While Computer-Aided Engineering (CAE) certainly benefits from HPC, configuring and deploying efficient HPC clusters can be a challenge. Altair has teamed up with Intel to take on this challenge and provide a suite of solutions for CAE in HPC. We caught up with Ravi Kunju, VP of Strategy and Business Development at Altair, to learn more.
insideHPC: How does Altair approach compute-intensive engineering? What is the underlying technology?
Ravi Kunju: Altair is unique in that we are the only company that addresses compute-intensive engineering from three angles:
- Application Software: We provide highly parallel CAE software (in Structures, CFD and Electro Magnetics) & Data Management (Simulation Cloud Suite)—all under the HyperWorks suite.
- Workload Management: We are a leader in comprehensive HPC workload management software (the PBS Works suite), including web portals to manage, monitor and visualize HPC data and provide in depth analytics on workloads.
- Product Design: In addition to developing and delivering software products and solutions, Altair employs over 500 engineers who employ the above two to solve real world problems everyday for our over 2000 customer worldwide.
insideHPC: Who should be interested in this?
Ravi Kunju: We think the following types of people will really want take a look:
- CIO and CTO’s looking to reduce IT spend on HPC, in the areas of configuration, support, application integrations and more
- VPs / Directors of Engineering who want to increase the pace of innovation
- CAE / FEA engineers who are looking to improve simulation throughput and massively improve product design
- Engineers looks to do massive explorations / simulations over a vast design space
- HPC Administrators looking for a turnkey HPC solutions
insideHPC: Do you have a user story that you might want to share?
Ravi Kunju: Our HyperWorks Unlimited customer Mubea says “With HyperWorks Unlimited we’re able to optimize our tailor rolled applications more efficiently than we ever could before. It’s an incredible product that gives our engineers access to all the software they need, fully configured and ready to use.”
insideHPC: What are the limitations to HPC in the CAE world? How does Altair overcome these?
Ravi Kunju: Here are the top limitations to HPC in CAE, and how Altair addresses these:
- Access to HPC: For many companies it is simply too hard and/or costly to get started with HPC in the first place. Altair addresses this through an easy to use web interface that is application aware (Compute Manager).
- Data Movement: Large amounts of data needs to be transferred from analysis, and can be inefficient, so remote visualization can mitigate this: Altair offers Remote Visualization through Display Manager—no data is moved, only pixels. Second, no software needs to be downloaded—all native pre and post processing applications can run through a web browser.
- Business / Licensing Model: Capital costs for procuring HPC hardware and the operational costs of running and supporting application on HPC gear can be prohibitive. Altair’s application softwares are uniquely licensed to encourage large scale exploration; our licensing for application software (HyperWorks) and HPC middleware & Web Portals (PBS Works) uses the same token system of HyperWorks Units; second we also offer both a private cloud turnkey managed private cloud appliance (HyperWorks Unlimited) and a public HPC Cloud (HyperWorks On Demand). HyperWorks Units can be exchanged for HPC cycles on HyperWorks on Demand increasing the value of Altair customer investment.
insideHPC: Intel is a partner of yours. How did this relationship come about and what does it mean as far as the technology goes?
Ravi Kunju: Altair works with Intel very closely on multiple levels:
- In developing HyperWorks solvers (highly parallel and also embarrassingly parallel solvers) , working on the latest chip sets to optimize our solver run times and performance – you can read about out latest performance results on Ivy Bridge EX here.
- Consuming Intel chipsets on all our HPC data centers around the world
- Ensuring our PBS Professional workload manager is optimized to take advantage of Intel’s coprocessor Xeon Phi – Altair is a leader in supporting this technology (see here for more).
insideHPC: You recently announced a free webcast that you made with Intel. What is this webcast all about and where can those interested find it?
Ravi Kunju: Yes indeed, we produced this webcast with Intel and Scientific Computing World to present the latest capabilities of HyperWorks Unlimited, which has just recently been certified Intel Cluster Ready. What’s also special about this webcast is that we announced our first HWUL customer, Mubea, live on the webcast. We also cover how HPC can transform what’s possible in engineering, and how to avoid roadblocks to HPC in CAE. You can read all about it here.