YT YouTube2Text

YouTube2Text → Transcripts → [CS61C FA20] Weekly Lecture 08.LIVE - RISC-V Datapath

[CS61C FA20] Weekly Lecture 08.LIVE - RISC-V Datapath — Transcript

by CS 61C Departmental · 12,233 words · 1,960 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00and record yeah perfect
0:03wonderful all right and we'll keep you
0:06on the side
0:06we're gonna we'll spotlight borrow
0:08myself and keep you on the background
0:10until we pull you up that's okay that
0:11way you can just relax for a second
0:13okay yeah exactly here we go let me
0:16spotlight this guy
0:18stop video spotlight for everyone
0:21and spotlight for everyone add spotlight
0:25perfect
0:25all right good stuff and we have 36 all
0:30right welcome everybody
0:31it's great to see you folks welcome
0:34welcome
0:37boy busy time on the political space if
0:40you're uh i just got my ballot i'm very
0:42excited by the way everybody vote this
0:44is
0:44recorded i want you all to vote i want
0:46you all to exercise
0:48your your rights folks this is very
0:49important so make sure you do that
0:52we're going to remind you every week
0:54we're going to remind you every week
0:56and and be informed like read up on the
0:58ballots look and look on that
1:00right there's lots of i mean
1:01propositions i mean look at the
1:02propositions on the ballot
1:04make sure you understand that craig we
1:05have these propositions which are kind
1:07of
1:08you know uh any any citizen can put a
1:11proposition forward and it becomes law
1:13if
1:13that people vote for it so i grew up in
1:15california so yeah oh yeah all right you
1:17know
1:18yeah and um
1:21i i yeah prop 13. i got
1:24my my ballot last friday or so
1:28yeah definitely and i've actually spent
1:29i spent uh
1:31like three hours with my best friend who
1:33lives in california we
1:34just went through every single one and
1:36we marked out and we talked about it and
1:37we fought it was great really great
1:39do that do that kind of thing all right
1:40let's get started ladies and gentlemen
1:42welcome to
1:43week eight once a week lecture risk five
1:45data path control and pipelining
1:50great to see everybody here wonderful
1:53wonderful
1:54all right uh let's go let's get started
1:56so here's our agenda
1:58uh we're gonna do as we always welcome
2:00to week eight we'll
2:01do a little computer in the news we have
2:02a wonderful guest a colleague and friend
2:04craig zillis
2:05from university of illinois
2:06urbana-champaign is going to join us in
2:08a moment or two
2:09then we'll talk about our week's plan
2:11and any uh
2:12we have some information about the
2:14upcoming exam i'm sure you want to care
2:15about that but we'll
2:16let greg take our leave after we after
2:18we get it done with that
2:20yeah all right important thing we are
2:22past the midpoint we are past the
2:23midpoint
2:24of the class exactly it's all just
2:27downhill from now
2:28i love it i love it no it really is it
2:30is it's it's it's like my
2:32you know i'm over 50 boy i think you are
2:33too so it's like just like our lives
2:35just downhill
2:36just coasting put it in neutral let my
2:39tesla recharge
2:40perfect i just got it i got it there uh
2:43computing in the news
2:44all right computer news um waymo
2:47so wait you know that google split uh
2:50from google to
2:51all of its different versions a through
2:53z i think they called about it and waymo
2:55was one of them
2:56uh and for a while they were having
2:58these little teeny kind of a mini
3:00it's like a vw minivan but a much
3:02smaller version that had no steering
3:03wheel and they were doing test drives
3:05around mountain view
3:06they went a little bit bigger uh and
3:08they went to arizona where the
3:10roads are really wide and there are a
3:12lot of people running around like new
3:13york city or a big city
3:14a very slow and soft kind of a kind of a
3:17space they did they did all this test
3:18prep for this
3:19and in fact i just read on the article
3:21the articles linked here from bloomberg
3:23um when people were doing these beta
3:26testing drives they were
3:27writing in this way more cars that had
3:29no uses for a while they had you know a
3:30person there with their hands on the
3:31wheel just making sure
3:32then for a while they actually had
3:34nobody there this has been beta test
3:36mode just to see if it worked
3:37really amazing and people everybody
3:39writing in those cars doing beta
3:41testament had to sign an nda a
3:42non-disclosure agreement saying
3:44if the car ride is terrible and it stops
3:48every two seconds they won't complain
3:50they won't post on social media like
3:51they literally had to silence like a
3:52silence
3:53clause they had everybody do i mean they
3:55certainly asked them to give feedback
3:56for you know how did it go blah blah
3:57blah
3:58they they're ready they launched can you
4:01believe it
4:02the future is here folks i mean elon
4:05musk keeps telling me my tesla will be
4:06able to drive me from here to there
4:07without my hands on i mean hands on the
4:09wheels still but like
4:10on roads my doesn't even only does
4:11highway so this is it look they've got
4:14the system
