RECRUIT

Reducing Energy Consumption in Radio-astronomical and Ultrasound Imaging Tools

image credit: shutterstock

When it comes to algorithms, technologies, and energy constraints, imaging efforts in radio astronomy and medical ultrasound share fundamentally similar challenges; both near the edge and further downstream in processing pipelines. Although time and space scales are orders of magnitude apart, the associated data processing and enabling hardware to image galaxy or brain both share the common requirements. That is, they must be processed in a local, real-time, and energy-efficient way. In this project, ASTRON (the Netherlands Institute for Radio Astronomy) and CUBE (the Center for Ultrasound and Brain imaging at Erasmus MC) join forces to tackle HPC and energy-efficiency challenges by utilizing new technologies and algorithmic improvements. The Adaptive Compute Acceleration Platform by Xilinx and Tensor Cores in NVIDIA GPUs are top examples of such enabling technologies. This project will unlock the potential of these highly-efficient technologies for use in radio astronomy and ultrasound brain imaging, delivering open-source libraries, innovation in limited-precision algorithms, and will develop a new tool to analyze energy efficiency. Ultimately, this will allow more (energy) efficient instruments to be built.

Participating organisations

ASTRON
Erasmus University Medical Center
Natural Sciences & Engineering
Natural Sciences & Engineering
Netherlands eScience Center

Impact

Output

Team

Laura Ootes
Laura Ootes
Hanno Spreeuw
Hanno Spreeuw
Ben van Werkhoven
Ben van Werkhoven
Senior RSE
Netherlands eScience Center
CS
Christos Strydis
JR
John Romein
Leon Oostrum
Leon Oostrum
Lead RSE
Netherlands eScience Center

Related projects

PADRE - The PetaFLOP AARTFAAC Data-Reduction Engine

Improving the AARTFAAC processing pipeline

Updated 6 months ago
Finished

Triple-A 2

Accelerating astronomical applications 2

Updated 22 months ago
Finished

Related software

cudawrappers

CU

The cudawrappers library is a C++ wrapper for the Nvidia C libraries such as the CUDA driver, NVRTC, and cuFFT.

Updated 17 months ago
12

Kernel Tuner

KE

Kernel Tuner greatly simplifies the development of highly-optimized and auto-tuned CUDA, OpenCL, and C code, supporting many advanced use-cases and optimization strategies that speed up the auto-tuning process.

Updated 12 months ago
105 15

PowerSensor3

PO

PowerSensor is a low-cost, custom-built device that measures the instantaneous power consumption of GPUs and other devices at a high time resolution.

Updated 16 months ago
6