Survey of HPC Applications Supporting GPU Computing and Their Adoption Across Domains

NVIDIA has largely driven the growth of GPU computing in high‑performance computing (HPC) by creating a robust software ecosystem—including the CUDA parallel programming API, OpenCL standards, and OpenACC directives—and by establishing more than 20 GPU centers worldwide to accelerate development and adoption.

With the rise of artificial intelligence, many organizations rely on deep‑learning algorithms that depend on GPUs, making AI a primary growth engine for NVIDIA.

The Intersect360 survey collected data from HPC sites in Q3 2017, recording 1,792 program entries that correspond to 534 distinct applications; the analysis focuses on the top 50 most frequently used applications.

Among these 50 applications, 34 already provide GPU support and two are under development, indicating that GPU acceleration has reached a critical point and is becoming mainstream in the HPC market.

Domain‑specific findings include: chemistry (20 applications, 16 GPU‑enabled), computational fluid dynamics (ANSYS Fluent and OpenFOAM with GPU support), structural analysis (7 of 8 applications GPU‑enabled), biosciences (GPU‑BLAST and a GPU‑accelerated Bowtie project), weather and environmental modeling, business intelligence (GPU versions of SAP and Oracle), physics, and pattern recognition (TensorFlow leveraging GPU acceleration).

The report underscores the pervasive adoption of GPU computing across diverse scientific and industrial fields, driven by both NVIDIA’s ecosystem and the expanding demand for AI‑enabled HPC workloads.