4:14in in arizona i think there are i think
4:16three or 500 cars so it's so small so
4:18small scale
4:19but they've got cars that have no driver
4:21this is unbelievable this is like
4:23disneyland or fairyland so i
4:25i think i i would hope that people who
4:27are using these
4:28these cars will you know you know share
4:31some video high def
4:32uh you know iphone video of here's me
4:34getting a car there's no driver look
4:36ah we're driving around i cannot wait so
4:38exciting exciting i hope
4:39i wish everybody safety i certainly wish
4:41for this to succeed but
4:42i mean it's here future is here all
4:45thanks to
4:45lidar sensors crazy machine learning the
4:49same thing we talked about last week bor
4:50we talked about this last week or the
4:51week before
4:52these high-end these high-end machine
4:54learning systems they are built into
4:56this system that's how they work
4:57it really is the key to thanks to all
4:58the hardware advances that we've talked
5:00about
5:01and and and thanks to a whole bunch of
5:03our alumni who went there
5:04uh i mean the the chief architect of
5:07this is uh
5:08pierre pierre trois um who
5:13who got the masters from berkeley yeah
5:15um
5:16yep so and a bunch of other people who
5:20just went through through as our alumni
5:23after
5:24just taking classes uh some of them uh
5:27with the advanced degrees some of them
5:29need bachelor's degrees
5:30and just made all of that yeah exciting
5:34exciting space all right i'm trying to
5:35be here wonderful all right let's keep
5:36on going
5:38ladies and gentlemen let me introduce
5:40our guest of the day professor craig
5:42zillis
5:42of uiuc university of illinois
5:44urbana-champaign
5:46so craig uh let me now spotlight him
5:48give a second i'll spotlight him here
5:50and add spotlight and i'll leave it on
5:53spotlight ourselves just to focus on you
5:55if that's all right remove spotlight and
5:56give it to you
5:57so um yep remove spot here we go there
6:00we go so
6:01uh so craig and i met each other
6:04because i was freaking out craig has
6:06been part of the computer science
6:07education he's a craig is a regular
6:08professor
6:09he does research in architecture uh and
6:11most recently he's been working on
6:14how to make exams scale online what's
6:16the future of exams could
6:18just like in a way piazza is some sense
6:20the future of
6:21online forums and as is slack and other
6:24discord and other things
6:25just like grade scope and uh is the
6:28future of exams with computers
6:30i mean future kind of homework and and
6:32auto grading and and
6:34supporting that and scanning exams
6:36craig's really research is thinking
6:38about and maybe quite correct
6:39is thinking about could we build a
6:41system that will allow us to randomly
6:44generate questions
6:45to allow us to be really creative rather
6:47than having to write exams every year
6:50what if we put the time into actually
6:52writing
6:53question generators and packing them all
6:54together and craigs really cared about
6:56how would you then assess a campus so
6:58he's worked on something called the
6:59computer-based testing facility or cbtf
7:02and that's a space where students come
7:03to take these exams so just like you're
7:05taking our exams
7:06on by the way just for fyi the system
7:09that you are using for your quest in
7:11your
7:11um midterm and your final is a system
7:14that
7:14again thanks to it's called prairie
7:16learn and it's thanks to craig and his
7:18team
7:18for building that priority loan system
7:20and craig's cared a lot about building a
7:21system
7:22a place of actual physical location
7:24called cbtf
7:25on campus to allow people to go and drop
7:28in whenever they want to when they show
7:29their id and they take an exam and they
7:31can
7:31take a range of time you can have as
7:32much as you want to or you can have a
7:33limited time and
7:34there might be a list of things you can
7:35bring in all those things are part of
7:36this eptf to try to make this really
7:39rigorous and what's
7:39nice about craig is he's done research
7:41to show how effective that is
7:43where i've seen cheating it's really
7:44interesting stuff that craig has done
7:45and so craig i've known each other for
7:46many many years
7:47and it's great to have you here welcome
7:49craig thank you
7:51so tell us a little bit about how you
7:53got to where you are did you always know
7:54you wanted to be a professor
7:56did you where did you grow up where did
7:58you what bloop when did the kind of
8:00academic bug
8:00bite you um yeah so i grew up in the bay
8:04area south bay area just north of santa
8:06cruz
8:07um i did my undergrad actually
8:09surprisingly
8:10my first two degrees were in mechanical
8:12engineering i was the kind of kid that
8:14i didn't know that uh you know always
8:17was taking things apart and trying to
8:18put them back together and
8:19and yeah so i studied mechanical
8:23engineering at mit
8:24and then uh i did my master's degree in
8:27robotics which i
8:29think of as the gateway drug between
8:31mechanical engineering and computer
8:32science
8:34i then worked for a year and went to
