling automatic mapping for tensor computations on spatial accelerators with hardware abstraction | GPU 编译与性能优化 / IR 与张量化 | 论文清单 | Paper / PDF / Zotero |
| Ansor: Generating High-Performance Tensor Programs for Deep Learning | GPU 编译与性能优化 / 编译器架构与调度 | 论文清单 | PDF / Zotero |
| AStitch: enabling a new multi-dimensional optimization space for memory-intensive ML training and inference on modern SIMT architectures | GPU 编译与性能优化 / Kernel Fusion | 论文清单 | Paper / PDF / Zotero |
| Automatic Horizontal Fusion for GPU Kernels | GPU 编译与性能优化 / Kernel Fusion | 论文清单 | Paper / PDF / Zotero |
| Bolt: Bridging the Gap between Auto-tuners and Hardware-native Performance | GPU 编译与性能优化 / Kernel Fusion | 论文清单 | PDF / Zotero |
| CAKE: Compiler-Agent Co-Design for Frontier Kernel Evolution | GPU 编译与性能优化；Zotero：Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| Chimera: An Analytical Optimizing Framework for Effective Compute-intensive Operators Fusion | GPU 编译与性能优化 / Kernel Fusion | 论文清单 | Paper / PDF / Zotero |
| ClusterFusion: Expanding Operator Fusion Scope for LLM Inference via Cluster-Level Collective Primitive | GPU 编译与性能优化；Zotero：Computer Science - Artificial Intelligence；Computer Science - Distributed, Parallel, and Cluster Computing | 论文清单 | Paper / PDF / Zotero |
| COCKTAILER: Analyzing and Optimizing Dynamic Control Flow in Deep Learning | GPU 编译与性能优化 / 编译器架构与调度 | 论文清单 | PDF / Zotero |
| CODA: Rewriting Transformer Blocks as GEMM-Epilogue Programs | GPU 编译与性能优化；Zotero：Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| CoSA: Scheduling by Constrained Optimization for Spatial Accelerators | GPU 编译与性能优化 / 软硬件协同设计 | 论文清单 | Paper / PDF / Zotero |
| DISTAL: the distributed tensor algebra compiler | GPU 编译与性能优化 / 分布式编译 | 论文清单 | Paper / PDF / Zotero |
| DNNFusion: accelerating deep neural networks execution with advanced operator fusion | GPU 编译与性能优化 / 计算图优化 | 论文清单 | Paper / PDF / Zotero |
| Dual-side Sparse Tensor Core | GPU 编译与性能优化 / IR 与张量化 | 论文清单 | Paper / PDF / Zotero |
| Effectively Scheduling Computational Graphs of Deep Neural Networks toward Their Domain-Specific Accelerators | GPU 编译与性能优化 / 软硬件协同设计 | 论文清单 | PDF / Zotero |
| EINNET: Optimizing Tensor Programs with Derivation-Based Transformations | GPU 编译与性能优化 / 计算图优化 | 论文清单 | PDF / Zotero |
| Event Tensor: A Unified Abstraction for Compiling Dynamic Megakernel | GPU 编译与性能优化；Zotero：Computer Science - Distributed, Parallel, and Cluster Computing；Computer Science - Machine Learning；Computer Science - Programming Languages | 论文清单 | Paper / PDF / Zotero |
| FlashFFTConv: Efficient Convolutions for Long Sequences with Tensor Cores | GPU 编译与性能优化 / GPU 算子；Zotero：Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| Graphene: An IR for Optimized Tensor Computations on GPUs | GPU 编译与性能优化 / IR 与张量化 | 论文清单 | Paper / PDF / Zotero |
| Graphiler: A Compiler for Graph Neural Networks | GPU 编译与性能优化 / 编译器架构与调度 | 论文清单 | PDF / Zotero |
| HASCO: Towards Agile HArdware and Software CO-design for Tensor Computation | GPU 编译与性能优化 / 软硬件协同设计 | 论文清单 | Paper / PDF / Zotero |
| HIDA: A Hierarchical Dataflow Compiler for High-Level Synthesis | GPU 编译与性能优化 / 软硬件协同设计 | 论文清单 | Paper / PDF / Zotero |
| Identifying Performance Limiters | GPU 编译与性能优化 / GPU 运行时与性能工具 | 教材／技术资料清单 | PDF / Zotero |
| KPerfIR: Towards an Open and Compiler-centric Ecosystem for GPU Kernel Performance Tooling on Modern AI Workloads | GPU 编译与性能优化 / GPU 运行时与性能工具；Zotero：Distributed, Parallel, and Cluster Computing (cs.DC)；FOS: Computer and information sciences；Programming Languages (cs.PL) | 论文清单 | Paper / PDF / Zotero |
| LADDER: Enabling Efficient Low-Precision Deep Learning Computing through Hardware-aware Tensor Transformation | GPU 编译与性能优化 / IR 与张量化 | 论文清单 | PDF / Zotero |
