GitHub 历史论文笔记归档
来源:KuangjuX/Papers-and-Learning-Notes 的历史 issues,共 34 个论文 issue。
包括 12 篇成段笔记、2 篇未完成摘记和 1 个外部笔记入口。论文阅读状态以主题清单为准。
历史笔记的性能结论和“当前系统”均以发表时的版本与实验背景为限。
补充:Unikraft 的外部文章现已完整本地化,原 issue 来源仍保留。
归档清单
| Issue | 论文 | 主题 | 材料与归档入口 |
|---|---|---|---|
| #1 | Nephele: Extending Virtualization Environments for Cloning Unikernel-based VMs | 计算基础 | 已有本地笔记;issue 本身仅有来源链接 |
| #2 | Theseus: an Experiment in Operating System Structure and State Management | 计算基础 | 论文链接;无笔记正文 |
| #3 | XPC: Architectural Support for Secure and Efficient Cross Process Call | 计算基础 | 论文链接;无笔记正文 |
| #4 | Fast Distributed Deep Learning over RDMA | 分布式训练与 MoE | 论文链接;无笔记正文 |
| #5 | RedLeaf: Isolation and Communication in a Safe Operating System | 计算基础 | 已有本地笔记;issue 本身仅有来源链接 |
| #6 | The Demikernel Datapath OS Architecture for Microsecond-scale Datacenter Systems | 计算基础 | 已有本地笔记;issue 本身仅有来源链接 |
| #7 | Every Walk’s a Hit: Making Page Walks Single-Access Cache Hits | 计算基础 | 论文链接;无笔记正文 |
| #8 | The benefits and costs of writing a POSIX kernel in a high-level language | 计算基础 | 论文链接;无笔记正文 |
| #9 | Unikraft: Fast, Specialized Unikernels the Easy Way | 计算基础 | 本地笔记与来源 |
| #10 | Alpa: Automating Inter- and Intra-Operator Parallelism for Distributed Deep Learning | 分布式训练与 MoE | 论文链接;无笔记正文 |
| #11 | Ray: A Distributed Framework for Emerging AI Applications | 分布式训练与 MoE | 论文链接;无笔记正文 |
| #12 | SkyBridge: Fast and Secure Inter-Process Communication for Microkernels | 计算基础 | 论文链接;无笔记正文 |
| #13 | Firecracker: Lightweight Virtualization for Serverless Applications | 计算基础 | 论文链接;无笔记正文 |
| #14 | Keystone: an open framework for architecting trusted execution environments | 计算基础 | 论文链接;无笔记正文 |
| #15 | The Deep Learning Compiler: A Comprehensive Survey | GPU 编译与性能 | 已有本地笔记;issue 本身仅有来源链接 |
| #16 | MLIR: Scaling Compiler Infrastructure for Domain Specific Computation | GPU 编译与性能 | 已有本地笔记;issue 本身仅有来源链接 |
| #17 | Microsecond-scale Preemption for Concurrent GPU-accelerated DNN Inferences | 推理与 KV Cache | 摘要摘记,未完成 |
| #18 | Fast Segment Anything | 模型架构与推理能力 | 论文链接;无笔记正文 |
| #19 | System Virtualization for Neural Processing Units | 计算基础 | 论文链接;无笔记正文 |
| #20 | Pushing the Limits of Machine Design: Automated CPU Design with AI | 计算基础 | 论文链接;无笔记正文 |
| #21 | Cocktailer: Analyzing and Optimizing Dynamic Control Flow in Deep Learning | GPU 编译与性能 | 历史笔记 |
| #22 | Rammer: Enabling Holistic Deep Learning Compiler Optimizations with rTasks | GPU 编译与性能 | 历史笔记 |
| #23 | The Tensor Algebra Compiler | GPU 编译与性能 | 无笔记正文;唯一评论来自他人,仅询问 data layout 论文推荐,不作为个人阅读笔记导入 |
| #24 | Roller: Fast and Efficient Tensor Compilation for Deep Learning | GPU 编译与性能 | 历史笔记 |
| #25 | Welder: Scheduling Deep Learning Memory Access via Tile-graph | GPU 编译与性能 | 历史笔记 |
| #26 | 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 | 历史笔记 |
| #27 | Graphene: An IR for Optimized Tensor Computations on GPUs | GPU 编译与性能 | 历史笔记 |
| #28 | BOLT: BRIDGING THE GAP BETWEEN AUTO-TUNERS AND HARDWARE-NATIVE PERFORMANCE | GPU 编译与性能 | 历史笔记 |
| #29 | PockEngine: Sparse and Efficient Fine-tuning in a Pocket | 分布式训练与 MoE | 历史笔记 |
| #30 | Chimera: An Analytical Optimizing Framework for Effective Compute-intensive Operators Fusion | GPU 编译与性能 | 历史笔记 |
| #31 | AMOS: Enabling Automatic Mapping for Tensor Computations On Spatial Accelerators with Hardware Abstraction | GPU 编译与性能 | 历史笔记 |
| #32 | Attention is all you need | Attention | 历史笔记 |
| #33 | FLASHATTENTION: Fast and Memory-Efficient Exact Attention with IO-Awareness | Attention | 历史笔记 |
| #34 | A Loop Transformation Theory and an Algorithm to Maximize Parallelism | 计算基础 | 开篇摘记,未完成 |
原 issue 中的推荐与外链
以下短评属于原 issue 作者的记录;保留来源,不将推荐语改写为论文结论。
- #9:Unikraft 知乎笔记。首次读取返回 403,后续已通过浏览器补入本地全文。
- #15:system-notes 仓库中的原笔记;当前已有本地版本。
- #18:原记录为“导师推荐的文章”,仅附 Fast Segment Anything 论文链接。
- #19:原记录为“导师推荐的关于 NPU 虚拟化的文章”,仅附论文链接。
- #20:原记录为“一篇比较有意思的使用 AI 自动设计芯片的文章,计算所和寒武纪一起做的”,仅附论文链接。