8:36graduate school and studied
8:38um that basically i worked for a year
8:41and i found like
8:42oh i really want to do more of this
8:44computing stuff and i got interested in
8:46digital electronics pretty deeply
8:48and i realized that my background even
8:50though i had taken a number of
8:51of computing and electronics courses
8:54that i didn't have
8:55even you know a full degree and so i
8:58went back to graduate school in
8:59electrical and computer engineering
9:01and ultimately transferred to
9:05computer science for my phd but um
9:08that i got uh super interested in
9:12computer architecture there so i did my
9:13my graduate work at the university of
9:15wisconsin-madison which has a
9:17very strong group of computer architects
9:20and so i did my phd um in computer
9:23architecture and that whole time
9:24i was studying uh my intention was to go
9:28and build processors to work for the
9:30ibms and intels and
9:33back when there was digital equipment
9:35corporation and you know
9:36there was a lot more variety of computer
9:39companies back then right all on route
9:41route 128 on the western part of
9:44massachusetts that's right
9:45the massachusetts group the the west
9:47coast group in portland um
9:49hp was in colorado at the time um
9:52yeah and so i actually um
9:56you know i almost left with a masters
9:58and and took a job and
10:00i had a very interesting conversation
10:02with um
10:03one of my uh mentors who was
10:07um joel emmer who's who's now a fellow
10:10at
10:10intel and actually retired from intel
10:13and teaches at mit
10:14um and he sort of said you know uh
10:18you know don't don't necessarily be in a
10:20hurry to work
10:21in that you know when you graduate
10:24you'll have 30 or 40 years to work that
10:26if you if you want to get a phd
10:29take the time that there's sort of no
10:30experience quite like it and that was
10:33um you know work will be there for you
10:36you know
10:36four years later um you'll still have
10:39plenty of time to work so that was
10:41pretty compelling but even then
10:43most of the time i was i was expecting
10:45to go into industry when i graduated and
10:47so
10:48again there was a mentoring experience
10:51that that some
10:52faculty that were on my phd committee
10:54pulled me aside and said
10:55we think you'd really be a really good
10:57professor um
10:59and so you should think about that and
11:00so um
11:02i guess my recommendation would be to
11:03seek out uh
11:05people to give you advice um
11:08about that so so i've taken sort of a
11:10crazy path from mechanical engineering
11:13through computer architecture and now as
11:15as dan was saying you know a lot of my
11:18work relates to
11:19how do we teach particularly teaching at
11:22scale
11:24um so the the project that that dan
11:28mentioned is really
11:29um focused on so our campus the
11:32university of illinois
11:33is the flagship campus for the state of
11:35illinois just like berkeley is for
11:36california um and we are the second
11:39largest engineering school in
11:41in the country and so we have some of
11:43the same challenges that you guys have
11:45which is
11:46we're a really good institution but we
11:48also have a lot of students going
11:50through our classes and so how do we
11:52maximize the quality in the presence of
11:55the number of students and so
11:57our project the the computer-based
11:58testing facility was really about
12:00how do we how do we make sure
12:04assessment in the form of you know exams
12:06and quizzes and things like that
12:08um you know meets the the demand of
12:12you know our large classes without
12:14losing a lot of quality that
12:16that you know you guys know how much
12:18time you spend
12:19you know studying for exams but you
12:21might not realize that
12:23how much time your professors and their
12:24course staff spend
12:26preparing and running exams um that
12:29typically
12:30the model you know the standard model of
12:32maybe like two midterms and a final that
12:34that was pretty common with pencil and
12:35paper exams is because
12:37it's really painful for faculty and
12:41their course staff to
12:42to run exams and we thought you know
12:44that's really not
12:45good for the students that having these
12:46sort of big heavyweight
12:48exams you know relatively uncommonly was
12:52uh would be problematic and so uh we
12:54were thinking like how could we
12:56sort of um make this
12:59experience for our students and that's
13:02sort of what led to
13:03us thinking about the computer-based
13:04testing facility and so tell me a little
13:06bit you so tell me if you're not going
13:08to give two midterms in a final what do
13:10you do
13:10how do you do i mean and by the way you
13:12teach a course i understand cs233 very
13:14similar to 61c
13:16it's almost exactly the same course it
13:18was mips and we do risk five but we also
13:20did mips about
13:20four or five years ago so we actually
13:22almost taught the exactly the same
13:23course
13:24tell me what you do in your in your
13:25vision of assessment
13:27yeah so um so there's there's two main
13:31things that we do one is we do
13:32very frequent testing so basically every
