All feedback below from CSE major Sam Courts, from 2013-14.
General Info
Vorlesung/Übung: For a 9 credit CS class, this is usually two two-hour lectures and one two-hour tutorial a week.
Seminar: I didn't actually take any seminars, but my understanding is that they meet once a week and have minimal readings/homework or perhaps a small group project. Usually only worth 3 ECTS credits, but be wary, they may be more time consuming than you expect.
Blockvorlesung: This is an intensive course offered at the end of the semester, after all the other lectures have finished. There are two week seminars and labs offered in this form, but the course I took was a 9 credit lecture. One block consisted of an hour of lecture, an hour and a half of working on exercises in groups, and a supervised hour going over those exercises in small groups. There was a block from 9-12:30 M-F and from 2-5:30 M-T & Th-F for a total 9 blocks or 31.5 hours a week. Do not plan on working on other things during this time! I had a compiler to finish writing and two other exams to take during this time and I strongly discourage following this example. On the plus side, if you don't have other things going on, it is fairly manageable. You spend so much time in class that there is no real outside work or studying expected. The course pace is pretty overwhelming for most students, but being used to a quarter system, I actually preferred learning everything in a shorter period of time than the more relaxed semester schedule.
Meeting Times: Make sure to check the individual course websites for the actual schedule because most courses don't meet for the first scheduled lecture. There may also be either an s.t. or a c.t. next to the meeting times. The s.t. means sharp time and I don't know what c.t. stands for, but it means not sharp time. Basically 10:00 s.t. means the class starts right at 10, whereas 10:00 c.t. means the class starts at 10:15. Most lectures are c.t., but most exams are s.t..
Registering for Courses: You may hear a lot of conflicting information about how or when to register for a course, and for the most part, that's because people won't know that international students can't register through HISPOS (the normal way of registering). In general, you don't need to do anything for the first few weeks of lecture. Just attend the courses you're interested in and stop going to the ones you decide not to take. Lectures may have a course specific registration during the first week that gives you access to the homework and puts you on a mailing list. This will be done through the course website, and you should have no problems registering like everyone else. This is also non-binding. The real registration is through HISPOS and must be done something like 6 weeks into the semester or two weeks before the first exam. All you have to do though is email your professors and say you're an international student and want to take the exam for a grade. You should do this before the exams, but it isn't really critical that you do it by the date that everyone else registers.
Specific courses
Algorithms and Data structures:
First student: 9 credits, Blockvorlesung. Grades were 40% midterm (no re- midterm) and 60% final exam or 100% re-exam. If you like the theory side of CS, I would highly recommend this course. Probably one of my all-time favorites. Professors and TAs were excellent, the material was challenging, but well taught and understandable. The material was both practically and theoretically relevant. The UdS equivalent of CSE 332 is much more theoretical than ours, so I did a times feel a little behind, but not in an insurmountable way. The exercises and exams required a lot of lateral thinking which was difficult, but also fun. Keep in mind that it's a block course, so you won't have a lot of free time while taking it.
2nd student: Class taught by Martin Hoefer / Michael Kerber: Overall, one of my favorite courses of all time. Good pacing (hard but doable), great content, teachers were great at imparting information. Assignments were optional, but you MUST do them in order to succeed. Course was entirely theory-based (zero practical implementation), which I unexpectedly loved.
Cryptography: 9 credits, Vorlesung/Übung. Grading was based on one midterm (with a re-midterm to improve your grade) and a final exam (also with re-exam). Weekly homework was optional and not graded, but I recommend doing it. I had a lot of fun with this course and didn't think the work load was too overwhelming. Much more theoretical than practical, but I felt sufficiently prepared. A lot of students complained that it was one of the worst lectures they'd taken, but don't think it was that bad. Lectures were a little hard to follow at times, but questions were encouraged and answered well. The lectures also closely follow the book (available in the library), which explains things really well. The exams were tough. By which I mean, 75% of people got less than 30/80 on the midterm. They don't curve and failing is not a big deal, so just be aware that good grades are not a given. Exams did improve throughout the semester though, so this may just have been because it was Prof. Schröder's first time giving the lecture.
Compiler Construction: 9 credits, Vorlesung/Übung. Grades were 50% final exam (with re-exam) and 50% compiler project. This course is a lot of work, but in a rewarding way. The project may vary by semester, but when I took it we wrote a fully functioning compiler in C++ for a non- trivial subset of C. In the end, we had over 5,000 lines of code in our src folder, so if you like programming, you'll get to do a lot of it. The project was designed for three people, but I only had a two person group, so that might be why I thought it was so much work. The course website was hands down the best I have ever seen. So much feedback, such graphs, much wow. There was incremental feedback on the project, but nothing was due or graded before the end of the course. Not so great if you're prone to procrastination, but also you can set your own pace. The lecture material and the project were not well integrated. Lecture material did not always seem relevant, but was a times interesting. The course covered a wide variety of topics in great detail, but the exam was less difficult than I expected (still not easy though). The general philosophy was to teach like the lecture was the only aspect of the course, but test with the knowledge that students had to spend a lot of time on the compiler.
Graphics: (With Prof. Phillip Slusallek): I had a love/hate relationship with this class. Assignments were extremely difficult, but extremely satisfying. Like I mentioned before, we implemented a ray-tracing renderer from the bottom up, which was awesome. If you want to take a look at what we did, here's a link to this year's ray-tracing competition.
Machine Learning: 9 credits, Vorlesung/Übung. I stopped attending this lecture after the first two weeks, but from what I heard, my first impressions were pretty accurate. UdS has a much heavier focus on theory than UW, so be prepared to teach yourself a lot of math on your own. There are no prerequisites for any courses, but I would recommend viewing Math 208, Math 309, Math 324, and statistics as minimum prereqs. Do not expect relevant concepts be reviewed. This course has required weekly homework, uses MATLAB, and is generally regarded by students as one of Saarland's most difficult classes. Prof. Hein does really interesting research in machine learning and is obviously brilliant, but seemed to be the sort of professor who doesn't realize that his students aren't understanding the material.
Non-CS Courses:
Despite what UW tells you about having to transfer all the classes you take, Saarland's international office says they will only send home the grades you want them to. Which doesn't mean a lot for CS courses, but it does mean you can take any of the free language courses or (if your German is good enough) some course in another department without getting a grade for it. The German courses fill up quickly, so try to sign up as early as possible. If you already know some German, I'd also (alternatively?) recommend signing up for a language partner. You'll get a lot more practice that way.
German 101: Easy, gut course, as expected.