Guides & Whitepapers.
In-depth engineering documentation and system architectures for distributed computing.
Whitepaper
AI Data Centre Design Playbook
A comprehensive guide to planning and building high-density facilities for AI workloads. Covers liquid cooling, power distribution, and floor loading.
Core Sections:
- Calculating power requirements for H100/B200 clusters
- Liquid cooling strategies: DLC vs. Immersion
- Network topology for non-blocking throughput
Blueprint
H100/B200 Cluster Reference Architecture
Validated network and power topology for deploying 1024+ GPU clusters. Includes switch fabric diagrams and rack layouts.
Core Sections:
- Power distribution schemes up to 100kW per rack
- Non-blocking InfiniBand/RoCE network topologies
- CDU and cooling loop design templates
Tool
TCO Calculator: On-Prem vs. Cloud
Interactive spreadsheet to model the 3-year total cost of ownership for training clusters. Input your own electricity and hardware costs.
Core Sections:
- Capital expenditure (CapEx) modeling for GPU nodes
- Operational cost variables (power, cooling, maintenance)
- PUE impact on long-term electricity outlays
Guide
Liquid Cooling Implementation Guide
Best practices for deploying Direct-to-Chip (DLC) and Rear Door Heat Exchangers (RDHx) in brownfield environments.
Core Sections:
- Direct-to-Chip (DLC) water chemistry and flow rate design
- Immersion cooling dielectric fluid selection criteria
- Retrofitting legacy facilities for liquid-to-air cooling