Cloud HPC for AI/ML

High Performance Computing for  AI/ML

Cloudam serves as a powerful engine for machine learning with datasets, algorithm models, and cloud GPUs that facilitates data creation, processing, annotating, model training, optimization, deployment, etc.

AI/ML Scenarios

computer vision

Computer Vision

voice-recognition.jpg

Voice Recognition

NLP natural language processing

NLP

knowledge graph

Knowledge Graph

auto pilot

Auto Pilot

AI-powered diagnosis

AI-powered Diagnosis

Solution Highlights

编组 24备份.png

Out-of-box Experience

With pre-installed Jupyter Notebook and Anaconda, you can harness the full power of popular Python libraries to analyze and visualize data.

flexible hardware scheduling

Flexible Hardware Scheduling

Various high-performance GPU choices are at your disposal.

full-stack HPC+AI envirionment

Full-stack HPC+AI Environment

Popular ML frameworks are pre-configured and tested on various hardware architectures on Cloudam.

cost-effective solution

Cost-effective Solution

Flexible pricing for different GPUs ensures efficiency and budget control at the same time.

Case Study

Tsinghua & Cloudam HPC Case Study

Tsinghua & Cloudam: Training NLP Model of 100 Billion Parameters

Tsinghua University AI Labs trained its 100bn parameters' NLP model on Cloudam's elastic cloud HPC-AI clusters 5 times faster than expected because of the attributes below:
  • Scalable & Manageable AI Clusters

 

Up to 3000 V100 & A100 GPU cards are available to deploy with SLURM as the job scheduler that allows 100% focus on training.

  • AI-powered Scheduling

 

The HPC system smartly schedules the latest hardware ensuring zero queues, and automatically terminate resources when there are no jobs.

 

  • Seamless Cooperation

 

Clusters on public cloud facilities seamless cross-regional cooperation of scientists worldwide to train NLP models.