UC Berkeley (where I did my CS degree) follows this model after admission. There's CS61A, an introductory class for people with some math/computers background, and CS3, an introduction for people coming in with no background. Successful and interested CS3 students are then funnelled into CS61A.
Yes but minus the math; what math there was in 61A was easy. 61A didn't really expect much prerequisite either, but then it took off quickly. A lot of work but not intellectually very hard. It was then taught in Scheme which no one knew. Scheme is an elegant language which we never used again. SICP was the course text which no one ever read.
What I really remember was EE40 taking like a week to explain Ohm's Law. I thought, Ok, I think I've got this class. But then it just went exponential and then the prof just dumped a bunch of chapters on us at the end that he didn't cover and said that'll be on the final. EE40 covered about a third of 105. EE40 just had an insane amount of material and labs. FWIW, I think 16A+B covers better material than EE40 did although I have a soft spot in my heart for what was EE20.
BTW, Calculus 1A at Berkeley, if you didn't take AP Calculus in high school, was brutal. Pointlessly so, since most math in non-math courses was pretty straightforward.
But all the freshman courses at Berkeley were hard. You either got up to speed or you got out. Upper division was even harder, but you were prepared by then.
61A was easy in terms of math, but hard in terms of everything else. Every student I met who had never programmed before taking 61A found it intellectually quite challenging, both before and after the switch to Python; even without math, thinking in terms of computation and execution is not natural.
Given that second-hand experience, I found the excision of all of SICP's math from the curriculum to be quite a good choice, as adding advanced math as another prerequisite would have raised the barrier to entry even higher.
(Side note: your description of EE40 sounds like a professor who didn't schedule well, not intentional difficulty.)
(As you know) most of the lower div EECS courses inherit slide decks and the coverage is set by the department. CS70 has a wonderful set of notes which should be a book. I regularly go through the CS61C, CS152 and CS252 decks; those are great Berkeley classes. CS162 has a good set and then Kubi has his which even better. Hilfinger's CS61A notes were quite good. The deck the EE40 prof used was based on a previous semester and it was pretty good. It's just that the class is an insane amount of material and work.
There was an intent behind the work. All of the lower div classes are hard with a massive amount of content. But you can show up a Berkeley from a substandard high school on the wrong side of the tracks and blow it out of the water; a buddy of mine was like that and went to Columbia for his PhD. It's very simple. You just have to work. Immediately. Day one. I don't think Berkeley EECS rewards creativity. Indeed I think EECS punishes creativity. But EECS teaches its students to work their asses off, regardless of whether they came from Lick Wilmerding or Skyline.
Don't know if the workload was all that hard compared to, say, the experience of my MIT friends, but the level of intellectual rigor and methodical thought was high.
