New accelerated modeling for automotive development is here
Siemens helps Arm prepare for Software-defined vehicles (SDV) with Arm Cortex-A720 AE pre-silicon support
Developers of SDVs are facing unparalleled software complexity. Software for SDVs could take up 100M lines of code rising to 300M+ lines in the near future causing untold difficulties in integration, testing and validation, and organizational strategy.
In SDVs, innovation and value will come from software and hardware optimized to run that software, not just from the conventional Electronic Control Unit (ECU) hardware. So software is an issue that won’t just go away.
Exploring the challenges of the SDV
Consumers want the same highly sophisticated software features in their cars that they are familiar with from their consumer electronics devices. Integration of these new software features is the first challenge facing developers. This is further compounded by a distributed development environment with OEMs relying on an increasing number of supply chain partners to fulfil consumers’ demand and retain competitiveness.
Testing and validating functionality of ever-increasing volume of software is the second challenge facing developers of SDVs. Everything needs to be unit tested, integration tested, and regression tested to achieve the highest quality. Unfortunately, the supporting tools available to SDV developers fall woefully short of enabling the level of validation needed.
Traditionally, software development waits until silicon is available before developing or porting software onto new architectures. New silicon is typically only available 12 to 18 months after the latest IP is launched. So, it takes a very long time before OEMs and Tier 1 developers can access the latest software on the latest architectures. This can mean that new vehicles can hit the market with technology that is potentially many years old, further stifling competitiveness.
Addressing the SDV challenge
The only real way to stay ahead of the game is to develop software first. A software-centric approach to SDV development means starting out with the software concepts to ensure the richest feature set possible. That means developers have to start software before hardware silicon is available. This is where the, now common, term shift-left comes from. Shifting software development further left than the hardware in the development cycle.
This software centric focus requires significant additional support to enable modern large scale software development methodologies for large teams. This includes agile development, large scale nightly regressions, and continuous integration/continuous delivery (CI/CD) systems, all of which impact the development environment of future automotive electronics systems across the entire development team.
Distributed development of SDV software means only one thing – it must move to the cloud. Developers across the automotive supply chain must embrace the cloud as a vital tool to aid in development, integration and testing of hugely complex software for SDVs.
The entire automotive supply chain must re-think strategies on SDV development. Tier 1 and 2 developers need updated and modern cloud-based tools focusing on digital twin-based software development. OEMs need the very same tools for informed decision making and to monitor supply chain partners.
Siemens EDA – a light at the end of the SDV tunnel
Siemens EDA has announced an accelerated pre-silicon development environment on the cloud as part of the PAVE360 digital twin solution that has the potential to address the SDV challenge. Digital twin-based development is at the heart of this philosophy, but existing solutions available on the market are desperately slow and unable to meet the challenge of large scale, complex software development that is inarguably needed by today’s advanced SDV.
Accelerating Software Development
The new cloud-based features that the PAVE360 digital twin solution brings can deliver the simulation speed required for iterative software development. Software developers can operate at comfortable speeds close to real hardware meaning that software development and integration can proceed at pace way – before silicon is available. This significantly alleviates the burden of the integration storm typical of SDV development. Software integration can be achieved early leaving more time later for smaller bugs to be fixed when the real hardware becomes available.
Unit, integration, and functional regression testing can leverage typical CI/CD flows maintaining the speed of development with which software engineers are already familiar.
Meaningful virtual representation of physical hardware
There is no point in accelerating software development if it can’t be done in a deeply embedded environment. Software should go through minimal changes when moving from a digital twin based solution to the real hardware. So, software engineers need a virtual solution that is as close to the physical hardware as possible. Siemens EDA can bring software engineers closer to the hardware without sacrificing simulation speed. Virtual representations of peripherals are made available, embedded console functionality, embedded storage, networking, and interrupt architecture are all present. This makes the solution a very fast and close representation of the embedded environment to which the software is being targeted.
Easing the integration storm
By offering cloud-based access for accelerated software development on the PAVE360 digital twin solution, there is now the potential for distributed development of SDV software. Supply chain partners can collaborate with OEMs via the cloud to ensure alignment of both technical requirements and time scales. Industries that embrace the cloud progress innovation at high speed.
Making the change together – Siemens and Arm
The first architecture supported on this early access accelerated solution in the cloud is the Cortex-A720AE CPU just launched by Arm as part of their latest Automotive Enhanced line up.
We map our modelling and simulation technology to cloud resources in order to optimize for time or cost, based on customers’ needs. We have optimized our Cortex-A720 AE based modelling environment by mapping this technology to the highest performance cloud resource configurations available. We use our close relationships with AWS and Arm to gain access to supporting technology, achieving the highest levels of performance
David Fritz VP of Hybrid and Virtual Systems, Siemens EDA
Supporting the market introduction
Developers and ecosystem partners can get started on SDV development on the latest automotive CPU architecture right after the launch of the IP. This means software can be completed, integrated, and tested incredibly early so when the hardware is available the debug burden is significantly smaller. This should result in the latest technology making it into next generation SDVs much earlier than previously possible.
Partnering with Arm to support the market introduction of the Cortex-A720 AE CPU is a strategic step forward. Early software development by Arm’s ecosystem partners on the latest generation of Arm IP is critical to adoption by SoC developers, Tier 1s, and OEMs. By adding accelerated simulation support for Arm’s Cortex-A720 AE to Siemens EDA’s already wide portfolio of EDA solutions, customers can be served by one solution stretching from early software development right through to silicon and systems validation. While software developers can focus on fast, iterative software development, system and SoC designers can leverage Siemens EDA emulation products later for accurate validation of this newly developed software.
Jean-Marie Brunet, Vice President and General Manager, HW Assisted Verification, Siemens EDA
Come and see Arm Cortex-A720 AE IP in action on PAVE360
Siemens EDA has worked together with select Arm ecosystem partners to showcase “Shift-Left” at Embedded World 2024 in Nuremberg April 9th-11th. In Hall 4, booth 4-64, you will see PAVE360 being used for early software development on Arm Cortex-A720 AE IP . Want to find out more about PAVE360 and our Arm collaboration? Then talk to the experts.