13:35two weeks
13:36we run a 50 minute midterm um and this
13:39starts in like week three or four of the
13:41semester and so the first one
13:42you know it's it's you know the idea is
13:45that
13:45we want to frequently test you
13:49so that if you find that you're falling
13:52behind
13:52you get that feedback early that what we
13:55don't want to have happen is that your
13:56first exam is in week
13:58six or seven of the semester and by then
14:00if if you things aren't going well
14:02you know it it's insurmountable to catch
14:04up so we want to do
14:06frequent tests and give you you know
14:08constant feedback
14:09but also um then you know if we're if
14:12we're doing frequent testing then each
14:14each exam isn't you know it's only like
14:17five or ten percent of your grade it's
14:19it's not quite as scary and sort of
14:21the amount of studying you have to do is
14:23not as much
14:24and then the second thing that we do um
14:27is we do second chance testing which
14:30means that
14:31and and i like the sound of that i'm
14:32sure the students do too tell us more
14:34the students love this so with the idea
14:36of section second chance testing is
14:38if you go in and you take an exam as as
14:41dan mentioned we use prairielearn
14:43you get immediate feedback so you know
14:45when the exam is over
14:47um because we do auto grading on on
14:49almost all of the exam
14:51um you know how well you did on the exam
14:55and so what we do is the the following
14:57week we run
14:58an optional second chance exam where if
15:01you want you can
15:03study the things that you didn't do well
15:05come back
15:06that you know schedule an exam that
15:08following week and take it and we do
15:10some form of grade replacement
15:12um so do you do a full replacement or
15:14just some frac like
15:15do you have to own some part of the exam
15:17the first time or can you kind of
15:19get a zero the first time and ace the
15:21second one and somehow get you know an
15:22ace
15:23by the by the two of them yeah so it's a
15:25different this is actually one of the
15:26things that we do research on is is
15:28understanding these policies
15:30what we found is that and i'm sorry to
15:32tell you these students that this is
15:33this seems to be true uh is that if if
15:37if there's no penalty for
15:40there's no consequence for doing poorly
15:42on the first exam people
15:44will tend to blow it off um there's a
15:47yeah
15:48because we have we have we have a full
15:49replacement policy here in our class
15:51it's called the clobber policy and if
15:53you know you do better on a later exam
15:54but only on the same question like you
15:56know the later the midterm might have
15:57one question that was on the previous
15:59exam
15:59so if you ace that one it's a full we're
16:02doing a full replacement usually it was
16:03a
16:03curved replacement but this year we're
16:05going to do a full replacement so if you
16:06whatever your grade of is higher it's
16:08higher if it's lower it's okay it's just
16:09averaged in but if it's higher
16:10completely replaces the early one but
16:12you're saying that people don't study
16:13for the first one
16:13i'm saying that some people don't so we
16:16definitely
16:17see you know we've had instances where
16:20people just don't even take the first
16:21one if
16:22we use that kind of policy um so we like
16:25to have some skin in the game for the
16:27first one because i think the
16:28the what the research suggests is that
16:31some of the benefit of second chance
16:32testing comes from
16:34taking the you know studying for the
16:35first exam taking it
16:37getting the feedback of what it is that
16:39you still don't know
16:41taking that feedback and studying those
16:43things and coming back and demonstrating
16:45you know that you now know them that
16:47have it going through that cycle
16:48twice um you know rigorously that
16:52uh is important so so i actually started
16:54doing second chance testing when
16:56we were doing pencil and paper and we
16:57were grading everything manually
16:59huh um and that's a lot of work
17:02a lot of work oh my gosh at that time we
17:05wanted to
17:07you know there you know there was a
17:08consequence of a student if they just
17:10didn't study and they went in they took
17:11an exam
17:12uh you just be like hey you know if i
17:14can get by without studying great you
17:16know and i totally understand
17:17you student you know you know your
17:19desire to optimize your life because you
17:21know you have three or four other
17:22classes that are going on
17:24and they and they don't have that policy
17:25that's that's the danger right you're
17:26the one who's soft and squishy with your
17:28policy
17:29they don't so they really have to study
17:30for those classes and they come to you
17:32so you know i'm going to lean on you
17:33to not you know to borrow time from your
17:35class for their class because that isn't
17:37you know no no retakes in that sense
17:39right yeah
17:41that yeah because yeah you know
17:43everybody's an optimizer right and
17:44actually this is you know you know tying
