Skip to content

Main Navigation Knowledge Base

Platform Overview

Scientific Problems

Scientific Cases

News & Insights

About Me AI Assistant

English

English

Appearance

Sidebar Navigation

Introduction

Overview

Architecture

Infrastructure

01. Hardware & Chips

Overview

AI Accelerator Overview

Cluster Selection & Acceptance

GPU Drivers & Lifecycle

Liquid-Cooled DC Design

GPU Server Topology

02. Cluster Architecture

Overview

CPU HPL Benchmark

GPU HPL Benchmark

AMD Tuning Guide

STREAM Benchmark

Benchmark Toolkit

03. Network (IB/RoCE)

Overview

RoCE AI Fabric

ClusterKit Testing

IB Admin Guide

NCCL Test

UFM Install

Linux NAT Gateway

InfiniBand Tuning for AI

04. Storage Systems

Overview

Advanced Features

Deep Dive: BeeGFS Deployment

Practical: GPFS ECE Deployment

Lustre Install

System Software

05. Automated Provisioning

Overview

Ubuntu Autoinstall

Diskless Boot (PXE+NFS)

Boot Failure caused by LD_LIBRARY_PATH

Cluster Delivery Automation

06. Cloud & Scheduling

Overview

Slurm Install (Rocky)

Slurm Install (Ubuntu)

Slurm User Guide

Slurm + Docker

K8s GPU Scheduling

07. Heterogeneous Computing

Overview

CUDA Install

Slurm Heterogeneous Scheduling

GPU Partitioning MIG/vGPU

CUDA MPS in Practice

Domestic GPU/NPU Porting

08. AI Compiler

Overview

Triton 101

Deep Dive into OpenAI Triton

PyTorch Compile Stack

MLIR Architecture

oneAPI Install

LLM Technology

09. Frameworks

Overview

PyTorch Autograd Deep Dive

PyTorch Internals: Tensor & Dispatch

Triton Kernel Practice

PyTorch Docker Env Setup

PyTorch Docker

10. Pre-trained Models

Overview

Pretraining Data Pipeline

Pretraining Engineering

Deep Dive: DeepSeek Arch

RTP Hybrid

11. Distributed Training

Overview

Parallelism: Single GPU to 10K

NCCL Tuning in Practice

LLM Lab 01 · DDP 101

LLM Lab 02 · DDP Minimal

LLM Lab 03 · AllReduce Autograd

LLM Lab 04 · Tensor Parallelism

LLM Lab 05 · Attention TP

LLM Lab 06 · Megatron Block

LLM Lab 07 · FlashAttention

Mixed Precision & Checkpoint

Loss Spike Debugging

12. Inference Engines

Overview

Long-Context Inference

vLLM Production Deployment

TensorRT-LLM in Practice

Deep Dive: vLLM Principles

Transformers vs vLLM Benchmark

Applications

13. Industry Apps

Overview

RAGFlow + K8s

HPC × AI Surrogates

Battery Lifetime Case

KB Q&A Production Stack

Private LLM Serving

Scientific Agent

LLM Inference SRE

14. AI×Science

Overview

Life Science

Weather

CFD

Materials

Foundation Models

On this page

Life Science: From AlphaFold to Protein Design

Core Technologies

MSA (Multiple Sequence Alignment) & Transformers
Equivariant GNNs
Diffusion Models in Molecular Generation

Case Studies

AlphaFold 2/3 Principles
RoseTTAFold

Pager

Previous page14. AI×Science

Next pageWeather

AI-HPC 智算超算专家 · Contact: openaihpc@gmail.com

Copyright © AI-HPC.org