| Mirage: A Multi-Level Superoptimizer for Tensor Programs | GPU 编译与性能优化 / GPU 运行时与性能工具 | 论文清单 | PDF / Zotero |
| mKernel: Fast Multi-GPU, Multi-Node Fused Kernels | GPU 编译与性能优化；Zotero：Computer Science - Artificial Intelligence；Computer Science - Distributed, Parallel, and Cluster Computing；Computer Science - Machine Learning；Computer Science - Networking and Internet Architecture | 论文清单 | Paper / PDF / Zotero |
| MLIR-based code generation for GPU tensor cores | GPU 编译与性能优化 / IR 与张量化 | 论文清单 | Paper / PDF / Zotero |
| MPK: A Compiler and Runtime for Mega-Kernelizing Tensor Programs | GPU 编译与性能优化 / GPU 运行时与性能工具；Zotero：Computer Science - Distributed, Parallel, and Cluster Computing；Computer Science - Machine Learning；Computer Science - Programming Languages | 论文清单 | Paper / PDF / PDF2 / Zotero |
| Open-CAKE-IR：跨硬件优化机制复用与编译器协同演进 | GPU 编译与性能优化 | 论文清单 | PDF / Zotero |
| PET: Optimizing Tensor Programs with Partially Equivalent Transformations and Automated Corrections | GPU 编译与性能优化 / 计算图优化 | 论文清单 | PDF / Zotero |
| PIT: Optimization of Dynamic Sparse Deep Learning Models via Permutation Invariant Transformation | GPU 编译与性能优化 / IR 与张量化 | 论文清单 | PDF / Zotero |
| RAMMER: Enabling Holistic Deep Learning Compiler Optimizations with rTasks | GPU 编译与性能优化 / 计算图优化 | 论文清单 | PDF / Zotero |
| ROLLER: Fast and Efficient Tensor Compilation for Deep Learning | GPU 编译与性能优化 / 编译器架构与调度 | 论文清单 | PDF / Zotero |
| SARA: Scaling a Reconfigurable Dataflow Accelerator | GPU 编译与性能优化 / 软硬件协同设计 | 论文清单 | Paper / PDF / Zotero |
| Slapo: A Schedule Language for Progressive Optimization of Large Deep Learning Model Training | GPU 编译与性能优化 / GPU 编程语言与 DSL；Zotero：Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| SparseTIR: Composable Abstractions for Sparse Compilation in Deep Learning | GPU 编译与性能优化 / IR 与张量化；Zotero：Computer Science - Artificial Intelligence；Computer Science - Machine Learning；Computer Science - Programming Languages | 论文清单 | Paper / PDF / PDF2 / PDF3 / Zotero |
| Task-Based Tensor Computations on Modern GPUs | GPU 编译与性能优化 / GPU 编程语言与 DSL；Zotero：Computer Science - Programming Languages | 论文清单 | Paper / PDF / Zotero |
| TASO: optimizing deep learning computation with automatic generation of graph substitutions | GPU 编译与性能优化 / 计算图优化 | 论文清单 | Paper / PDF / Zotero |
| Tawa: Automatic Warp Specialization for Modern GPUs with Asynchronous References | GPU 编译与性能优化 / GPU 运行时与性能工具；Zotero：Computer Science - Hardware Architecture；Computer Science - Machine Learning；Computer Science - Programming Languages | 论文清单 | Paper / PDF / Zotero |
| The Deep Learning Compiler: A Comprehensive Survey | GPU 编译与性能优化 / 编译器架构与调度 | 论文清单 | Paper / PDF / Zotero |
| The tensor algebra compiler | GPU 编译与性能优化 / 编译器架构与调度 | 论文清单 | Paper / PDF / Zotero |
| ThunderKittens: Simple, Fast, and Adorable AI Kernels | GPU 编译与性能优化 / GPU 编程语言与 DSL；Zotero：Computer Science - Artificial Intelligence；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| TileLang: A Composable Tiled Programming Model for AI Systems | GPU 编译与性能优化 / GPU 编程语言与 DSL；Zotero：Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| TileSight: A First-Principles Tile-Centric Analytical GPU Performance Model from Cores to Clusters | GPU 编译与性能优化；Zotero：Computer Science - Distributed, Parallel, and Cluster Computing；Computer Science - Performance | 论文清单 | Paper / PDF / Zotero |
| Tilus: A Virtual Machine for Arbitrary Low-Precision GPGPU Computation in LLM Serving | GPU 编译与性能优化 / GPU 编程语言与 DSL；Zotero：Computer Science - Artificial Intelligence；Computer Science - Machine Learning；Computer Science - Programming Languages | 论文清单 | Paper / PDF / Zotero |