17:46this back to computer architecture that
17:48you know this
17:49is you know that that's very much what
17:51computers you know the way we design
17:53you know computers is is all about sort
17:55of optimization problems of like okay
17:57how can i
17:57you know given a finite amount of
17:59resources how can i
18:01do the best with those available
18:03resources but so
18:04so i don't i don't fault any student
18:06from being optimizers but it just means
18:08that
18:09there's pressure on us as educators to
18:11design
18:12policies that you know uh
18:15get the behavior that's going to be most
18:17beneficial for the students and so what
18:19we found is having some
18:20amount of skin in the game of the first
18:23exam so
18:23you know an example policy that that
18:25people seem to like is
18:2710 from your lower score
18:30plus 90 percent of your higher score so
18:32it's it's almost
18:35complete um clubber yeah yeah
18:39complete clobber but not perfect you
18:40know so so that's what we found
18:43and and overall what's the reaction i'm
18:45sure the students love the retakes but
18:47overall do the students appreciate
18:48because
18:49to some things it's funny i have this
18:50theory which is there are some things
18:52that are better for students
18:54but it's like broccoli eat your broccoli
18:56and it's even though you're not going to
18:57like it you're going to tell me at the
18:58surveys you don't like this but it's
19:00better for you it's better for you to do
19:01the reading
19:02and force you to the reading before you
19:03come to lecture because you'll get more
19:04out of lecture
19:05so i don't have time but they don't do
19:07it so
19:08aside from the retakes i'm sure they
19:10love what do they think about
19:11the more rapid assessment versus the
19:13bigger high-stakes things do they
19:15do they appreciate having exam every two
19:17weeks is that overall
19:18because it sounds like you do they do
19:19better at the end that's like from the
19:21point of view of better learning that's
19:22that's better
19:23but what's the student perspective on
19:24that i i think
19:26uh they they in general like it
19:29up to a point so um like there's no
19:32break is like i gotta examine every two
19:34weeks i can't go break here
19:36right um and i think especially if you
19:39imagine that you know
19:40you're starting to get three or four of
19:41your classes that are testing
19:44frequently um it can be fatiguing
19:47um but i i think um so that's that that
19:51is a concern but in general
19:52students are very positive about smaller
19:55more frequent assessment
19:56especially if if they have flexibility
19:59um so
20:00so our system back before covid we have
20:02a physical space
20:04we have a couple computer labs on campus
20:06and we give students a three-day window
20:08to make a reservation and they come and
20:10they take their exam whenever it's
20:11convenient for them
20:13and so that's that's another feature
20:14that they really like is because
20:16you know if you know if their psychology
20:19class has an exam on thursday they can
20:21choose to take the
20:22exam in my class on saturday instead of
20:24you know the thursday that i would have
20:26run it on
20:27so they get to pick a time that's
20:28convenient for them and then
20:30another thing that they like is you know
20:33picking the time of day like some some
20:34kids work well
20:35at 9 00 p.m and they'd like to take
20:37their exam then other people
20:38you know want to get it done at 10 a.m
20:40in the morning and so having that
20:41flexibility
20:43is really uh beneficial what do you
20:46think about having untimed exams and we
20:48had a 24-hour window to take
20:50exam that was some number of hours for
20:52some people here we're now gonna have a
20:5348-hour 48-hour window for our midterm
20:56and we've played with it we played with
20:58having you know basically untimed 24
20:59hours exam and the second one we're
21:01actually trying
21:02we're going to talk about this i'm going
21:03to be limiting that time per question
21:05we'll draw it up with the question so
21:06what do you think about the trade-off
21:07between untimed versus timed exams in
21:09terms of that
21:11um so you know i strongly believe that
21:14you should have enough time for the exam
21:18so you know i think one of the the other
21:20things about
21:21you know like why we
21:25are interested in frequent testing and
21:26early testing you know
21:28uh relates also to sort of having enough
21:30time that one of the things that we want
21:32to be true is that we want your score
21:34and our courses to be
21:36dictated by how well you understand the
21:38material not necessarily
21:40by how good your test taking skills are
21:42so we don't want to design
21:44exams where you know oh the right way to
21:46take the exam is to skip the first
21:47three problems and do problem four then
21:49if you have time go back you know like
21:51right you know so you know definitely
21:53having enough time set aside for your
21:55exam i think is super important
21:56um there's definitely a point in you
21:59know for a bunch of students where
