2021 · Theory & algorithms

Software that keeps up with supercomputers

Numerical libraries connect mathematical algorithms with changing hardware.

Jack Dongarra

2021Turing Award

Visual guide

The prize, pictured

Open full-size ↗Download poster ↓

Software that keeps up with supercomputers: Numerical libraries connect mathematical algorithms with changing hardware. — Original conceptual diagram by Curionode. Read the text version below for the full explanation.

Read the poster text

Formulate
Express the numerical task.
Organise
Arrange computation and data movement.
Scale
Adapt algorithms to powerful hardware.

A benchmark or one linear-algebra kernel does not measure every aspect of a computer’s usefulness.

01 · The idea

In everyday language

Jack Dongarra developed algorithms and software libraries for high-performance numerical computing. Solving systems of equations efficiently requires attention to both mathematics and the way processors move and share data. His work helped numerical software evolve across successive generations of vector, parallel and distributed machines.

Why it matters

Reusable numerical software makes advanced computing accessible to scientific and engineering applications without rebuilding every kernel.

02 · The people

Who brought the idea to life?

Jack Dongarra

Jack Dongarra developed numerical algorithms and software libraries that let scientific applications benefit from successive generations of high-performance computers. His work on widely used linear-algebra software connects mathematical reliability with the realities of parallel hardware. He also helped build shared benchmarks and scientific software infrastructure. His appointments at Tennessee, Oak Ridge, and Manchester reflect a career of collaboration across laboratories, universities, and national boundaries.

Read the biography ↗

03 · Read further