参考资料¶

内容优先采用厂商官方文档、项目官方仓库与原始论文。工具接口和产品规格会变化，实际实验前应回到对应版本的原始资料确认。

芯片架构与 CUDA¶

• NVIDIA A100 Tensor Core GPU Architecture Whitepaper
• NVIDIA A100 Datasheet
• CUDA Programming Guide
• NVIDIA PTX ISA
• NVIDIA Ampere Tuning Guide

Groq LPU 与张量流处理器¶

• Think Fast: A Tensor Streaming Processor for Accelerating Deep Learning Workloads（ISCA 2020）
• A Software-defined Tensor Streaming Multiprocessor for Large-scale Machine Learning（ISCA 2022）
• GroqCard Accelerator Product Brief
• Answer Fast: Accelerating BERT on the Tensor Streaming Processor

架构仿真与数据流¶

• gem5 Documentation
• Accel-Sim Framework
• Timeloop
• MAESTRO: Data-Centric DNN Dataflow Analysis
• SCALE-Sim

AI 编译器¶

• MLIR
• Apache TVM Design and Architecture
• TVM TensorIR
• Triton Documentation
• LLVM Documentation

模型架构¶

• Attention Is All You Need
• GQA: Training Generalized Multi-Query Transformer Models
• Switch Transformers
• FlashAttention

量化与推理加速¶

• PyTorch Quantization API
• SmoothQuant
• GPTQ
• AWQ

建站与发布¶

• Material for MkDocs
• Mermaid
• Cloudflare Pages：MkDocs

内容更新记录¶