| Tiramisu: A Polyhedral Compiler for Expressing Fast and Portable Code | GPU 编译与性能优化 / 编译器架构与调度；Zotero：Computer Science - Distributed, Parallel, and Cluster Computing；Computer Science - Mathematical Software；Computer Science - Neural and Evolutionary Computing；Computer Science - Performance；Computer Science - Programming Languages | 论文清单 | Paper / PDF / Zotero |
| WELDER: Scheduling Deep Learning Memory Access via Tile-graph | GPU 编译与性能优化 / Kernel Fusion | 论文清单 | PDF / Zotero |
| X-Stage: Modeling Post-Issue Backpressure in GPU Communication--Computation Fusion | GPU 编译与性能优化；Zotero：Computer Science - Artificial Intelligence；Computer Science - Distributed, Parallel, and Cluster Computing | 论文清单 | Paper / PDF / Zotero |
模型架构与推理能力
| 论文／资料 | 标签 | 阅读入口 | 原文 |
|---|---|---|---|
| Better & Faster Large Language Models via Multi-token Prediction | 推理服务与 KV Cache；模型架构与推理能力；Zotero：Computer Science - Computation and Language | 论文清单 | Paper / PDF / Zotero |
| DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning | 模型架构与推理能力；Zotero：Computer Science - Artificial Intelligence；Computer Science - Computation and Language；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| DeepSeek-V3 Technical Report | 模型架构与推理能力 | 论文清单 | PDF / Zotero |
| DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models | 模型架构与推理能力；Zotero：Computer Science - Computation and Language | 论文清单 | Paper / PDF / Zotero |
| DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression | 模型架构与推理能力；Zotero：Computer Science - Computation and Language | 论文清单 | Paper / PDF / Zotero |
| Forking: Sudden Overfitting Under Replay | 模型架构与推理能力；Zotero：Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| mHC: Manifold-Constrained Hyper-Connections | 模型架构与推理能力；Zotero：Computer Science - Artificial Intelligence；Computer Science - Computation and Language；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
| Qwen3 Technical Report | 模型架构与推理能力；Zotero：Computer Science - Computation and Language | 论文清单 | Paper / PDF / Zotero |
| SANA-WM: Efficient Minute-Scale World Modeling with Hybrid Linear Diffusion Transformer | 模型架构与推理能力；Zotero：Computer Science - Computer Vision and Pattern Recognition | 论文清单 | Paper / PDF / Zotero |
| Scaling Latent Reasoning via Looped Language Models | 模型架构与推理能力；Zotero：Computer Science - Computation and Language | 论文清单 | Paper / PDF / Zotero |
| Virtual Width Networks | 模型架构与推理能力；Zotero：Computer Science - Artificial Intelligence；Computer Science - Machine Learning | 论文清单 | Paper / PDF / Zotero |
计算基础
| 论文／资料 | 标签 | 阅读入口 | 原文 |
|---|---|---|---|
| A loop transformation theory and an algorithm to maximize parallelism | 计算机系统与并行基础 / 并行计算与体系结构 | 论文清单 | Paper / PDF / PDF2 / Zotero |
| A Survey on Register Allocation | 计算机系统与并行基础 / 编译原理 | 论文清单 | PDF / Zotero |
| Characterizing Off-path SmartNIC for Accelerating Distributed Systems | 计算机系统与并行基础 / 网络与 SmartNIC | 论文清单 | PDF / Zotero |
| Exploiting Memory Access Patterns to Improve Memory Performance in Data-Parallel Architectures | 计算机系统与并行基础 / 并行计算与体系结构 | 论文清单 | Paper / PDF / Zotero |
| Honeycomb: Secure and Efficient GPU Executions via Static Validation | 计算机系统与并行基础 / 虚拟化与安全 | 论文清单 | PDF / Zotero |
| Nephele: Extending Virtualization Environments for Cloning Unikernel-based VMs | 计算机系统与并行基础 / 虚拟化与安全 | 论文清单 | Paper / PDF / Zotero |
| Security and Performance in the Delegated User-level Virtualization | 计算机系统与并行基础 / 虚拟化与安全 | 论文清单 | PDF / Zotero |
教材与技术参考
| 论文／资料 | 标签 | 阅读入口 | 原文 |
|---|---|---|---|
| CMU 15-418/618 Assignment 2: A Simple CUDA Renderer | 书籍与技术资料 / 课程资料 / CMU 15-418 · 并行计算 | 教材／技术资料清单 | Paper / PDF / Zotero |
| CMU 15-418/618 Lecture 1: Why Parallelism? Why Efficiency? | 书籍与技术资料 / 课程资料 / CMU 15-418 · 并行计算 | 教材／技术资料清单 | Paper / PDF / Zotero |
