Services
Fluid Numerics provides a wide range of services for high performance scientific computing for both on-premise and cloud. You may be looking to accelerate your code with GPUs, leverage cloud computing resources, or design numerical methods from the ground up.
As a Google Cloud, Google Workspace, and Supermicro reseller staffed with applied mathematicians and computational scientists, Fluid Numerics is a one-stop-(boutique)shop for designing and deploying scientific software on appropriately matched hardware. Whatever the case may be, we can help.
Hourly Consulting
Sometimes your project just needs an extra set of eyes to stay on track for success. With hourly consulting we can guide you through software development, porting and optimization work, Cloud-HPC systems design, engineering, operation, and management.
Software Development
Outsource your scientific software development work to Fluid Numerics. We have experience in C/C++, Fortran, Python, and Julia programming languages, in addition to parallel programming paradigms including MPI, OpenMP, OpenACC, CUDA, HIP, and PyTorch.
We’ll work with you to develop project scope, timeline, and performance & correctness expectations. Software development projects are typically contracted as fixed-fee or hourly service.
Mentored Sprint
Mentored sprints are a popular service for porting & optimizing software for new systems, like GPUs. Over the course of 8 weeks, we help you establish benchmarks for your code and use these to guide porting & optimization work. During one week, we meet daily with your team to rapidly port your code to new hardware. The sprint is concluded with a retrospective and complete write-up of the progress made and lessons learned. Mentored sprints are typically contracted as a fixed-fee service.
Cloud-HPC Systems Design, Engineering, & Deployment
Google Cloud has compute resources to offer, but out-of-the-box, it doesn’t have the same look-and-feel of the traditional HPC user experience. Fluid Numerics can help you quickly get started with solutions that combine auto-scaling clusters with high performance parallel file systems, following best practices.
You provide us specifications and we provide you with complete engineered solutions alongside administrator and user documentation.
Bring your own domain or work with us to get started from scratch. As a Google Workspace and Google Cloud reseller, we are your full service shop for getting started with HPC on Google Cloud.
Design, engineering, and deployment is a fixed-fee service, per project service.
Try it before you buy it
Supermicro systems + AMD GPUs
Fluid Numerics houses a small cluster of compute platforms equipped with AMD Instinct and Radeon GPUs for experimentation and software development purposes.
If you are interested in purchasing Supermicro systems and would like to work with Fluid Numerics, we can provide you with access to the resources we have on-site to help you gauge the viability of AMD Instinct platforms for your organization. Additionally, our team has been working with the ROCm ecosystem since 2018 and can provide you with guidance on porting your applications to the AMD Instinct GPU platform.
Should you decide to move forward with a systems purchase, you can work with Fluid Numerics to procure and commission your next HPC system.
Remote System Administration & User Support
Whether you’ve purchased a Supermicro system from Fluid Numerics or you are using a Cloud-HPC system we’ve deployed for you, our team can provide remote support. This service can include HPC Systems Consulting, Cloud-HPC System Administration, and Research Software Engineer Support and is available as a fixed-fee monthly subscription
Reviews
Dr. Melissa Kozul - University of Melbourne
Joe from Fluid Numerics was our mentor during a code development sprint organised by the Pawsey Supercomputing Centre in Perth. He helped us develop tailored solutions for our software and moreover was very responsive and organised before and during the sprint. Joe is a true professional and it was great to work with him!
Dr. Emily Kahl - University of Queensland
The Fluid Numerics team is extremely knowledgeable and wonderful to work with. My team and I worked with Joe during a sprint to implement a tricky new feature to a large, legacy scientific codebase and our progress would've been much slower without his assistance and deep knowledge of GPU architectures. Would absolutely work with them in future.
Dr. Waseem Kamleh - University of Adelaide
Fluid Numerics provided extremely helpful profiling and performance analysis of my scientific software code. The resulting optimisations yielded significant performance improvements on the AMD ROCm platform. Joe was fantastic to work with during our coding sprint.
