Simcenter STAR-CCM+: Steady State External Aerodynamics
11M Cell Shakedown Runs: On-Premise Cluster vs. Corvid HPC Bare Metal
November 6th, 2026
Simcenter STAR-CCM+ v.19.06
On-Premise Cluster vs. Corvid HPC OD3 Results
Wall-Clock Time
Cost Per Job (USD)
Why Bare Metal Outperforms Aging On-Premise Clusters
In CFD, a CPU core is not just a CPU core. Solvers like Simcenter STAR-CCM+ are heavily memory-bandwidth bound, and Intel architectures from 2018 through 2025 simply cannot compete with the memory throughput of AMD EPYC Genoa.
However, node architecture matters just as much as the chip. Because every dual-socket AMD Genoa server shares the exact same memory bandwidth ceiling, packing 192 cores onto a node (like AWS hpc7a.96xlarge) cuts the available bandwidth per core in half compared to a 96-core layout. Corvid HPC’s OD3 bare-metal nodes use 96 cores across two sockets—delivering 2x the memory bandwidth per core over hpc7a.96xlarge and cutting this 480-core run from 90 minutes to 9 minutes.
Unlocking Next-Level Speed
The goal of this benchmark was to demonstrate how shifting from an aging or over-utilized on-premises cluster to optimized bare-metal infrastructure can transform an engineering workflow.
The baseline on-premises cluster took 1.50 hours to solve the case on 480 cores. Moving that exact workload to Corvid HPC’s OD3 platform delivered massive, immediate speed improvements:
At 480 cores: Runtime dropped to just 9 minutes (0.15 hours)—a 90.33% reduction.
At 960 cores: Runtime dropped to just 6 minutes (0.10 hours)—a 93.67% reduction.
For an engineering team, cutting a 1.5-hour solve time down to under 10 minutes fundamentally changes the design workflow. Rather than stepping away and waiting for a local queue to clear, engineers can treat complex fluid dynamics as a rapid iteration tool. This speed allows a team to test, analyze, modify, and re-test a model multiple times within a single morning. Ultimately, this massive increase in daily throughput enables organizations to explore a much wider design space, validate better solutions faster, and extract significantly more value from the simulation software licenses they already own.
Secure Infrastructure Built for Engineering
Managing complex engineering workloads requires an environment explicitly designed for those solvers. Corvid HPC delivers a secure, fully managed infrastructure that removes the IT burden of maintaining and upgrading an on-premises cluster. Our bare-metal approach ensures rapid job deployment, predictable billing, and a completely protected environment for sensitive intellectual property, including ITAR and CMMC compliance.
By migrating to purpose-built HPC hardware, simulation teams aren't just saving time—they are fundamentally expanding their capacity for design exploration without needing to purchase additional software licenses or procure new local servers.
| Platform | Core Count | Wall-Clock Time (hr) | Time Reduction | Cost Per Job (USD) |
|---|---|---|---|---|
| On-Premise Cluster | 480 | 1.50 | Baseline | $0.00 |
| Corvid HPC OD3 | 480 | 0.15 | 90.33% | $4.18 |
| Corvid HPC OD3 | 960 | 0.10 | 93.67% | $5.57 |