| CMU 15-418/618 Lecture 2: A Modern Multi-Core Processor | 书籍与技术资料 / 课程资料 / CMU 15-418 · 并行计算 | 教材／技术资料清单 | Paper / PDF / Zotero |
| CMU 15-418/618 Lecture 3: Parallel Programming Abstractions | 书籍与技术资料 / 课程资料 / CMU 15-418 · 并行计算 | 教材／技术资料清单 | Paper / PDF / Zotero |
| CMU 15-418/618 Lecture 4: Parallel Programming Basics | 书籍与技术资料 / 课程资料 / CMU 15-418 · 并行计算 | 教材／技术资料清单 | Paper / PDF / Zotero |
| CMU 15-418/618 Lecture 5: Performance Optimization — Work Distribution and Scheduling | 书籍与技术资料 / 课程资料 / CMU 15-418 · 并行计算 | 教材／技术资料清单 | Paper / PDF / Zotero |
| CMU 15-418/618 Lecture 8: Parallel Programming Case Studies | 书籍与技术资料 / 课程资料 / CMU 15-418 · 并行计算 | 教材／技术资料清单 | Paper / PDF / Zotero |
| CMU 15-418/618: GPU Architecture & CUDA Programming | 书籍与技术资料 / 课程资料 / CMU 15-418 · 并行计算 | 教材／技术资料清单 | Paper / PDF / Zotero |
| CS336 Assignment 1: Building a Transformer LM | 书籍与技术资料 / 课程资料 / CS336 · 大模型基础 | 教材／技术资料清单 | PDF / Zotero |
| CS336 Assignment 2: Systems and Parallelism | 书籍与技术资料 / 课程资料 / CS336 · 大模型基础 | 教材／技术资料清单 | PDF / Zotero |
| CUDA by example: an introduction to general-purpose GPU programming | 书籍与技术资料 / 教材与参考书；Zotero：Application software；Computer architecture；Development；Graphics processing units；Parallel programming (Computer science)；Programming | 教材／技术资料清单 | PDF / Zotero |
| CUDA C++ Programming Guide | 书籍与技术资料 / 课程资料 / CMU 15-418 · 并行计算 | 教材／技术资料清单 | PDF / Zotero |
| Introduction to Algorithms, Fourth Edition | 书籍与技术资料 / 教材与参考书 | 教材／技术资料清单 | PDF / Zotero |
| Introduction to CUDA Programming | 书籍与技术资料 / 课程资料 / CMU 15-418 · 并行计算 | 教材／技术资料清单 | PDF / Zotero |
| Professional Cuda C programming | 书籍与技术资料 / 教材与参考书；Zotero：Parallel programming (Computer science) | 教材／技术资料清单 | PDF / Zotero |
| 深入理解 AI Agent：设计原理与工程实践 | 书籍与技术资料 | 教材／技术资料清单 | PDF / Zotero |
| 深入理解 AI Infra：量化分析与系统设计 | 书籍与技术资料 | 教材／技术资料清单 | PDF / Zotero |
| 编译原理（第二版） | 书籍与技术资料 / 教材与参考书 | 教材／技术资料清单 | PDF / Zotero |
归档源文件（146 份）
路径相对于 ~/Desktop/PapersPDF/，保留归档文件名。
01_Attention/00_Surveys/4J28YBFT-A Survey of Efficient Attention Methods_ Hardware-efficient, Sparse, Compact, and Linear Attention.pdf· PDF01_Attention/01_Sparse_Attention/2606.13392v2.pdf· PDF01_Attention/01_Sparse_Attention/2VJX3AWK-Efficient Context Scaling with LongCat ZigZag Attention.pdf· PDF01_Attention/01_Sparse_Attention/AW9DQY8W-Quest_ Query-Aware Sparsity for Efficient Long-Context LLM Inference.pdf· PDF01_Attention/01_Sparse_Attention/Block Sparse Attention with Log-Linear Complexity.pdf· PDF01_Attention/01_Sparse_Attention/DOSA- DOTS SPARSE ATTENTION.pdf· PDF01_Attention/01_Sparse_Attention/FSA- AN ALTERNATIVE EFFICIENT IMPLEMENTATION OF NATIVE SPARSE ATTENTION KERNEL.pdf· PDF01_Attention/01_Sparse_Attention/IndexCache.pdf· PDF01_Attention/01_Sparse_Attention/LIRH3V9J-Big Bird_ Transformers for Longer Sequences.pdf· PDF01_Attention/01_Sparse_Attention/Native Sparse Attention- Hardware-Aligned and Natively Trainable Sparse Attention.pdf· PDF01_Attention/01_Sparse_Attention/P4ZJ8KM9-SpargeAttention_ Accurate and Training-free Sparse Attention Accelerating Any Model Inference.pdf· PDF01_Attention/01_Sparse_Attention/PA5MCZGA-MoBA_ Mixture of Block Attention for Long-Context LLMs.pdf· PDF01_Attention/01_Sparse_Attention/PAGEWEAVER- KV-GUIDED QUERY UNIONS FOR SPARSE ATTENTION.pdf· PDF01_Attention/02_Efficient_Attention_Kernels/8QQKNGYR-FlashDecoding++_ Faster Large Language Model Inference on GPUs.pdf· PDF01_Attention/02_Efficient_Attention_Kernels/FlashAttention-4.pdf· PDF01_Attention/02_Efficient_Attention_Kernels/PP6LRECT-FlashAttention_ Fast and Memory-Efficient Exact Attention with IO-Awareness.pdf· PDF01_Attention/02_Efficient_Attention_Kernels/PV68CNSP-FlashAttention-2_ Faster Attention with Better Parallelism and Work Partitioning.pdf· PDF01_Attention/02_Efficient_Attention_Kernels/QL7BJ5SD-Faster Attention with Better Parallelism and Work Partitioning.pdf· PDF01_Attention/02_Efficient_Attention_Kernels/ThunderMLA_ FlashMLA, Faster and Fused-er! · Hazy Research.pdf· PDF01_Attention/02_Efficient_Attention_Kernels/flashattention-3.pdf· PDF01_Attention/03_Linear_Attention_and_Memory/2609.28399v2-Memory Attention.pdf· PDF01_Attention/03_Linear_Attention_and_Memory/2RFQ4ZTQ-Conditional Memory via Scalable Lookup_ A New Axis of Sparsity for Large Language Models.pdf· PDF01_Attention/03_Linear_Attention_and_Memory/3GQRHRJX-Kimi Linear_ An Expressive, Efficient Attention Architecture.pdf· PDF01_Attention/03_Linear_Attention_and_Memory/Fast Weight Attention for Continual Learning-with-annotations.pdf· PDF01_Attention/03_Linear_Attention_and_Memory/QZ5TMHJT-Gated Linear Attention Transformers with Hardware-Efficient Training.pdf· PDF02_Inference_and_KV_Cache/2211.17192v2-Fast Inference from Transformers via Speculative Decoding.pdf· PDF02_Inference_and_KV_Cache/2401.10774v3-M EDUSA_ Simple LLM Inference Acceleration Framework with Multiple.pdf· PDF02_Inference_and_KV_Cache/2401.15077v3-EAGLE_ Speculative Sampling Requires Rethinking Feature Uncertainty Yuhui Li♠.pdf· PDF02_Inference_and_KV_Cache/2406.16858v2-EAGLE-2_ Faster Inference of Language Models with Dynamic Draft Trees Yuhui Li♠.pdf· PDF02_Inference_and_KV_Cache/2503.01840v3-EAGLE-3_ Scaling up Inference Acceleration of Large Language Models via Training-Time Test.pdf· PDF02_Inference_and_KV_Cache/2503.18292v1-J ENGA_ Effective Memory Management for Serving LLM with Heterogeneity.pdf· PDF02_Inference_and_KV_Cache/2508.16984v2-HiCache_ A Plug-in Scaled-Hermite Upgrade for Taylor-Style Cache-then-Forecast Diffusion Acceleration.pdf· PDF02_Inference_and_KV_Cache/2508.18572v1-Strata_ Hierarchical Context Caching for Long.pdf· PDF02_Inference_and_KV_Cache/9ICECIEG-Efficient Memory Management for Large Language Model Serving with PagedAttention.pdf· PDF02_Inference_and_KV_Cache/CUXI564S-streamingllm.pdf· PDF02_Inference_and_KV_Cache/DFlash.pdf· PDF02_Inference_and_KV_Cache/HiSparse.pdf· PDF02_Inference_and_KV_Cache/KCJRN274-2501.01005v1.pdf· PDF02_Inference_and_KV_Cache/MoonCake.pdf· PDF02_Inference_and_KV_Cache/QLN9WPPQ-SGLang_ Efficient Execution of Structured Language Model Programs.pdf· PDF02_Inference_and_KV_Cache/Shadow-KV.pdf· PDF02_Inference_and_KV_Cache/SparseEngine- Sparse-First Inference Engine.pdf· PDF02_Inference_and_KV_Cache/XTL5JC62-ByteTransformer_ A High-Performance Transformer Boosted for Variable-Length Inputs.pdf· PDF02_Inference_and_KV_Cache/tensorcast.pdf· PDF03_Distributed_Training_and_MoE/01_Distributed_Runtime/ECO_ An AI-Driven Code Efficiency Optimizer for.pdf· PDF03_Distributed_Training_and_MoE/01_Distributed_Runtime/JYZPKSY8-Ray_ A Distributed Framework for Emerging AI Applications.pdf· PDF03_Distributed_Training_and_MoE/02_RL_and_Agent_Training/YKUNW5YJ-Seer_ Online Context Learning for Fast Synchronous LLM Reinforcement Learning.pdf· PDF03_Distributed_Training_and_MoE/03_On_Device_Training/B37PRC33-PockEngine_ Sparse and Efficient Fine-tuning in a Pocket.pdf· PDF03_Distributed_Training_and_MoE/ANUU3GWJ-MegaScale_ Scaling Large Language Model Training to More Than 10,000 GPUs.pdf· PDF03_Distributed_Training_and_MoE/DeepSeek Elastic Compute (DSec)- A Sandbox Infrastructure for Effective Agentic Training at Scale.pdf· PDF03_Distributed_Training_and_MoE/FoldMoE.pdf· PDF03_Distributed_Training_and_MoE/IGB2EYWT-Alpa_ Automating Inter- and Intra-Operator Parallelism for Distributed Deep