Live data from Hacker News

Purdue Starts Comprehensive Semiconductor Degree Programs in U.S.

eetimes.com

121–130 of 165 posts

Re: Purdue Starts Comprehensive Semiconductor Degree Programs in U.S.

#121

Earlier quoted context omitted.

I worked at Intel, and with process/fab people for a number of years. Everything you're seeing in the comments is true. Compensation is not that bad. Clearly, it'll not pay SW salaries - no engineering does. But if you're a fab person, you'll work long hours, be on call often (and you will get woken up often), and eventually will own a tool that you'll be responsible for, even when not on call. Lots of abusive and pa…

> Clearly, it'll not pay SW salaries - no engineering does. For some reason, I assumed that in order to be a chip designer you had to be as good at programming as a developer. Then I became friends with two intel chip designers/engineers and was surprised that neither knew much of anything about programming, despite both having multiple PhDs in their fields. That's when I began realizing that engineering and programm…

It does kind of depend - I've made a career of being both: being the log designer that understands software (and makes sure the registers make sense etc) and/or the software guy that understands the c hips (and can read the verilog and find the bug) etc etc

Both require deep domain knowledge that doesn't always overlap - but being able to do both makes you more valuable

Re: Purdue Starts Comprehensive Semiconductor Degree Programs in U.S.

#122

As someone working full-time in Canada, wish I could roll back my life 10+ years and get into one of the semiconductor schools. Good luck to all who are admitted into this program, looks very interesting. Link to online degree: https://engineering.purdue.edu/semiconductors/degrees#online...

Warning: nerd trap. Low wages, horrendous deadlines (tapeout), hard work. Perhaps this has changed, given that we're in a semiconductor boom cycle now, but I have my doubts. When SMIC started trying to poach TSMC talent, the response was not to make pay competitive (or even half decent), let alone in line with the massive geopolitically-relevant value being created. No, the response was a mask-off legislative crackdo…

I suspect the reason is that semiconductor industry relies much heavier on captial (comparing to talents) than software industry given the same level of risk. One or two talented software enginners can bootstrap a company with reasonable amount of success that can be a good indictor of future larger success. You can't do that in semiconductor industry.

Re: Purdue Starts Comprehensive Semiconductor Degree Programs in U.S.

#123

Ha, I'm interning with their chip design team right now. They've done 7 tapeouts but there's a lot of mistakes and other issues down in the plumbing. I wouldn't say their program is significantly better or worse than any other, and my understanding is this focus on a "Semiconductor Degree" is just a reshuffling of courses already offered in EE and ECE degrees. Certainly there are teams far ahead of Purdue. The progra…

Is OSU Oklahoma or Ohio?

Oklahoma. I'm actually working with a kid from Ohio State who's constantly ragging on their program for not even teaching Verilog much less how to build an ASIC.

Purdue's program is ok, but it's a veritable bastion of academic excellence compared to schools that have no staff that have ever performed a tapeout, which is most of them.

Electrical Engineering departments have tons of faculty who know what they're talking about. CS departments have people who mostly kinda sorta know what they're talking about (even if they're 10 years out of date, which is an eternity in CS).

I would wager most schools don't have a single faculty member who's performed a small tapeout, much less a big project for an industry leader. Actual ECE, VLSI, the building of chips, is a wasteland on the academic side of the house IMHO.

Re: Purdue Starts Comprehensive Semiconductor Degree Programs in U.S.

#124

As someone working full-time in Canada, wish I could roll back my life 10+ years and get into one of the semiconductor schools. Good luck to all who are admitted into this program, looks very interesting. Link to online degree: https://engineering.purdue.edu/semiconductors/degrees#online...

Warning: nerd trap. Low wages, horrendous deadlines (tapeout), hard work. Perhaps this has changed, given that we're in a semiconductor boom cycle now, but I have my doubts. When SMIC started trying to poach TSMC talent, the response was not to make pay competitive (or even half decent), let alone in line with the massive geopolitically-relevant value being created. No, the response was a mask-off legislative crackdo…

I worked in a semiconductor school at a tenant organization. When the crunch was on, the parking lot was full at midnight before Christmas.

Pay was not so great as judged by the cars. The other thing was that the workforce was split - you were either old or young.

