3 ms·

You're not going to learn electrical and computer engineering by reading stuff on Sci-Hub. You might get some knowledge, but so much is just simply never publis

You're not going to learn electrical and computer engineering by reading stuff on Sci-Hub. You might get some knowledge, but so much is just simply never published that it's essentially tribal knowledge. Heck, even with professors walking you through every step of learning the topics, it's hard. As for computer architecture, OSU is not heavily into computer architecture. It's much more focused on custom accelerators and ASICs as well as semiconductor research. And it is recognized as one of the best ECE programs in the world with one of the main problems with it being that the CS department is comparatively weak. My graduating class, in a December, was still over 1,000 engineers. Spring graduation typically had 2.5-3.5K/yr.

kragen 5y ago
I appreciate the correction about OSU's EE department. I'm probably unfairly prejudiced against them because mostly I read in CS. I'd make a stronger statement than "You're not going to learn electrical and computer engineering by reading stuff on Sci-Hub." You aren't going to learn by reading. You have to actually do stuff in order to construct the knowledge inside your own mind, and as you say, that's hard. But the information is available, even if it isn't academically published; in electrical engineering, in addition to sci-hub stuff, you have Bob Pease columns, appnotes (especially the ones from Analog and Linear), The Art of Electronics, EEVblog, allaboutcircuits, Hackaday, Marco Reps, Camenzind's book, Jeri Ellsworth's Short Circuits, Don Lancaster's website, Forrest Mims's books, patents, Adafruit, etc. LTSpice is proprietary but gratis, while QUCS, ngSpice, and WR-SPICE are free software, and now you can get stuff mostly automatically fabbed by JLCPCB for ridiculously low prices. And with the rise of iceSTORM it's becoming practical to design gateware without signing a support contract with Xilinx. Just like the undergraduate EECE curriculum, a lot of this stuff is aimed at beginners (you can learn a lot by fiddling with Falstad's circuit.js even if it doesn't do Monte Carlo simulation), so it doesn't go very deep, but some of it is very much not. (Still, there is a lot of information that isn't publicly available at all, because it's a trade secret of TSMC or Lam Research or Intel. But the professors aren't walking undergraduates through that stuff either.) The huge thing about custom accelerators and ASICs and semiconductor research, though, is that you can't really do it in your basement. MPW/shuttle programs like MOSIS and CMP go a long way to making chip design accessible, but it's going to be tough to even run DRCs on your designs without proprietary software, and then it still costs you US$10k to get back 10 chips that don't work. Going to a university for EECE, even for undergrad, might give you the chance to go through that painful cycle enough times to actually get something working, and the mentorship you need to find out what you don't know.

kragen 5y ago

I appreciate the correction about OSU's EE department. I'm probably unfairly prejudiced against them because mostly I read in CS. I'd make a stronger statement than "You're not going to learn electrical and computer engineering by reading stuff on Sci-Hub." You aren't going to learn by reading. You have to actually do stuff in order to construct the knowledge inside your own mind, and as you say, that's hard. But the information is available, even if it isn't academically published; in electrical engineering, in addition to sci-hub stuff, you have Bob Pease columns, appnotes (especially the ones from Analog and Linear), The Art of Electronics, EEVblog, allaboutcircuits, Hackaday, Marco Reps, Camenzind's book, Jeri Ellsworth's Short Circuits, Don Lancaster's website, Forrest Mims's books, patents, Adafruit, etc. LTSpice is proprietary but gratis, while QUCS, ngSpice, and WR-SPICE are free software, and now you can get stuff mostly automatically fabbed by JLCPCB for ridiculously low prices. And with the rise of iceSTORM it's becoming practical to design gateware without signing a support contract with Xilinx. Just like the undergraduate EECE curriculum, a lot of this stuff is aimed at beginners (you can learn a lot by fiddling with Falstad's circuit.js even if it doesn't do Monte Carlo simulation), so it doesn't go very deep, but some of it is very much not. (Still, there is a lot of information that isn't publicly available at all, because it's a trade secret of TSMC or Lam Research or Intel. But the professors aren't walking undergraduates through that stuff either.) The huge thing about custom accelerators and ASICs and semiconductor research, though, is that you can't really do it in your basement. MPW/shuttle programs like MOSIS and CMP go a long way to making chip design accessible, but it's going to be tough to even run DRCs on your designs without proprietary software, and then it still costs you US$10k to get back 10 chips that don't work. Going to a university for EECE, even for undergrad, might give you the chance to go through that painful cycle enough times to actually get something working, and the mentorship you need to find out what you don't know.