There has been recent chatter in the automotive industry news about high performance computing (HPC) as it relates to speed, price and automotive applications. I’d like to break some of this down a bit further and explore why automotive is the next great area for HPC.
Automotive AI, Design & Security
In an Auto Futures article about HPC and AI revving up the automotive industry, Tractica’s principal analyst Keith Kirkpatrick reported a convergence of AI and high performance computing in the automotive industry. He stated “HPC enables faster data processing, and permits the development of high-fidelity environments that can simulate vehicle crash tests and wind-tunnel tests... HPC speeds up the data-intensive process of training object-recognition, object-avoidance, and response systems used in autonomous vehicles.”
One of the ways automakers have used HPC has been in automotive neural networks, object training, crash testing and fluid dynamic designs of airflow for such things as turbulence and noise. Automakers previously had to use soundproofing to deal with noise. Something as simple as the placement of a bolt can change the noise heard in the vehicle, which can be reduced by automotive modelling with computational fluid dynamics. This type of application requires a lot of compute power, making it an ideal application for HPC.
The design of vehicles is mandated by safety standards and fuel efficiency. One change in a safety standard could require a major overhaul in the design. Instead of a labour intensive re-design, HPC can quickly and cost-effectively accomplish this task.
Understanding Autonomous Vehicle Data
There is an incredible amount of data that comes from autonomous vehicles. Sam Abuelsamid, a principal research analyst at Navigant Research and contributor at Forbes, reports that autonomous vehicles can collect four terabytes or more of raw sensor data a day. Tractica’s Kirkpatrick predicts that the data generated by autonomous and connected vehicles is projected to increase exponentially over time.
Neural networks easily amalgamate large quantities of data to better understand or process a single result or action. With this quantity of data coming from autonomous vehicles, deep learning and neural networks will start to become the norm as the volume of data generated will be too great otherwise.
Security, Speed, & Insurance
As discussed in the Auto Futures article, HPC can find and heal software problems or a security vulnerability. While other methods could take up to six months, HPC systems are able to find and update the system in as little as three days. HPC computing can also train algorithms in cases when something goes wrong for providers of car insurance, warranty and rental agencies.
Where Do You Go for Automotive HPC Applications
Not surprisingly, I’m going to tell you Iceland. With a long history of servicing automotive clients, like BMW and VW, Verne Global provides the automotive industry the power needed to advance safety, improve designs, heal software vulnerabilities, empower autonomous vehicles and enable data collaboration amongst industry leaders with great speed and competitive pricing.
The way we do this is through HPC colocation and our HPC Cloud platform - hpcDIRECT, a bare-metal cloud ideally suited for HPC applications without any of the drawbacks that come with running HPC on the public hyperscale cloud - unpredictability of performance, unpredictable capacity at a certain configuration, and cost.
Timothy Prickett-Morgan of The Next Platform recently reported in “The Gap Between Clouds and On Premises for HPC”, “Growth of public cloud is certain. What’s more uncertain is the performance of the calibre required by HPC. One thing, however, is sure: if hpcDIRECT performs better and costs less than public cloud, then that sure makes the decision of eschewing the big, hyperscale providers for Verne Global a lot easier.” He notes that Verne Global offers HPC specialists that can help customers find the right optimised environment and keep it optimised as conditions and applications change.
Verne Global’s HPC has an advantage over generic clouds because generic clouds such as Google and Amazon are designed for basic operation of basic software such as the top clouds apps of Office 365 Google Office, Salesforce and Dropbox which requires just moderate power to operate. HPC is very dense compute working on practical functions required by the automotive industry, such as machine vision, autonomous driving or crash testing.
For information about Verne Global and HPC for the automotive industry, please contact me for a free initial consultation and trial: email@example.com.