GPU Acceleration

GPU Center of Excellence

Rescale’s GPU Center of Excellence is a competence center for GPU Accelerated Workflows, collaborating with your Rescale Solutions and Commercial team to drive value and accelerate innovation. Our dedicated task force of global specialist architects and engineers brings expertise to harness the latest GPU architecture, ensure efficient workflows, and uncover innovation opportunities in productivity and experience.

Connect With a GPU Expert

Please reach to the GPU Center of Excellence to discuss how this persuasive technology can accelerate your simulation workflows

GPU Accelerated High Performance Computing

Graphics Processing Units (GPUs) play a pivotal role in High-Performance Computing (HPC) by significantly enhancing computational speed and efficiency. Originally designed for rendering graphics, GPUs have evolved to handle complex mathematical and computational tasks in parallel. Their architecture, with numerous cores optimized for parallel processing, enables simultaneous execution of thousands of threads. 

In HPC, GPUs accelerate simulations, data analytics, scientific modeling, and deep learning tasks, allowing researchers and scientists to achieve results in a fraction of the time compared to traditional Central Processing Units (CPUs). By offloading intensive computations to GPUs, HPC systems can achieve substantial performance gains and facilitate breakthroughs across various domains, making them an indispensable component in modern supercomputing environments.

Meet Our GPU Experts

Our global task force of specialist architects and engineers brings extensive expertise and experience to partner with your simulation experts, ensuring you get the most from your GPU-driven computational processes.

Benchmarking Industrial Simulation Workflows on GPUs

GPU implementation in the simulation world is still very much in its adolescence. Rescale takes an agile and dynamic approach to ensure all bases are covered in this ever-evolving market. We are here to work together with you to make the best of what is on offer.

Examples of the benefit of running your HPC workloads on GPU’s and how they compare to traditional CPU-based ones are illustrated below:

ANSYS Fluent GPU Acceleration – Benchmark

ANSYS Fluent is an industry standard commercial, general purpose CFD code. It is used aerospace, automotive and energy verticals 

Fluent’s native GPU solver has increased performance compared to its CPU based counterpart. The chart illustrates a typical external aerodynamics benchmark. Lower is better.

ANSYS Speos GPU Acceleration – Benchmark

ANSYS Speos predicts the illumination and optical performance of systems to save on prototyping time and costs while improving your product’s efficiency.

Its new GPU accelerated solver also shows favorable performance characteristics when compared to its CPU solver sibling. Lower is better.

Siemens STAR-CCM+ GPU Acceleration – Benchmark

Much more than just a CFD solver, STAR-CCM+ includes an entire suite of applications for solving problems involving complex geometries, multi-physics, flow (of fluids or solids), heat transfer, and stress.

The GPU solver’s raw speed performance is also mirrored in its economical value as shown by this plot comparing prices with the traditional CPU solver. Lower is better. The boost in performance and cost efficiency is clearly illustrated, as well as decreasing the data center energy footprint.

Expanding to AI/ML Workflows Using GPUs

Closely linked to the use of GPUs for traditional HPC workflow, are the predictive physics using AI/ML implementations. These algorithms also rely on massively parallelizable infrastructure that GPU provides and can be used to accelerate the engineering and scientific design cycles using commercial as well as open source development platforms. Our partnership with Nvidia and other accelerator infrastructure manufacturers ensures we are on the bleeding edge of this paradigm shift.