Learning.pdf· PDF03_Distributed_Training_and_MoE/LoongTrain.pdf· PDF03_Distributed_Training_and_MoE/UniEP.pdf· PDF03_Distributed_Training_and_MoE/XPT4R2PX-Insights into DeepSeek-V3_ Scaling Challenges and Reflections on Hardware for AI Architectures.pdf· PDF03_Distributed_Training_and_MoE/pith-train.pdf· PDF04_GPU_Compiler_and_Performance/01_DSL_and_Programming_Models/2U3YQMCK-TileLang_ A Composable Tiled Programming Model for AI Systems.pdf· PDF04_GPU_Compiler_and_Performance/01_DSL_and_Programming_Models/3DBN55V2-Slapo_ A Schedule Language for Progressive Optimization of Large Deep Learning Model Training.pdf· PDF04_GPU_Compiler_and_Performance/01_DSL_and_Programming_Models/RAQT8GSR-Task-Based Tensor Computations on Modern GPUs.pdf· PDF04_GPU_Compiler_and_Performance/01_DSL_and_Programming_Models/SSZK6BP2-ThunderKittens_ Simple, Fast, and Adorable AI Kernels.pdf· PDF04_GPU_Compiler_and_Performance/01_DSL_and_Programming_Models/ZUCBLEW5-Tilus_ A Virtual Machine for Arbitrary Low-Precision GPGPU Computation in LLM Serving.pdf· PDF04_GPU_Compiler_and_Performance/02_Compiler_Architecture_and_Scheduling/A4GT2NTU-Ansor_ Generating High-Performance Tensor Programs for Deep Learning.pdf· PDF04_GPU_Compiler_and_Performance/02_Compiler_Architecture_and_Scheduling/IJTC8RXW-COCKTAILER_ Analyzing and Optimizing Dynamic Control Flow in Deep Learning.pdf· PDF04_GPU_Compiler_and_Performance/02_Compiler_Architecture_and_Scheduling/IYTUH2XV-Graphiler_ A Compiler for Graph Neural Networks.pdf· PDF04_GPU_Compiler_and_Performance/02_Compiler_Architecture_and_Scheduling/M6EXK7WY-The Deep Learning Compiler_ A Comprehensive Survey.pdf· PDF04_GPU_Compiler_and_Performance/02_Compiler_Architecture_and_Scheduling/RTVMUM4F-Tiramisu_ A Polyhedral Compiler for Expressing Fast and Portable Code.pdf· PDF04_GPU_Compiler_and_Performance/02_Compiler_Architecture_and_Scheduling/WZ72K59S-osdi22-zhu.pdf· PDF04_GPU_Compiler_and_Performance/02_Compiler_Architecture_and_Scheduling/ZVZ6LMVF-The tensor algebra compiler.pdf· PDF04_GPU_Compiler_and_Performance/03_Kernel_Fusion/2TRNJBSP-AStitch_ enabling a new multi-dimensional optimization space for memory-intensive ML training and inferen.pdf· PDF04_GPU_Compiler_and_Performance/03_Kernel_Fusion/6ZV83ISM-WELDER_ Scheduling Deep Learning Memory Access via Tile-graph.pdf· PDF04_GPU_Compiler_and_Performance/03_Kernel_Fusion/DBCSL5U7-Automatic Horizontal Fusion for GPU Kernels.pdf· PDF04_GPU_Compiler_and_Performance/03_Kernel_Fusion/ELTDBRRZ-Bolt_ Bridging the Gap between Auto-tuners and Hardware-native Performance.pdf· PDF04_GPU_Compiler_and_Performance/03_Kernel_Fusion/YHXE9DXB-Chimera_ An Analytical Optimizing Framework for Effective Compute-intensive Operators Fusion.pdf· PDF04_GPU_Compiler_and_Performance/04_Graph_Optimization/637ESUF5-RAMMER_ Enabling Holistic Deep Learning Compiler Optimizations with rTasks.pdf· PDF04_GPU_Compiler_and_Performance/04_Graph_Optimization/6EEYF2XM-DNNFusion_ accelerating deep neural networks execution with advanced operator fusion.pdf· PDF04_GPU_Compiler_and_Performance/04_Graph_Optimization/B8UQIZHU-PET_ Optimizing Tensor Programs with Partially Equivalent Transformations and Automated Corrections.pdf· PDF04_GPU_Compiler_and_Performance/04_Graph_Optimization/C8455AUK-EINNET_ Optimizing Tensor Programs with Derivation-Based Transformations.pdf· PDF04_GPU_Compiler_and_Performance/04_Graph_Optimization/S64TB9IQ-TASO_ optimizing deep learning computation with automatic generation of graph substitutions.pdf· PDF04_GPU_Compiler_and_Performance/05_Hardware_Software_Co_Design/3MIDF5NM-HIDA_ A Hierarchical Dataflow Compiler for High-Level Synthesis.pdf· PDF04_GPU_Compiler_and_Performance/05_Hardware_Software_Co_Design/7W5N7ECJ-Effectively