Ask HN: How to get into the chip industry?

This is like asking, "How do I get into the medical industry?" It's overly broad. There are lots of different jobs in the chip industry.

1. Device engineer - want to build transistors? Grab a physics/chemistry masters and /or PHD. Go to work at a foundry house like TSMC, Samsung, Intel, SMIC or Global Foundries.

2. Analog circuit design - grab a masters in EE (w/ circuit design specialty).

3. Layout - the person who translates the analog circuit design into the physical drawing that ultimately gets turned into a photolithography maskset. You can get an associates degree (maybe not even that much) at a local community college - learn the cadence/mentor EDA tools.

4. Verification - learn verilog HDL and computer architecture. A bachelors degree at a good Electrical Engineering program will get your foot in the door.

5. CPU design/architects - masters/phd in computer architectures.

6. Software - lots of different jobs here. Firmware guys write low level boot-rom code. Some guys focus on linux drivers and kernel porting to company xyz's latest chip. Some software guys focus on specific IP blocks of a chip like maybe working around bugs in the latest GPU IP integrated into an SOC like that used on the Raspberry Pi.

7. Test engineers - the guys that write test patterns to test a chip after it has been fabricated and packaged. You could go software or hardware for these jobs. Bachelor degree needed.

8. Packaging engineer - the people that design the packages that the bare silicon gets mounted to. All kinds of folks in this field - physics, mechanical engineers, materials guys. This is a fairly broad field.

9. Hardware engineer - designs the pcb's that the chips get mounted too. Your "classic electrical engineering" job IMO. Bachelors in EE with a focus on electronics and circuit design.

In general you'll need a degree from a decent engineering college to get your foot in the door. Software generally doesn't have this barrier. Look for internships at chip companies.

Relative to the software industry, the chip industry has a vast number of highly-obfuscated dependencies, nearly-infinite supply chains, and zero margin for error. Robust profitability, let alone innovation, is very difficult to achieve without monstrous scale. The majority of recent semiconductor industry gains have come only through consolidation, giving chip suppliers more bargaining power while letting them cut staff.

The only area in the semiconductor industry that I'd consider marginally worth getting into is analog or RF IC design, but that's predicated on lots of VC bucks chasing the last scraps of quick-win innovation (mostly in the Bay Area).

So even before picking an area to focus on, I'd take a deep look at your motivation for wanting to get into the chip industry in the first place. As someone who's been in & out of the chip industry a few times, I can tell you that there must be hundreds of other industries that are more lucrative and rewarding than the chip industry, while still giving you whatever it is you're seeking.

Some of the most fun I have as a graduate student in computer architecture is trying to write the fastest possible code snippets for complicated chips. Working your way "down" from C/C++ is a great place to start, assuming you're familiar with either of those languages.

Take some collision detection algorithm for a game or some linear algebra kernel used in your machine learning model of preference.

If you're referencing a library that's already high-performance (e.g. Intel's MKL), what does the assembly look like? What's happening to your caches as the assembly is being executed? What are the trade-offs your compiler has to make when it comes to optimizations? How does Intel's compiler output differ from Clang? I mention these compilers in particular because they provide optimization reports.

My advisor guided me through "Computer Architecture: A Quantitative approach" by Hennessy & Patterson and it's been very eye-opening to pull back the curtain on the last logical layer of "magic" I had never fully understood. There's an enormous of amount of circuitry in the latest chip designs dedicating predicting the outcome of pending work so that we can "hide" the latency introduced by having to wait for data to arrive into the CPU core all the way from RAM. The way an instruction set architecture and compiler (e.g. x86-64, AArch64) unite to abstract away the work underneath is really one the coolest technical feats of the 20th century. Hopefully this inspires you (or anyone reading this) to go explore it. I think if you can document your exploration and communicate your findings effectively then that would be a great first step towards getting your bearings at a high level of how the "chip industry" operates.

As others here have mentioned, the chip industry is an enormous field and you could build an entire career around just the physical/mat.sci. side of fabrication for these chips. The physical side is still somewhat foreign to me, but I hope to explore it more in the years to come.

EE Master here, this is a wide topic but usually from my experience (friends around who work in chip industry, ie TSMC, MediaTek).

Generally, you have the hardware design (Layout designer, R&D, Process engineer ) this typically need a Master and mostly PhD in EE or related fields. From what I heard in Taiwan, TSMC only accepts Master/PhD for these position and internship.

Then you have the interface side which writes firmware codes for these chips ( SSD controler ). These are more like your typical software engineering, you will need at least BC degreee and well verse in C/C++. But usually it's hard to get in with a BC degree unless you are recommended by one of the employee.

