Virtual Reference Platforms: No Need to Reinvent the Wheel
I have to admit that it’s pretty rare that I do much programming anymore. I spend far more time creating documents, presentations and talking to customers now than anything else. All of these tasks though do share a very common trait: they’re all greatly simplified when you can reuse content from a previous project or borrow it from someone else who has done a similar task. It’s very rare that I have to create a presentation completely from scratch. I can pretty much always find content that I can reuse. Back in my programming and design days, I never created a new design module from scratch. I took a previous design and stripped out the stuff I didn’t need and reused what I could, even if it was just the declarations and formatting.
Virtual platforms, and tasks performed using virtual platforms, can of course be similarly streamlined. Before standards like TLM-2.0 existed however, it was pretty tough to reuse much from project to project since there always seemed to be enough differences in how the design blocks communicated to make it easier to rewrite a block in many cases rather than try and debug why it wouldn’t work in a new environment. As the leading provider of cycle accurate models and model creation technology, Carbon of course is very familiar with this problem. Our blocks are reused in a variety of environments and we’ve spent years working on technologies to streamline that reuse process. Reuse however isn’t just a model problem. A virtual platform is only interesting when it represents an entire system or subsystem. That necessarily means bringing in other blocks, configuration information and of course, software. Models are great but in order to streamline reuse you need to supply more content.
Carbon’s recent virtual reference platform announcement is a huge step forward in enabling this reuse. Our virtual reference boards contain a processor and memory subsystem and also include some sample software to get the user up and running quickly. We’ve been testing these with customers for a while now and I’ve been impressed with just how quickly they’re able to get up and running not only with the design we provide them but also on customizations to this design to represent their own environment.
Just last month, a Carbon user with no ARM programming background took one of our A9 reference platforms from Carbon IP Exchange, got it up and running quickly and then started modifying things to fit his own environment. He used the supplied software as a starting point and quickly ported over Dhrystone benchmarks, he then replaced the supplied fabric with one which he compiled using Carbon Model Studio to represent the system configuration from a previous design. Finally, he replaced the generic memory with a memory controller and memory downloaded from Carbon IP Exchange.
Even better, these results aren’t at all atypical. We’ve seen them several times now at multiple customers. Our current virtual reference platforms are fairly straightforward and focused around a single processor. As you might imagine however, with the substantial and increasing amount of IP on Carbon IP Exchange, we’ll be rolling our more complex platforms in the upcoming months.
Life is always easier when you don’t have to reinvent the wheel every time you start a new project. Our virtual reference platforms aim to get you up and rolling in your project faster than ever before.
