BLOG

I always enjoyed writing, but my current journey started in September 2017 with writing on Quora. Credit where credit is due, I unexpectedly received a message in 2018 that I got a top writer badge on Quora. And looking back at my initial writing, perhaps only related to the niche topics VLSI and semiconductors, that was a surprise. So surprising, that it kept the ball rolling for a few years. Today, January 2021, I have over 4 million answer views on Quora. And a Quora space about “HW accelerators eating AI” with over 60.000 followers. A space is like a blog, for the people unfamiliar with the concepts on that site. In 2020, I started boosting my LinkedIn presence. So, January 2021, I have a semiconductor related page with over 10.000 followers.

Overall, I am thrilled that so many people are interested in ASIC and FPGA design. That makes me optimistic for the future of chips and silicon overall.

Here are a few recommendations and links to things I have written about:

ASIC design fundamentals every front-end digital chip designer needs to know.

1st post: What is synchronous design?
2nd post: How do we design an ASIC?
3rd post: What’s a clock?
4rd post: What are the important clock parameters?
5th post: Do we choose an internal or external clock?
6th post: How do we deal with multiple clocks?
7th post: How do synthesis and STA deal with clocks?

8th post: How do the back-end engineers deal with clocks?
9th post: ASIC clocks in FPGA prototyping (emulation)

Similar to the above ASIC design posts, I am starting a series of posts about verification. Soon, I will be adding specific UVM related verification posts, so bear with me.

In January 2021, I am working as an ASIC consultant, advisor and freelancer. My main customer today is Mythic AI, they demonstrated silicon that does real time openPose and Resnet-50 end of 2020. This startup, the era of analog compute, is a series B funded company with mixed-signal design expertise. So, I am extremely interested in chip design and the different tasks that are part of any chip design project. I do design, verification, synthesis, Static Timing Analysis (STA), Design For Test (DFT), gate level simulation, Formal Verification, ECO (netlist metal patches, spare cells), support for back-end floor planning, Clock Tree Synthesis (CTS) and congestion in Place And Route (P&R). Follow-up wafer test, packaging, packaged test, yield and yield optimisation. I have been Project Manager and Project leader for many years. In many different sectors, consumer, automotive, industrial and many more. In small startups and big companies and as subcontractor for big companies. I have met and worked with unknown but brilliant people. My mission is to share their stories and teach you a thing or two about good design practices. Most of today's tape-out cost explosion can be attributed to wasted time and resources. The race to the bottom for salaries and the "agentiation" of freelancers create havoc in the ASIC design world. Single skilled people cost less but have no knowledge of the cohesion and the sequential nature of IC design. The dino's from the nineties, like me, know the tricks we use in the concept and design phase to avoid issues in verification, synthesis, STA, DFT, gate level sims, ... That information worth millions for a $500mn tape-out project was lost over time. Todays agencies and recruitment firms make those millions and they don't care about quality, experience or talent stacked people. They are business people interested in ROI for themselves. And the companies using external recruitment know very well they don't get the best, they get the best the agency presented. Which is not the same as the best they found, they will obviously only present those that are interesting to themselves. And the company engaging agencies need obedient in-the-box thinkers even though their ads are claiming the opposite.

Long story short, I will be writing a lot more about Artificial Intelligence hardware, startups, employment, IC design and much, much more. But I will tell you the real story, because the marketing and sales hyperbole, some call it lies allegedly, you can find on the internet via your favourite monopolistic search engine.

Dilbert Creator, Scott Adams, has written many interesting books over the past decade. His main point is that we all should build a talent stack. To be able to eat fish and expecting an employer to give us a fish every month, is something the boomer and older generations enjoyed. But things have changed over the past decade. With covid, what was happening in 10 years from now, remote work, more flexible workforce and many other things, have happened in a time frame of just a few months. I am old enough to remember the days that all managers in semiconductors told me remote work wasn't possible. Even though ASIC designers work with exceed or VNC sessions on a remote server somewhere. Turns out everybody switched to remote work in 2020 just like that. And end of 2020, CEO's of major companies, like Apple CEO Tim Cook, were actually "surprised" they did much better than expected when they switched to work form home (wfh). So, back to the point I was making, a talent stack is much needed. To learn how to fish and have several ways to catch fish, is superior to waiting every month for a fish to be delivered to your home. Luck is simply an opportunity and a match in skill and expertise. Therefor, outside my regular job, I am deep into Raspberry Pi, Arduino and FPGA development boards. I build my own 3D printers and own a few original Prusa 3D printers too. And I am not afraid to criticise Prusa printers from the standpoint of a decade long ASIC designer. A free graduate that was put on a pedestal (Forbes 30 under 30) is probably very successful, but probably has the arrogance to go with it. I do feel that printers could have been much better, faster, safer and more efficient than what we have now. And don't get me started on the 10 week lead time on the Prusa Mini. That is just, what is the word I'm looking for? Nevermind.

I am also looking into 3D printer assisted construction of houses, but haven't written much about this yet. I plan to write about it on this blog when time permits. Not sure this is a topic you all want to read about, so comment or reply if that is the case. My wife's activity is real estate, property management and thus something we will pursue for the future for sure.

And obviously, I use Arduino and Raspberries to automate the irrigation of the lawn (we live in Portugal, so summer is hot and dry) and to automate other things as well.

On top of electronics, there are many other topics that interest me. Thanks to Star Trek Discovery (the spore drive) and Paul Stamets, I am interested in longevity and neurogenesis. Legal mushrooms like Lion's mane are my focus now and I am trying to grow them at home. Hericium Erinaceus is a mushroom (fungus) that is used for culinary reasons but is also used as a supplement to help regrow neurons. BTW, I hate the spell checker that insists on changing the word "neurons" into "neurone". An example of Artificial Intelligence that is so dumb one can't believe it is still out there. The focus of AI is clearly conversion of ads and to make advertising great again, instead of for the good of mankind. I am surprised how many entrepreneurs over 40 are into supplements and focussed on maximising their mental capacity. It is a thing nobody talks about in public. Nootropics for example. Of course medically sound and under supervision of a doctor (MD). I am expecting good things from supplementing a few mg home grown Lion's mane. Chain of custody, you know. I like my brain very much and I want to stay sharp as long as possible. Balaji (and Naval too) are two persons that believe that age is a disease that might be cured in the future. Like all cures, it triggers a lot of discussions. And I like to discuss rather than, (a) fall on my knees and be 100% sure I will live forever or (b) mock the initiative and scream about overpopulation. Those are the two extremes and in between there is a honest pro and contra discussion I hope we can have.

As you might have guessed, I am not particularly good at writing. I write like and engineer that writes for an audience of engineers. And I am fine with that since the engineers that read me like it and upvote it regularly. That makes me happy and as long as people think it is interesting, I will continue writing.

I am open to connect with any semiconductor profile on LinkedIn, btw, so feel free to invite me. That way we create more opportunity for both of us on LinkedIn. Connections increase the reach of your network. For jobs and influence, it is always a good idea to grow your network. Most semiconductor work is network based. Shared through a network, shared towards the biggest accounts, ... Your network and reach matters a lot.

I would also be honored if you would follow my company showcase page on LinkedIn. Thank you for your support!

Here is a screenshot of my content views and a few of the spaces I am active in:

Here is an overview of the all-time statistics for my answers on Quora:

Some statistics for the space "HW accelerators eating AI":