25 Years ago, had I shared the prediction I am about to share now, I would have been laughed out of the conference room in any FPGA company. Not in a rude way, but with the sort of polite pat on the back you would give an intern who is just entering the real world from academia.
Schematic capture was the way ‘real engineers’ did FPGA design, inch thick data books outlined each and every detail of the logic and routing structures, and there was a sense of pride when someone completed a ‘boot camp’ on an FPGA vendor’s proprietary toolchain that often shipped on a stack of 3 1/2” floppy disks as long as a loaf of bread. We scanned newspaper ads for the latest Fry’s Electronics sales so that we could upgrade our desktop PC or Unix workstation with the fastest processor and largest amount of RAM available so that we could get our larger FPGA designs to place and route in less than a day.
Brian Faith, CEO, QuickLogic
Fast forward to today. We have come so far in many ways. For FPGA users, schematic design is a lost art, replaced long ago by languages such as Verilog and VHDL, and increasingly by languages like System C and Python. Data books are now merely electrons.
And yet in other ways we are still stuck in that time. Proprietary FPGA toolchains are the very definition of bloatware, still requiring very capable and expensive computers. Moreover, the deep experience an engineer has in using a proprietary toolchain appears to be a benefit, but that benefit is only an illusion – much like a mirrored room – one that appears to have endless capacity but is in fact very much constrained.