If you want to make money in semiconductors, find a place they are building fabs and become a plumber or electrician. Those are the folks who make money, and often there’s security clearance requirements that adds job security.

Re: Purdue Starts Comprehensive Semiconductor Degree Programs in U.S.

#125
I have fond memories of my semiconductor physics classes. The were not easy whatsoever, rather intense.

As cvccvroomvroom had mentioned, I had my VLSI chip fabbed at the MOSIS facility in Westborough MA, during my EE tenure at UMASS Amherst in the 1990s as part of a state-wide chip design competition (yeah - didn't win..). There was much activity on VLSI and computer system design at that time, things have certainly changed in the US.

I recall many entries were implementations of various (and obscure) mathematical algorithms, mostly in the frequency domain (hence DSP chips), that tried to solve problems that are now trivial in SW today.

My school did win occasionally, others were mostly MIT and WPI - I think in a large part due to the excellent professors at all competting schools and the work ethic of the students.

My final submission was the design for an Arithmetic Fourier Transform processor - trying to remove the need for expensive ROM by aliasing the signup up-frequency - which I did the barrel-based ALU for, and thank the grad students for including me in the project.

cvccvroomvroom: https://news.ycombinator.com/item?id=31756704#31760353 MOSIS: https://en.wikipedia.org/wiki/MOSIS Barrel ALU: https://en.wikipedia.org/wiki/Barrel_shifter (my ALU did much more than multiply/divide..)

(edit - remove double paste)

Re: Purdue Starts Comprehensive Semiconductor Degree Programs in U.S.

#126

Lynn Conway, co-author along with Carver Mead of "the textbook" on VLSI design, "Introduction to VLSI Systems", created and taught this historic VLSI Design Course in 1978, which was the first time students designed and fabricated their own integrated circuits: >"Importantly, these weren’t just any designs, for many pushed the envelope of system architecture. Jim Clark, for instance, prototyped the Geometry Engine an…

The Mead Conway book is absolutely beautiful, I have like 3 copies.

A copy is available here https://ai.eecs.umich.edu/people/conway/VLSI/VLSIText/PP-V2/...

Re: Purdue Starts Comprehensive Semiconductor Degree Programs in U.S.

#127

I have fond memories of my semiconductor physics classes. The were not easy whatsoever, rather intense. As cvccvroomvroom had mentioned, I had my VLSI chip fabbed at the MOSIS facility in Westborough MA, during my EE tenure at UMASS Amherst in the 1990s as part of a state-wide chip design competition (yeah - didn't win..). There was much activity on VLSI and computer system design at that time, things have certainly…

Neat, what was your dev process like? Are you still in contact with those grad students?

Re: Purdue Starts Comprehensive Semiconductor Degree Programs in U.S.

#128

Skimmed the article; what's the difference between this program and what's already taught in electrical and computer engineering classes? I studied ECE for my BS and MS and there were already a couple of chip design-related courses: digital design, VLSI, FPGA, semiconductor physics, IC fabrication. I guess more specialized coursework on verification or manufacturing would have been nice, but I don't think that would…

Which company did you work for?

In chronological order: a fabless IP company, FPGA company, FAANG company. I've since taken a hiatus from chip design.

Re: Purdue Starts Comprehensive Semiconductor Degree Programs in U.S.

#129

Skimmed the article; what's the difference between this program and what's already taught in electrical and computer engineering classes? I studied ECE for my BS and MS and there were already a couple of chip design-related courses: digital design, VLSI, FPGA, semiconductor physics, IC fabrication. I guess more specialized coursework on verification or manufacturing would have been nice, but I don't think that would…

There is a big difference between chip design and running a factory...which generally isn't taught anywhere. When I got my ChE degree many years ago we had like 3 or 4 courses for semiconductors and in one course we learned about the manufacturing aspect. Even then it was super outdated with respect to what I actually dealt with when working as a Process Engineer at a fab.

That makes sense. It kind of seems like the new program will be sort of a mish-mash of ChE, systems eng, and ECE (i.e. what it takes to actually run a fab).

In fairness, I think overall there are no curricula specifically about how to run a factory. Most engineering degrees focus on design and theory rather than industrialization. For example, there are no classes on how to run datacenters for computer science majors.

Post reply on HN