22:01uh they've done what they can but they
22:03still have 20 minutes left and so they
22:05stay there for 20 minutes
22:06and you know my one of my concerns would
22:09be
22:10you know uh is this a an extra burden on
22:13those students that feel like they
22:15should spend more time even when
22:16it's not being productive um i guess the
22:20other
22:20concern i have um you know one of the
22:23things that the literature suggests
22:25that is a mitigation for uh cheating and
22:28online
22:29exams is having time for them um so
22:32if that is a concern i know none of the
22:34students at berkeley have ever cheated
22:36in history so
22:37that's certainly true actually i was
22:39gonna bring that up tell me you've done
22:40so many
22:41research into cheating tell me what you
22:42found cbtf is a place where you can
22:45guarantee as much as you can that
22:47cheating wasn't happening because
22:48there's somebody watching this space as
22:49a staff member you're
22:50in a booth and a kiosk and it's just you
22:52and nothing else and whatever you were
22:53allowed to bring into that exam
22:54calculator or whatever
22:56um but now that you can't go to that
22:58place and you're kind of
22:59asking these students to take these
23:00exams at home what are you doing in
23:02terms of is it is it an honor code
23:04are you having them bring a camera are
23:06you recording the space
23:08or what are you recording the screen
23:09what is that how are you managing
23:10cheating because cheating is a reality
23:12even though we know
23:13no one you know has ever cheated before
23:14either what are they what are you doing
23:16at least to make sure it doesn't happen
23:17in the future
23:18yeah i think it's it's uh i think
23:20preventing cheating
23:23in online exams is basically impossible
23:27so actual prevention so we we go for a
23:30number of mitigation strategies
23:32um so we we've instituted so
23:36since we can't run the physical cbtf a
23:39group of people have organized what we
23:40call the cbtf online
23:42which is some zoom proctoring so
23:45we do have people that are
23:49you know one-to-many proctoring um
23:53so that's one thing we're not trying to
23:55to lock down the machines i think
23:57one of the big things that we do is as
23:59you mentioned earlier
24:00we do a lot of randomization on our
24:02exams and this sort of came from
24:05uh when we were started running exams
24:07asynchronously that
24:08you know the thought is is if you let
24:10different people take the same the exact
24:12same exam at different times then
24:14student a goes in and takes the exam
24:16comes out and talks to student
24:18b and and student b has an advantage
24:20over student c that doesn't have friend
24:22a
24:22um so if we if we randomize the exam so
24:25different students have different
24:26questions then
24:27um it means that it's you know very
24:30difficult
24:31you're more dependent on sort of what
24:33you know than what
24:34other people have told you and it also
24:37means that we can give out
24:38you know practice exams um that are very
24:41similar in structure to
24:42the actual exams sure um so
24:46a heavy degree of randomization is is i
24:48think one of our keys
24:50uh this semester we've been running
24:51synchronous exams
24:53um to try to you know
24:56mean that at least students in the class
24:58at the same time
24:59have to be focusing on their of their
25:01exams rather than potentially helping
25:03other students
25:04um good right yeah
25:08we don't know it's hard i mean we've got
25:09to support multiple time zones and so
25:11all of that with 1200 students has been
25:13just a challenge and so we're
25:14mostly leaning on the honor code and
25:16doing what we can in terms of
25:17randomization some other things we can
25:18to to try to
25:19to help with that tell me the things
25:20you're excited about what are the kind
25:21of where are you kind of cutting edge
25:23research
25:24the next five years this is what we
25:26always ask any any candidate for a
25:27faculty
25:28candidate what are you excited about
25:29next five years of terms of
25:31ai mixing with prairie learn mixing with
25:33cbtf but you know what what's the space
25:35what
25:35what what gets you up in the morning
25:37yeah so um
25:39i've been recently teaching one of our
25:41cs1 classes for
25:43for for berkeley people that um this is
25:45close to the cs0 that you guys teach
25:48the beauty and joy of computing and i
25:51have a couple things that we're working
25:52on there we've been focusing on a very
25:54uh reading before writing kind of
25:57curriculum
25:58and one of the key things there is the
26:01the research literature in learning the
26:03program suggests
26:04that being able to read and
26:08explain code in english is a is a step
26:12um is a building block step to actually
26:14being able to write code
26:15and so my group has developed a natural
26:18language processing
26:20system that allows us to present pieces
26:24of code to students and have them type a
26:25description in english and automatically
26:27score that
26:28wow wow it's kind of the pseudo code
26:31like how they're describing the
