AI Cloud

Browser-Based HPC For AI

Revolutionize your research and innovation with CR8DL’s AI Cloud, providing a comprehensive suite of domain-specific solutions engineered to meet the needs of researchers and organizations, driving breakthroughs and acceleration discovery.

Accelerate discovery

CR8DL AI Cloud

The power of supercomputing, accessible through your browser with expert support. CR8DL AI Cloud offers a comprehensive solution for AI and computational intensive workloads with our core products Origin, Base, Explore and Scale.

Origin

A web-based control console that provides users with a flexible and cost-effective way to access high-performance computing resources. It allows users to access all of their CR8DL products and services from a single dashboard, providing users with a unified view of their resources.

Base

CR8DL’s HPC Infrastructure as a Service (IaaS) offering including GPU on-demand, allowing for the scale up and scale out of AI applications and models launched in Explore. Base is housed in a secure, sustainable, and water-free private data center with secure access, live engineering support/on-boarding, and dedicated capacity.

Explore

A no-code, low-code and full-code HPC workspace that is focused on AI, molecular biology, quantum simulation, image processing, and financial analytics use cases. It offers pre-configured containers for common AI frameworks, enabling our subscribers to accelerate their development and training processes. Explore provides user-friendly self-guided training and documentation, as well as expert on-call engineering support available through multiple channels.

Scale

CR8DL’s AI Platform-as-a-Service (PaaS) for AI is a cloud-based platform that provides on-demand access to a cluster of high-performance computing resources, specifically designed to run compute-intensive workloads in AI, machine learning, and deep learning. It is designed to be scalable, allowing users to quickly add more resources as needed, and is accessible from anywhere with an internet connection. This makes it easier for organizations to implement and utilize AI technologies, without the need to invest in costly hardware and infrastructure.

Redefining AI Cloud: The CR8DL Advantage

Accessibility

BROWSER-BASED HPc PLATFORM.

Scalability

SCALE UP OR DOWN WITH EASE.

Transparency

ONLY PAY FOR WHAT YOU USE. nO HIDDEN FEES.


CR8DL AI Infrastructure

The CR8DL AI Infrastructure built on NVIDIA hardware is housed in a secure and sustainable private data center and includes high-capacity and redundant network and storage resources to ensure reliable and uninterrupted service.

CR8DL Base GPU Cluster

Node and Cluster Technical Specifications

Compute Nodes
  • CPU – AMD EPYC 7713 Processor
    Sockets – 2
  • Cores/Threads – 64 Cores/128 Threads per socket
  • L3 Cache – 256MB per socket
  • Clock – 2.0Ghz boost to 3.675Ghz
  • Internal Storage – 46TB over 6 x 7.7TB NVMe
  • Network – 16 x 100G Ethernet/InfiniBand
Node AcceleratorsGPU
  • 8x NVIDIA A100/80GB RAM
  • HGX backplane
  • 600GB/s inter-GPU throughput/3rd Gen NVIDIA NVLink
Cluster Storage and LAN
  • 8x Cluster capacity 256TB/16x100GE Interconnects
  • Multi-fabric 100G Ethernet/InfiniBand
Internet Access
  • Multi-homed Internet Access
  • 100Gbps Fabric
  • IX Services powered by DE-CIX

Available System Benchmarks

AMD EPYC 7763, NVIDIA A100-SXM-80GB
MXNet NVIDIA Release 22.04, PyTorch NVIDIA Release 22.04, TensorFlow NVIDIA Release 22.04

Image Classification

Imagenet/ResNet
28.11 Minutes

Object Detection (Heavy Weight)

COCO/Mask R-CNN
43.787 Minutes

Natural Language Processing (NLP)

Wikipedia/Bert[1]
19.828 Minutes

Speech Recognition

LibriSpeech/RNN-T
31.291 Minutes