Recently, I received an email from my old chum Dave Millman, who is CEO of SalesDev.Global. Before we plunge headfirst into the fray (this column) with gusto and abandon (and panache, of course), I would first like to note that I only just realized that there even was a .Global generic top-level domain (gTLD).
Historically, the collection of generic TLDs commenced with .com (commercial), .edu (education), .gov (government), .mil (military), and .org (organization), with .net (network infrastructures) added soon after. How things have changed since then. Currently, there are more than 1,500 of the little rascals, including country code TLDs and ICANN-era gTLDs (there’s even a .accountant gTLD, for goodness sake, which should make the guys from Monty Python happy).
But we digress… In his email, Dave spake as follows: “Hi Max, I just got off the phone with a friend, Mohamed Kassem of efabless.com, who has done something truly revolutionary: enabled any digital designer to build an ARM chip (ASIC/SoC) without any tools or infrastructure. Google and ARM and other companies are all backing it. And nobody knows about it. ‘It’ doesn’t even have a name.”
Dave went on to say: “As we were finishing our call, I started thinking about who could tell such an enormous story and I remembered your book, Bebop to the Boolean Boogie (as you know, I bought my paperback copy in the 1980s and you signed it for me in 2006).”
But wait, there’s more, because Dave closed by saying: “You’re the only person I can think of who can tell this big story in the way it needs to be told. Please allow me to introduce Mohamed to you (he’s copied on this message). Mohamed was already aware of your work. When he saw a copy of Bebop, he immediately recognized the genius of your unique voice. You two will have an interesting phone call. Best regards, Dave.”
“You are making me blush,” I thought.
“Don’t stop,” I thought.
As an aside, on the off chance you are interested in this sort of thing — which you definitely should be if you are part of a startup — I would just like to say that I am very impressed with the SalesDev.Global website. Right from the get-go, the landing page tells you what they do and why you need them in clear, unambiguous terms. I know so many technical websites whose designers could learn from this approach instead of going all-out, Full Monty, “Death by Bullets” because they laughingly assume you will be delighted to see a list of “50 things we think you will want to know about our product” (you really don’t).
But we digress…
Suppose you are a small company and you want to create your own application-specific integrated circuit (ASIC). If this contains one or more processors and some memory and suchlike, you may also refer to your ASIC as a System-on-Chip (SoC). (See also What the FAQ are ASICs, ASSPs, SoCs, SOMs, etc.?) There are all sorts of considerations involved here — you need some way to capture the design at the register transfer level (RTL) using an appropriate hardware description language (HDL), such as Verilog or VHDL. You are then going to require a bunch of electronic design automation (EDA) tools to analyze, synthesize, and verify your design. Ultimately, you are going to need to work with a silicon chip manufacturer, for which you will require access to their process development kit (PDK), which is the piece of the puzzle that lets you map your design onto their silicon.
If you travel the traditional route using proprietary tools and flows — and if you are using the latest and greatest process nodes — then all of this is going to cost you a fortune (it’s like they say, “It’s easy to make a small fortune making your own ASICS — all you have to do is start off with a large fortune!”).
This is where a number of players leap onto the stage with a fanfare of sarrusophones. Why sarrusophones? Because “Trumpets are so 20th century, my dear!” (In the same way that a saxophone is reminiscent of an ophicleide with a clarinet reed, so is the sarrusophone redolent of an ophicleide with an oboe or bassoon reed. You’re welcome.)
First, we should make mention of the non-profit Free and Open Source Silicon (FOSSi) Foundation, whose mission it is to promote and assist free and open digital hardware designs and their related ecosystems. In addition to hobbyists and academics, the FOSSi Foundation is of interest to an increasing number of companies that use open source IP blocks and make their own development open source.
Next, we have efabless. Mohamed Kassem, who was introduced to me by Dave Millman, is the co-founder and CTO of this company, whose mission is “To bring back the ‘two guys in a garage’ to the semiconductor industry.” If you are looking to have a custom design created for you, you can connect with the efabless network of design firms and professionals to design and deliver a custom ASIC/SoC for your product. Alternatively, if you wish to create your own design, you can use the efabless cloud-based platform — without any upfront costs or commitment — to develop an ASIC/SoC for your own use or deliver it to your customer.
We also have a little company you may have heard of — Google — which seems to have something to do with almost everything, such as the Google Summer of Code (GSoC), which is an international annual program in which Google awards stipends to students who successfully complete a free and open-source software (FOSS) coding project during the summer.
Last, but certainly not least, we have SkyWater Technology, which is the only U.S.-owned and U.S.-based pure play semiconductor foundry — also a DoD-accredited Trusted supplier — specializing in custom design and development services, design IP, and volume manufacturing for integrated circuits and micro devices.
Earlier this summer, Google and SkyWater announced the first manufacturable, open-source process design kit in the form of the SkyWater PDK (accessible via Github), which allows users to produce chips at the 130nm node (although there are other open-source PDKs, they either aren’t manufacturable or they support only older 0.35/0.5 micron nodes).
Now, your knee-jerk reaction may be to say, “The 130 nm node? Is anyone still using that?” Well, yes, they are, actually, because it’s well known and stable and low risk and affordable. Also, any ASIC/SoC you create at the 130 nm node is going to offer you a much better value proposition than implementing the equivalent functionality using a bunch of off-the-shelf components.
The cool news is that efabless is collaborating with Google and SkyWater to accelerate design using the SkyWater PDK. The folks at efabless say they will make design for this PDK simple and affordable by integrating resources on the efabless cloud-based design platform including:
- An open source-based end-to-end ASIC design flow featuring OpenRoad, Electric, Magic, and others.
- The open source striVe family of full ASIC references designs.
- A marketplace for monetizing your IP or designs.
Even cooler news is that efabless is managing an open source shuttle program sponsored by Google. (Since chip fabrication costs are extremely high, it makes sense to share mask and wafer resources to produce designs in low quantities, and this concept is referred to as a shuttle program).
And the really thrilling news is that the first such shuttle will provide 40 project slots, free of charge, to any fully open-source design. I don’t know about you, but I’m almost giddy with excitement. I can’t wait to see the innovative designs that come out of this program. How about you? Do you have any thoughts you’d care to share on any of the topics present