26:32pseudocode in english they're kind of
26:33saying it well it first
26:34grabs a number but are they doing like
26:36because when i've asked you is to do
26:37this they they go
26:38step by step they literally read well
26:40this this one for i
26:41like they're they're not able to step
26:43back and say no i don't care about the
26:44four right tell me
26:45that you're iterating over the list and
26:46doing something but they can't do that
26:48at the early stone
26:49so what what are you finding your
26:50students are doing are they line by line
26:52or they're able to pull back and and see
26:54right and well so so uh we've trained
26:56this auto grader to
26:58to not give credit to the low-level
27:00description so
27:01as you as you say our goal is to
27:04you know we want them to see the forest
27:07for the trees so it's like
27:09right this is you know this this
27:12function
27:12is is taking the in making a new list
27:15out of the
27:16um you know even numbers that are you
27:19know in the provided list
27:21right right right right how many regrade
27:23requests are there there's some it's not
27:25perfect yet uh
27:27it's still research um it's about 90
27:30accurate which is really good for
27:32homework
27:33we put it on exams but every time we put
27:35it on exams we have a human go through
27:37and
27:37make sure it doesn't make mistakes um
27:41but i think the key thing is you know
27:43and i think the whole point of
27:44automation
27:45you know the power of automation is that
27:48um
27:49you know anytime a student wants to
27:51study they can get immediate feedback
27:53on their work um and so
27:56to me it's almost a more important
27:58formative tool than
27:59you know actually putting it on exams
28:01definitely definitely
28:03um there's a question actually it's very
28:04good question from benjamin uh who's
28:06also asked good questions in the past
28:08he's saying for a heavy degree of
28:09randomization
28:11how can you ensure that each given exam
28:13is the quote-unquote
28:14same difficulty um since it seems the
28:17same question with different numbers and
28:18parameters would be easier to randomize
28:19but also make cheating easier
28:21whereas truly randomizing the questions
28:22could lead to differences in difficulty
28:24of the exam before between students
28:26what was a great question it's a very
28:28good question and it's something that we
28:30studied
28:31um so i think the the key thing is is
28:34we're not perfect we admit we're not
28:35perfect um
28:37the our goal is to
28:40be at least as good as semester over
28:43semester
28:44so so historical practices in fall
28:46semester dan would write an exam and
28:48spring semester dan would write a
28:49different exam
28:50right and right you know those exams
28:52wouldn't be perfectly fair so maybe if
28:54you took it in fall
28:56you know dan wrote a slightly easier
28:58exam and you know
28:59versus his spring so right and and even
29:02within a question like my sts question
29:04might be easier
29:05secret digital systems easier in the
29:06fall but the cache is as easy in the
29:08spring like you know
29:09that's even right some variants even
29:12and so basically we're translating that
29:15variance that used to be semester
29:17between semesters to within a semester
29:19um
29:20and so that that variance is there
29:24um we've studied it uh it's you know
29:27sometimes we make mistakes and
29:28and we we need to fix them that you know
29:31i've definitely had the case where
29:33you know a problem that had a a an
29:35average score of 37
29:37was in the same pool as a problem that
29:39had a seventy percent
29:40you know average or like wow something
29:42bad happened
29:43you know and so how do you correct it do
29:45you throw it out i mean that's the hard
29:46one
29:46do you bubble it up just double the what
29:49the temper the lower one
29:50yeah so one of the key things is is i
29:53tend to have
29:54many many questions on my exams um and
29:56that's another
29:57thing that you know makes me helps me
29:59sleep at night um is that
30:00you know there's a bit of central limit
30:02theorem going on is that if you ask
30:04enough questions then even if
30:06large law of large numbers you know that
30:08um
30:09yeah you know you're going to get some
30:11hard ones and some easy ones and
30:12somebody else is going to get different
30:13hard ones and easy ones and and
30:15it's going to work out yeah yeah so we
30:18do that
30:19um one of the things that i'm looking to
30:20do going forward
30:22is um you know the there's something
30:25called
30:25uh item response theory that indeed irt
30:28yeah i served on the college board for
30:30develop and we spent
30:31i learned that from them i was on the
30:32meeting i had no idea what it was and
30:34they said
30:34let me teach what i'm response theory is
30:36and i learned about why this is so
30:38cool most people don't know this you
30:39didn't know the same teacher's school
30:40you know
30:41so so all of you took some exam like the
30:43sat or act in order to get into college
30:46or most of you did
30:47um in general those are different exams
30:50but somehow they
