Introduction to High-Performance Computing


1. From Room-Sized Machines to Life-Saving Simulations

Before diving into what High-Performance Computing (HPC) means today, let’s take a quick trip through time — from the machines that filled entire rooms to the systems that help us model the human brain, predict pandemics, and design new cancer drugs.


1940–1960s: The First Supercomputers

The birth of electronic computing

The first “supercomputers” emerged during World War II. They weren’t built for medicine, but for codebreaking, ballistics, and physics — yet they laid the foundation for every modern medical simulation.

  • 🖥️ ENIAC (1945) — the first electronic general-purpose computer, built for U.S. Army ballistics.

  • These early machines were massive — the size of an operating room, consuming as much electricity as a hospital wing.

  • Their mission: calculate faster than any human could.

../../_images/eniac.jpg

ENIAC (1945) — one of the first electronic supercomputers, programmed manually with cables and switches. Figure taken from Wikipedia.


1970–1990: The Cray Era

When speed became a race

In the 1970s, speed became the focus.
Supercomputers like the Cray-1 used vector processors to perform thousands of operations at once — a breakthrough that inspired modern GPUs (used in AI today).

../../_images/Cray-1.jpg

“Cray-1 (1976) — elegant design and groundbreaking parallelism.”
(Original: Wikipedia – Cray-1)


1990–2010: The Cluster Era

From super-machines to super-teams of computers

Scientists realized that connecting many smaller computers could outperform one massive one — the birth of clusters.

  • 1994: NASA engineers built the first Beowulf Cluster using standard PCs and Ethernet cables.

  • This democratized HPC — universities could now build their own “mini supercomputers.”

../../_images/Beowolf2-300x174.jpg

Photo of a Beowolf Cluster. Figure taken from https://memim.com/beowulf-cluster.html.


2010–Present: The GPU & Hybrid Era

From physics to AI and medicine

Today’s supercomputers combine traditional CPUs with GPUs and accelerators — the same hardware used for medical AI, radiology, and genomics.
Instead of a few super-fast processors, modern systems have thousands of smaller ones working together.

../../_images/supek-gpu-server.png

Modern HPC combines CPUs and GPUs for AI, simulations, and data analytics.


2. What Is HPC — and Why It Matters to You

So what exactly is HPC?

High-Performance Computing (HPC) means using many powerful computers at once to solve problems that are too big, too detailed, or too urgent for a single machine.

It’s the engine behind modern data-driven medicine — where algorithms, images, and molecules meet computation.

What is High-Performance Computing - YouTube (Croatian).


HPC in Everyday Healthcare Research

Here’s how HPC touches your field, even if you don’t see it directly:

Application

HPC’s Role

Radiology AI

Train deep neural networks on thousands of MRI/CT scans.

Drug discovery

Simulate how molecules bind to proteins.

Genomics

Process hundreds of patient genomes in parallel.

Epidemiology

Model virus spread and hospital resource needs.

Surgery planning

Run 3D simulations of patient anatomy.


3. Why Do We Need HPC?

Most modern medical and pharmaceutical challenges are too large for desktop computers:

  • A single CT scan can produce hundreds of megabytes of data.

  • One genome = 100 GB of raw data.

  • A deep learning model might require weeks of computation without HPC.

../../_images/laptop-vs-cluster.png

Laptop vs. Cluster


The Grand Challenges (and Opportunities)

The U.S. National Science Foundation once defined “Grand Challenges” — problems only solvable with HPC.
Many now directly relate to healthcare and life sciences:

  • Cancer detection and therapy

  • Drug design and molecular simulation

  • Climate and disease modeling

  • AI and deep learning

  • Understanding biological systems


In One Sentence:

🧠 HPC allows us to ask bigger questions — and get answers faster.

You can:

  • Solve bigger problems in the same time, or

  • Solve the same problems in less time.


Reflection

Where could HPC help you?

Think of a task in your daily work — an analysis, simulation, or AI model — that you wish could run faster or on more data.
Would HPC make it possible?