One of our mantras at QuickLogic has always been “software before silicon.” Why? Just as a computer is a useless collection of hardware without an operating system and applications software, the same is basically true for FPGAs without our customers’ designs mapped into our architecture. Sure, we provide a lot of clever integrated functional building blocks – some quite sophisticated – but deep down we appreciate the fact that you are the ones who “make the machine fly”.
Since, as you know better than anyone, end application designs can be extremely complex and take many months to develop or evolve, we know that it is important to provide you with a complete and highly functional set of development tools as early in your design cycle as possible. At that stage, it is critical for the software to be intimately familiar with the details of the device architecture, but having actual silicon is practically irrelevant. In fact, many customers tell us that their ideal lead time between first availability of development software and first available of silicon engineering samples is three to six months. While your particular needs may be longer or shorter, the basic utility of having software before silicon is the same.
The advent of eFPGA technology creates a new paradigm, however (and we don’t use words like “paradigm” lightly). Now SoC developers can also have post-manufacturing programmability built into their devices. So how does the “software before silicon” mantra relate to this new category of FPGA user?
In one sense it stays very much the same, and that is the case of the initial integration of eFPGA technology into a particular SoC. In that case the developers are probably still going to want to begin the FPGA portion of their design before their initial production run of silicon, as it is very likely that the initial FPGA design is critical to completing the overall application design which may be mostly implemented in ASIC gates.
In another sense, though, the paradigm changes. Developing SoCs is itself a long, complex, and expensive process. The allure of integrating eFPGA technology into SoCs is that the same basic piece of silicon can be used over and over again with the eFPGA portion being reprogrammed post-production to fix bugs, add new features, support evolving standards, or address different market requirements. In this case (and although we will have delivered the FPGA development tools already), our customers will continue to use their SoC for new designs, and so in a way they will have their silicon at the same time they need software. And that will be true over and over again with each new iteration of the eFPGA portion of their logic.
We’re perfectly okay with this, by the way, as it is just one more way for us to add value for our customers and ultimately that’s all that matters. So, our new mantra when it comes to eFPGA technology is “software before silicon, and then silicon and software together – again and again and again.”
2 thoughts on “Software Before Silicon?”
Show me the money now Brian!
If the lead times just from development software to silicon engineering samples is 3 – 6 months what might a lead time from a production agreement to actual shipping of finished goods ( a recognizable sale) be all in? Is it about a year? 18 months? Curious as we have heard great things about QUIK solutions from the Eos S3 to now the Artic Pro eFPGA. That there are still no confirmed available products with either Eos S3 or eFPGA inside is this a product of the lengthy semi cycle or more that OEM’s have not yet embraced the alternative technology.