Multicore architectures have opened up a new challenge for embedded developers: how to effectively tap the pure processing potential of multicore platforms. Multicore platforms require software engineers to spend significantly more time thinking about software architecture. Optimizing the performance of multicore processors means embracing parallel processing, which can be a challenge given the long and successful history of single-core systems that are largely self-synchronizing. A related challenge is that the majority of the networking installed base is still operating on a mixture of single-core processors, ASICs and DSPs. Developers have the enormous task of effectively migrating millions of lines of code to multicore architectures.

So what development considerations matter when it comes to addressing these challenges and unleashing the power of embedded multicore technology in your applications?

Comprehensive Virtualization

Virtualization is one approach to simplify the multicore programming challenge. Stated simply, virtualization is typically a trusted layer of software that manages resources and access to them. It dedicates resources to cores, as appropriate, when the system is initialized and manages access to shared resources during system runtime.

As part of our Multicore Platform, Freescale is implementing an embedded hypervisor in the processor core to enable a more streamlined implementation with less software overhead. This is made possible with the Power ISA™ embedded hypervisor privilege mode. Through this embedded hypervisor, Freescale’s Multicore Platform will provide hardware mechanisms that help to ensure cores only access the resources (memory, peripherals, etc.) that they are designated to access. In addition to this embedded hypervisor, Freescale’s multicore solution will include data path accelerators that inherently virtualize access to network interfaces by decoupling them from the cores. Any core can send or receive packets on any network interface without the need for inter-core synchronization in software.

Freescale is also developing a lightweight virtualization layer that will be available to the open source community, and we are supporting community projects like the kernel virtual machine (known as KVM) and Power.org’s new eHYPE technical subcommittee, which will be addressing this issue with more advanced and sophisticated Linux® solutions.

Advanced Simulation

With multicore architectures, advanced simulation technology is also becoming paramount. The simulation environment can present a programmer’s view of the system, which accelerates the development cycle, provides more flexible debug capability and improves the overall quality of software. Freescale is partnering with Virtutech to offer a virtualized software platform which enables pre-silicon and post-silicon development. This hybrid functional and cycle-accurate simulation tool gives developers the speed of a functional model for early development and the precision of cycle-accurate models to identify performance bottlenecks. This combination enables fast hardware concept testing and evaluation, as well as performance verification.

Debugging Visibility

Multicore SoCs are complex devices with multiple cores, hardware accelerator blocks, memory, caches and network peripherals. As such, debugging solutions should strive to provide greater system visibility to all of these elements. Freescale is engineering the necessary hooks in our Multicore Platform to enable advanced debugging of the platform, and we are also working in tandem with industry leading vendors to assure the availability of tools that can take advantage of these features. These hooks include, among others, trace capabilities on the cores, watchpoint triggers, cross triggering capabilities, performance monitoring capabilities on the cores and SoC and the Power ISA defined debug features.

For more information about Freescale’s Multicore Platform, please visit www.freescale. com/multicore.