Our HPC team has been hard at work upgrading Bramble’s compute fleet. That work is nearly complete and the results are in. The upshot? Simulations on bramble are now significantly faster, and that means real cost savings for you.

Here’s what’s changed, and what it means for your engineering workflows.

What have we upgraded?

This year, we’ve made four major improvements to our HPC systems. Each of these changes helps squeeze more performance out of every core. Combined, the effect is dramatic:

simulations run on SSDs

1. All simulations now run on solid-state drives (SSDs)

 Drastically reducing I/O bottlenecks.

upgraded servers

2. We’ve upgraded network interfaces to 10Gb/s

Improving node-to-node and storage communication speeds.

Upgraded clusters

3. Clusters now run on the latest Rocky Linux OS

Bringing stability, compatibility, and performance improvements.

upgraded the MPI

4. We’ve upgraded the message passing interface (MPI)

Reducing communication overhead in parallel workloads.

Benchmarking results

As always, we validated the upgrades by running a suite of OpenFOAM benchmarks designed to mirror real-world bramble workloads.

  • Core counts tested: 96, 192, and 384
  • Mesh sizes: 80M, 140M, and 200M cells
  • Model: The DrivAer Model, steady-state RANS with k-omega SST
  • Run length: 4000 iterations, including mesh generation with snappyHexMesh

We captured both simulation time in hours and total cost (£) for each case before and after the upgrades.

Benchmarking Graph
Benchmarking Data showing faster and cheaper cfd simulations

Key takeaways

rocket fast simulations

Faster simulations

Across the board, simulations are running faster.

  • A 140-million cell model on 192 cores (our most common case) now runs 17% faster.

  • At 96 cores, you’ll still see a 7% improvement on the same case.

  • The biggest gains? At 384 cores, where a 140M cell model runs 31% faster

cfd scalability

Improved scalability

Previously, doubling core count didn’t always pay off, especially with smaller meshes. For example:

  • A 140M cell model running on 384 cores used to cost £23 more than on 192 cores.

  • After the upgrades? That cost difference is down to just £4, thanks to improved parallel scaling.

What does this mean for you?

In short: faster turnaround times and lower simulation costs, with no change to our pricing.

bramble’s OpenFOAM pricing remains at just £0.0615/core-hour, and we’re passing every bit of these performance gains directly to you.

Whether you’re running aerodynamic studies, cooling simulations, or something more niche, you’ll now get results quicker, and pay less to get them.

Let us know how you get on with the improved performance. As always, we’re here to help.

Contact us

See how bramble CFD could work for you

Discover more!

 

Find out more about some of bramble’s key features or bramble’s tailor built hardware for CFD options.

Related articles

Secure Your CFD Data with Mandatory MFA for All Bramble Accounts from 1 Sept

Multi-factor authentication (MFA) has been available on bramble for some time, and many users have already enabled it. From 1 September 2026, it…

Bramble & White Motorcycle Concepts CFD Partnership

Supporting the Next Phase of Aerodynamic Innovation We’re proud to be partnering with White Motorcycle Concepts (WMC), a company that’s…

Business Development Manager

A new position of Business Development Manager is being created by Bramble CFD to spearhead our ongoing acquisition process in UK.

Business Development Executive

Bramble CFD are creating a new Business Development Executive role to drive our active customer acquisition.

Analysing the Ferrari SF16-H

Ferrari SF16-H A model of the F1 Ferrari SF16-H is run through Bramble’s CFD software, with the CFD results analysed by our experts Photo by Morio -…