Scheduling Computational Graphs of Deep Neural Networks toward Their Domain-Specific Accelerat.pdf· PDF04_GPU_Compiler_and_Performance/05_Hardware_Software_Co_Design/9B35YJJ3-CoSA_ Scheduling by Constrained Optimization for Spatial Accelerators.pdf· PDF04_GPU_Compiler_and_Performance/05_Hardware_Software_Co_Design/CZXLU8MM-HASCO_ Towards Agile HArdware and Software CO-design for Tensor Computation.pdf· PDF04_GPU_Compiler_and_Performance/05_Hardware_Software_Co_Design/U55E5DCP-SARA_ Scaling a Reconfigurable Dataflow Accelerator.pdf· PDF04_GPU_Compiler_and_Performance/06_IR_and_Tensorization/3PFGF8PF-PIT_ Optimization of Dynamic Sparse Deep Learning Models via Permutation Invariant Transformation.pdf· PDF04_GPU_Compiler_and_Performance/06_IR_and_Tensorization/4F4IIX46-SparseTIR_ Composable Abstractions for Sparse Compilation in Deep Learning.pdf· PDF04_GPU_Compiler_and_Performance/06_IR_and_Tensorization/4Z3T48EM-LADDER_ Enabling Efficient Low-Precision Deep Learning Computing through Hardware-aware Tensor Transforma.pdf· PDF04_GPU_Compiler_and_Performance/06_IR_and_Tensorization/8L7I4TM7-MLIR-based code generation for GPU tensor cores.pdf· PDF04_GPU_Compiler_and_Performance/06_IR_and_Tensorization/8QAZ5LTU-[ASPLOS2023]SparseTIR.pdf· PDF04_GPU_Compiler_and_Performance/06_IR_and_Tensorization/9GGQREBY-SparseTIR_ Composable Abstractions for Sparse Compilation in Deep Learning.pdf· PDF04_GPU_Compiler_and_Performance/06_IR_and_Tensorization/B4QA8WLK-Graphene_ An IR for Optimized Tensor Computations on GPUs.pdf· PDF04_GPU_Compiler_and_Performance/06_IR_and_Tensorization/LYZ6VBHY-Dual-side Sparse Tensor Core.pdf· PDF04_GPU_Compiler_and_Performance/06_IR_and_Tensorization/M5MVQC36-AMOS_ enabling automatic mapping for tensor computations on spatial accelerators with hardware abstractio.pdf· PDF04_GPU_Compiler_and_Performance/07_Distributed_Compilation/AMVNYVPH-DISTAL_ the distributed tensor algebra compiler.pdf· PDF04_GPU_Compiler_and_Performance/08_Runtime_and_Performance_Tools/3PFQFIXR-Mirage_ A Multi-Level Superoptimizer for Tensor Programs.pdf· PDF04_GPU_Compiler_and_Performance/08_Runtime_and_Performance_Tools/7EYAKB7L-Identifying Performance Limiters.pdf· PDF04_GPU_Compiler_and_Performance/08_Runtime_and_Performance_Tools/9D4CNY6C-Mirage Persistent Kernel_ A Compiler and Runtime for Mega-Kernelizing Tensor Programs.pdf· PDF04_GPU_Compiler_and_Performance/08_Runtime_and_Performance_Tools/9YVGSWWH-KPerfIR_ Towards an Open and Compiler-centric Ecosystem for GPU Kernel Performance Tooling on Modern AI W.pdf· PDF04_GPU_Compiler_and_Performance/08_Runtime_and_Performance_Tools/UM2GUBDY-Tawa_ Automatic Warp Specialization for Modern GPUs with Asynchronous References.pdf· PDF04_GPU_Compiler_and_Performance/09_GPU_Operators/QDPRZPX2-FlashFFTConv_ Efficient Convolutions for Long Sequences with Tensor Cores.pdf· PDF04_GPU_Compiler_and_Performance/AffineGraph.pdf· PDF04_GPU_Compiler_and_Performance/CAKE.pdf· PDF04_GPU_Compiler_and_Performance/CODA.pdf· PDF04_GPU_Compiler_and_Performance/ClusterFusion- Expanding Operator Fusion Scope for LLM Inference via Cluster-Level Collective Primitive.pdf· PDF04_GPU_Compiler_and_Performance/event-tensor.pdf· PDF04_GPU_Compiler_and_Performance/mKernel- Fast Multi-GPU, Multi-Node Fused Kernels.pdf· PDF04_GPU_Compiler_and_Performance/open-cake-ir-technical-report.pdf· PDF04_GPU_Compiler_and_Performance/osdi26-cheng-MPK.pdf· PDF04_GPU_Compiler_and_Performance/tilesight.pdf· PDF04_GPU_Compiler_and_Performance/x-stage.pdf· PDF05_Model_Architecture_and_Reasoning/2404.19737v1-Better & Faster Large Language Models via Multi-token Prediction.pdf· PDF05_Model_Architecture_and_Reasoning/466NDG9P-DeepSeek-R1_ Incentivizing Reasoning Capability in LLMs via Reinforcement Learning.pdf· PDF05_Model_Architecture_and_Reasoning/8IZQ6MZA-DeepSeek-V3.2_ Pushing the Frontier of Open Large Language Models.pdf· PDF05_Model_Architecture_and_Reasoning/D664IKR9-Qwen3 Technical Report.pdf· PDF05_Model_Architecture_and_Reasoning/D6RIU87T-DeepSeek_V3.pdf· PDF05_Model_Architecture_and_Reasoning/DeepSeek-V4.1-Flash- Pushing the Limits of KV Cache Compression.pdf· PDF05_Model_Architecture_and_Reasoning/Forking- Sudden Overfitting Under Replay.pdf· PDF05_Model_Architecture_and_Reasoning/G5UP6HY5-mHC_ Manifold-Constrained Hyper-Connections.pdf· PDF05_Model_Architecture_and_Reasoning/SANA-WM- Efficient Minute-Scale World Modeling with Hybrid Linear Diffusion Transformer.pdf· PDF05_Model_Architecture_and_Reasoning/Scaling Latent Reasoning via Looped Language Models.pdf· PDF05_Model_Architecture_and_Reasoning/virtual-width-network.pdf· PDF06_Computing_Foundations/01_Parallelism_and_Architecture/A Loop Transformation Theory and an Algorithm to Maximize Parallelism.pdf· PDF06_Computing_Foundations/01_Parallelism_and_Architecture/G7G54FHS-Exploiting Memory Access Patterns to Improve Memory Performance in Data-Parallel Architectures.pdf· PDF06_Computing_Foundations/01_Parallelism_and_Architecture/R3PQDJEZ-A loop transformation theory and an algorithm to maximize parallelism.pdf· PDF06_Computing_Foundations/02_Compiler_Foundations/9WWM8C8A-A Survey on Register Allocation.pdf· PDF06_Computing_Foundations/03_Virtualization_and_Security/3IE2M56Z-Honeycomb_ Secure and Efficient GPU Executions via Static Validation.pdf· PDF06_Computing_Foundations/03_Virtualization_and_Security/4FAXKU7Z-Security and Performance in the Delegated User-level Virtualization.pdf· PDF06_Computing_Foundations/03_Virtualization_and_Security/LSGBB2YL-Nephele_ Extending Virtualization Environments for Cloning Unikernel-based VMs.pdf· PDF06_Computing_Foundations/04_Networking/RTBRCBWY-Characterizing Off-path SmartNIC for Accelerating Distributed Systems.pdf· PDF07_Books_and_Technical_References/01_Books/5S4FSPFX-Professional Cuda® C Programming.pdf· PDF07_Books_and_Technical_References/01_Books/BQFJTKME-Introduction to Algorithms, fourth edition (Tho... (z-lib.org).pdf· PDF07_Books_and_Technical_References/01_Books/IXB5S4XQ-编译原理 第二版-6.pdf· PDF07_Books_and_Technical_References/01_Books/SUQKUXL8-CUDA by example_ an introduction to general-purpose GPU programming.pdf· PDF07_Books_and_Technical_References/02_Courses/CMU15418/54SQR92R-05_progperf1.pdf· PDF07_Books_and_Technical_References/02_Courses/CMU15418/9CCPD6WX-06_gpuarch.pdf· PDF07_Books_and_Technical_References/02_Courses/CMU15418/DM3FZJZJ-08_casestudies.pdf· PDF07_Books_and_Technical_References/02_Courses/CMU15418/G6EJBPIF-asst2.pdf· PDF07_Books_and_Technical_References/02_Courses/CMU15418/GBQTWU8D-CUDA C++ Programming Guide.pdf· PDF07_Books_and_Technical_References/02_Courses/CMU15418/GJKKSHXL-04_progbasics.pdf· PDF07_Books_and_Technical_References/02_Courses/CMU15418/QE42E4DH-01_whyparallelism.pdf· PDF07_Books_and_Technical_References/02_Courses/CMU15418/UI4J9J7B-Introduction to CUDA Programming.pdf· PDF07_Books_and_Technical_References/02_Courses/CMU15418/W5PU2L6H-03_progmodels.pdf· PDF07_Books_and_Technical_References/02_Courses/CMU15418/YUQG9U4G-02_basicarch.pdf· PDF07_Books_and_Technical_References/02_Courses/CS336/ACNW7CEX-cs336_spring2025_assignment2_systems.pdf· PDF07_Books_and_Technical_References/02_Courses/CS336/N9VEMX6Z-cs336_spring2025_assignment1_basics.pdf· PDF07_Books_and_Technical_References/AI-Agents-in-Depth-zh-CN.pdf· PDF07_Books_and_Technical_References/AI-Infra-Book.pdf· PDF