If you don't have any of the mentioned degree, the fastest way is to get one. It is really hard to learn yourself since this is a highly automated industry with most of the work available in maintainence and cutting edge R&D.

Otherwise, your job scope in chip industry is limited to a few ( Firmware is a good start )

ASML's architects is a relatively new book on the photolithography part .. it is on my to read list but kinda expensive. There are a couple of nice Youtube videos - recordings from the computer history museum on the history of semiconductor fab processes. Might scratch your itch to understand how it works (it started out "simple" but today is insanely complex .. the physics phds are really pulling their weight on the latest process nodes). The few responses you have gotten here are gold. Only thing not mentioned is FPGAs .. if you are a software person, you can learn a ton by playing around with cheap FPGAs. Oh .. if you meant you just want to learn high-level electronics, I have to recommend an incredible course on Udemy by Andre Lamothe. Good luck.

I work in this field

The pay is poor compared to SaaS/AIML/etc

Hardware folks (and firmware) are always asking how to retrain and move into mainstream SWE. Its not worth it.

For learning below courses and books are very helpful:

1) Online courses: Computation Structures: Parts 1, 2 and 3 - all course materials available on EdX MIT as well as MIT OCW

2) Books: a) Computer Systems: A Programmers Perspective and b) Computer Organization: Hardware/Software Interface

3) Practice Verilog HDL programming for small/medium complexity building blocks such as FIFOs, RAMs, bridges, etc using free simulators like Xilinx ISIM. Courses such as Coursera's Hardware Description Languages for FPGA Design are a good starting point. Lookup most starred repos on Github in verilog and systemverilog languages.

Depends on what you mean by "the chip industry"?

- Designing CPUs and digital logic chips? You'd start with a computer engineering degree focusing on computer architecture and VLSI design.

- Designing analog or RF chips? You'd start with an electrical engineering degree.

- Fabricating chips from bare silicon wafers? That's an intersection of chemistry, applied physics, material science, and electrical engineering.

Look up the course requirements for degrees at universities with engineering programs and you can see what courses and textbooks are used for each.

I have no context into your background knowledge, but I'll assume you know how to write some software.

Coming from a software background, my recommendation would be to come from above and work down towards the chips. The perfect route to take is to work with a soft RISC-V cpu running on an FPGA with a totally open toolchain. By utilizing open tools, you have the agency to both modify the tools for your own use, gaining knowledge into how the tools work as well as participating in the technology graph. As soon as you add even a minor feature to the tooling, you then have the ability to add features for other people. This makes you valuable and will open other opportunities.

Get a Lattice FPGA supported by Yosys [1] and learn Chisel [2] and Verilog [3] while working through Nand To Tetris [4]

Once you understand the hardware enough, those soft designs can start to be turned into hard designs using the Skywalker PDK [5] and get your Open Source designs made for free.

[1] http://www.clifford.at/yosys/

[2] https://www.chisel-lang.org/

[3] https://hackaday.io/project/160759-nand-to-tetris-in-verilog...

[4] https://www.nand2tetris.org/

[5] https://www.theregister.com/2020/07/03/open_chip_hardware/ https://github.com/google/skywater-pdk

> How to get into the chip industry?

If your idea of a chip industry is that part of the industry that physically makes chips... then most likely you don't.

Very likely more than half of people who venture into this career, and successfully graduate will never see any work in the field.

Just like in the medical field, a tiny amount of people end up being practicing doctors, and more than half end up in "near med" fields, from pharmacists, to nurses, and .. to an analogy of quack doctors despite high salary potential for the most lucky.

Let's begin:

1. First, be born Taiwanese

2. Second, finish EE PhD at around 15-20 years ago, at maximum

3. Third, get through the glut of other young engineers who ventured into the field at the crest of the dotcom boom wave

4. Fourth, be lucky to selected for hire upon completing a long unpaid internship

5. Fifth, spend your first 3-5 years in a role of office coffee porter, and such

6. Sixth A, be lucky to be finally selected by somebody for actual RnD work

6. Sixth B, go and work as a fab tech at $2000 USD a month

Your question has been well answered. I'm surprised you even ask it, because the answer is so obvious for such a poorly formed question: get a graduate EE degree.

I would be interested to know what you are doing now, that you don't like, and why you find the "chip industry" attractive?

For someone competent at coding, a good way to get into the chip industry is as an embedded programmer, writing firmware. These are usually in high demand in the industry.

If you do want to go straight on, FPGAs are a good and fun option.

Turn into a "technovangelist", advocate https://www.yokogawa.com/yjp/solutions/solutions/minimal-fab...

FUCKING DISRUPT IT!

Profit.

Check out everything around RISC V. There the rabbit hole is as deep as it can get.

Start with potato chips. Then, go where the spud takes you. When you ace your yams, you know you've made it.