AI Hypercomputer overview

AI Hypercomputer is a supercomputing system that is optimized to support your artificial intelligence (AI) and machine learning (ML) workloads. It's an integrated system of performance-optimized hardware, open software, ML frameworks, and flexible consumption models.

AI Hypercomputer uses best practices and systems-level designs to boost efficiency and productivity across AI pre-training, tuning, and serving.

System architecture

AI Hypercomputer is comprised of the following layers:

  • Performance-optimized infrastructure: contains accelerators, networking, and storage resources that provide the computing capabilities to support your workloads.
  • Open software: optimized versions of popular machine learning frameworks such as TensorFlow, PyTorch, and JAX. Google provides operating systems (OS) that are configured with essential software for leveraging the compute resources provisioned in your clusters. To deploy and manage a large number of accelerators as a single unit, you can also use Cluster Director for Google Kubernetes Engine, or Cluster Director for Slurm, or directly through Compute Engine APIs.
  • Consumption options: multiple options to provision clusters that optimize costs and hardware availability based on your specific needs and workload patterns.

Benefits

AI Hypercomputer has the following benefits:

  • High performance and goodput: Goodput metrics measure ML Productivity. AI Hypercomputer optimizes the scheduling, runtime, and orchestration layers.
  • Get up and running quickly: AI Hypercomputer provides tools and blueprints that let you reliably and repeatedly deploy large numbers of accelerator-optimized resources that are configured to support your most demanding AI and ML workloads.

Use cases

AI Hypercomputer was designed to meet the needs of the following use cases:


Use case
Example workloads

Large-scale AI and ML workloads
  • Generative AI distributed training
  • Generative AI inference
  • Fraud detection
  • Recommendation models

High performance computing (HPC)
  • Complex simulations
  • Drug discovery, protein folding, and genomic analysis
  • Risk analysis and quantitative trading

What's next?