Since the early days of computational fluid dynamics (CFD) innovation, Software Cradle—founded in 1984—has consistently set industry benchmarks. Now a vital part of Hexagon’s Design & Engineering division, following its integration into MSC Software in 2019, Cradle continues to pioneer advanced CFD solutions. Its flagship solvers include scFLOW, ideal for complex general-purpose flow analysis with polyhedral meshing, and scSTREAM, a high-speed Cartesian solver tailored for thermal and electronic cooling simulations, and scFAST, a next-generation GPU-first solver designed to deliver ultra-fast performance for demanding simulations.
Table of Contents
Introducing scFAST: GPU-First CFD for the Next Generation
Driven by the relentless demand for faster, more scalable simulations, Hexagon has unveiled scFAST—a groundbreaking CFD solver re-engineered for full GPU acceleration. Specifically tuned for NVIDIA Hopper and Blackwell architectures, scFAST redefines how engineers simulate large-scale fluid systems with efficiency and precision.
Why Shift from CPU to GPU in CFD?
Traditional CFD solvers rely on CPUs, which, while versatile, are limited by their serial processing capabilities. In contrast, GPUs offer massive parallelism, enabling simulations to run faster, more efficiently, and at higher fidelity. By eliminating CPU bottlenecks and optimizing GPU memory usage, scFAST delivers an entirely new tier of performance.
For those new to the concept, this shift builds on core CFD principles where simulation accuracy and speed are directly linked to processing power.
scFAST: Breakthrough Technologies at a Glance
1. End-to-End GPU Execution
scFAST operates entirely within the GPU environment—no data shuttling between CPU and GPU—ensuring minimal latency and unmatched processing speed.
2. Memory-Efficient RM-MCGS Algorithm
The solver’s Reduced-Memory Multicolor Gauss-Seidel (RM-MCGS) method replaces traditional Jacobian matrix storage with a Fréchet derivative approach, significantly lowering memory demand—a perfect match for GPU architecture.
3. Cross-Platform Compatibility via OpenMP
scFAST features portable OpenMP code, allowing seamless execution on both GPUs and multicore CPUs. This ensures flexible deployment across different hardware setups.
4. Adaptive Code Optimization with C++ Templates
By leveraging C++ templates, scFAST dynamically tailors its code paths based on simulation parameters, offering custom-optimized performance for every case.
A Performance Milestone: 450x Faster Simulations
On a single NVIDIA H100 GPU, scFAST delivers 15% higher performance compared to a legacy solver running on 384 CPU threads across four AMD EPYC 7763 processors. The result? A staggering 450x speedup, cutting simulation times dramatically while preserving accuracy.
Figure 1: Performance comparison for a 53-million-cell Onera OM6 wing mesh, demonstrating optional memory savings with RM-MCGS. (Source: Hexagon)
Mega Simulations on a Single GPU
scFAST redefines scalability by enabling over 50 million elements per GPU, allowing single-card simulations that once required an entire CPU cluster. This is a game-changer for industries like aerospace, automotive, and energy, where large-scale CFD problems are mission-critical.
Figure 2: CRM-HL model executed with a 68-million-element mesh on an NVIDIA A100 GPU using scFAST. (Source: Hexagon)
The Road Ahead: AI, CAD, and 1000x Gains
“We’re thrilled to offer early access to scFAST,” shares Subham Sett, VP of R&D and Innovation at Hexagon. “With broader release and full compatibility for NVIDIA Blackwell GPUs on the horizon, the potential is enormous.”
Looking forward, Hexagon is investing in:
- Faster meshing technologies
- AI-enhanced reduced-order models
- Tighter CAD integration.
One of the upcoming features includes improved CAD integration, aimed at streamlining geometry preparation and reducing preprocessing time.
These initiatives aim to amplify productivity by up to 1000x, making CFD simulation smarter, faster, and more accessible across industries.
Final Thoughts
With scFAST, Hexagon is rewriting the rulebook for CFD performance. This GPU-native solver isn’t just an upgrade—it’s a revolution in engineering simulation. As industries demand ever-more-accurate, real-time insights, scFAST positions itself as the essential tool for tomorrow’s most challenging fluid dynamics problems.
🔗 Reference & Further Reading
- To learn more about the technical architecture and performance benchmarks of scFAST, visit the official Hexagon resource library:
- 👉 Moving Towards 1000x Productivity – Introducing Hexagon’s scFAST, a Fully NVIDIA GPU-Optimized CFD Solver