Computational Core

The computing power behind research

MBI’s Computational Core provides the infrastructure — network, storage, high-performance computing and scientific software — that makes large-scale mechanobiology research possible. From terabyte-scale imaging data to GPU-accelerated modelling, we keep the science moving. 

1.6PB

Parallel Storage

574 TFLOPS

Deep learning computer

224

GPU Core

1.5TB

Main Memory

CORE MISSION

Built for data-intensive science

Modern mechanobiology generates data at a scale that demands infrastructure. A single imaging experiment can produce several terabytes across thousands of files. Multiple microscopes run concurrently, creating continuous high-volume data streams that must be stored reliably, processed efficiently, and made accessible to researchers in real time.

The Computational Core was purpose-built to meet this challenge, designed from the ground up when MBI moved to its current research space.

Store

High-capacity, high-reliability storage for imaging and experimental datasets at petabyte scale.

Strategized storage

A tiered storage strategy ensures that active research data is fast and accessible, while lower-priority and archival data is managed cost-effectively.

Primary (GPFS parallel): 1.6PB scalable to 3PB, 10Gbps throughput

Secondary (SAN): 400TB cluster and ILM managed

Archive / low-priority: 200TB software-defined storage

Connect

High-speed network fabric linking microscopes, servers, storage and end users —including connectivity to NSCC.

Multi-tier network architecture

MBI delivers high-speed connectivity from microscope to storage to researcher through a multi-tier network architecture and a dedicated high-bandwidth link to the National Supercomputing Centre (NSCC).

Storage fabric: 56Gb/s Infiniband (scalable to 100Gb/s)

End-user access: 1Gbps - 10Gbps to centralised storage

Core switching: 40GbE with 100GbE upgrade path

External link: 100GbE to NSCC

High-performance computing

GPU-accelerated computing for image analysis, deep learning and large-scale biological modelling.

Showcasing MBI science

A cluster of current-generation GPU servers — equipped with NVIDIA P100 and V100 (NVLink) GPUs alongside large-memory nodes — supports everything from image processing pipelines to deep learning model training.
50 TFLOPS
Double-precision compute

100 TFLOPS
Single-precision compute

574 TFLOPS
Deep learning performance

CPU cores (total)
224 cores

Main memory (total)
1.5TB

Analyse

Bioinformatics, genomics, dabase mining and data visualisation tools for complex biological data.

Tools for every stage of analysis

The Core provides and supports a broad suite of commercial and ope-source applications, covering the full analytical workflow for mechanobiology research.

Image processing | Computer modelling | Bioinformatics | Genomics | Database mining | Data visualisation | Archival storage

Full-stack IT expertise on site

Our team of IT professionals covers systems design, implementation and management, application development, and independent web and network administration —providing MBI researchers with direct access to specialist support without external dependency.

Meet the team

MANALAD, Manuel Santiago

Desktop and Software Support

SHAGIROV Murat

High-performance Computing (Research Fellow)

Mazlan Bin ITHNIN

IT Support Officer