30:51uncovered you know a score for you it's
30:53it's not a particularly transparent
30:56mechanism and so i don't think it's the
30:58main mechanism that we want to use when
31:00grading students because
31:01students value transparency um but i
31:04think
31:05as a background mechanism of like let's
31:07go ahead and run this in the background
31:09and if somebody
31:10got hurt by the main mechanism will
31:12boost their score up to what irb
31:15would have said they should have gotten
31:17so if you are that one person out of a
31:20thousand that gets really unlucky
31:22and got you know hard you know more than
31:25the average number of hard versions
31:27you know will bump you up at the end of
31:29the semester
31:31so those are some of the things that
31:32we're thinking about go ahead
31:34okay we spent quite a bit we we thought
31:36about this i mean what happens if you
31:38discover this in
31:39i mean in grade scope that we have very
31:41different
31:42um outcomes and scores so our approach
31:45was that we're going to go ahead and
31:47re-normalize somehow but we'll see you
31:50know we're not hoping that
31:51we're hoping that we are going to make
31:54our exams perfect
31:56for now right but we are preparing for
31:58for
32:00[Music]
32:0377 students here on the call we actually
32:06threw away
32:06we wrote i think seven borrowed seven
32:09different versions
32:11of uh the the the question the second
32:15programming question
32:16section three the splitting question um
32:18and then we threw away
32:19three of them because they said these
32:21aren't the same difficulty so we already
32:22have been thinking about it
32:23we already decided that only the bottom
32:25four or whatever some number were the
32:27same
32:27and so we and we had we made some more
32:29that were variants but the point is we
32:30threw three of them away because they
32:31said you know these are definitely
32:32and actually i want to give cc credit my
32:34our head ta cc
32:36uh came up and said you know dan i don't
32:37think one two and three the same as four
32:39four through seven so
32:40let's throw one through three away and
32:42not use them so i thank her for for
32:43thinking about that but that's that's an
32:44issue
32:45and and they're perfect practice
32:47questions for going forward i mean once
32:49when we use
32:50all of that in the exam they're good for
32:52for practicing i mean
32:53we don't have to be completely fair i'm
32:55getting i mean there is not the same
32:56scrutiny when
32:57when we give something uh give out as a
33:00practice question make sense
33:01and we definitely have pretty hard ones
33:03also
33:04not going live in any exams anytime soon
33:10yeah i love this natural light i just
33:12want to say i love the natural language
33:13to work that's really exciting i'd love
33:15to see it at 60 or any other conference
33:16i want to see
33:17some results from it that's one there's
33:18a there's a paper that got accepted to
33:1960 on it
33:21congratulations yeah it was a good year
33:23it's a good year
33:24any more questions absolutely over there
33:26i i have a quick i have a quick question
33:28uh i mean the
33:29the the a different different thing so
33:33i i'm assuming some of our students want
33:34to go to grad school um
33:36so how do they get do you have any tips
33:37of how could they get into
33:39uiuc we heard you at uc is good in
33:42architecture
33:44yeah so so we're we're good on a number
33:46of things um we have a very strong
33:48uh computer and education group now um
33:52uh you got colleen lewis got colleen
33:54lewis
33:55our own student yeah she's one of our
33:57all-stars
33:58uh yeah definitely um she's amazing um
34:02yeah like so i think the the key things
34:05um i would recommend
34:07um applying to grad school so um
34:10you know one of the biggest challenges
34:12of computer science is it's so big so
34:14trying to figure out what it is you're
34:16interested in
34:16um is is a challenge i think that's
34:18something that you guys as
34:20undergrads need to figure out sort of as
34:21you work to your elective courses
34:24um but get to know faculty in that area
34:27so um letters of recommendation are
34:30super important that you know what
34:32um what grad schools are looking for is
34:34somebody who understands what research
34:36is that can
34:37sort of say oh yeah this this person has
34:40the characteristics that
34:42means that they're likely to be able to
34:43do research and whether that
34:45the person says that because you've done
34:47research with them or because
34:49they've seen enough of you in other
34:50contexts that they can say that
34:52that's what you're looking for so i
34:53think you know making sure
34:55um that you are making some
34:57relationships with potential letter
34:59writers early
35:01and then the other thing that you know
35:03faculty know is um
35:04if you if you say i'm interested in this
35:06subject you know what are some
35:07interesting schools because um you know
35:10not every school is
35:11equally good in in every dimension you
35:14know not
35:15know again cs is so broad these days
35:18that not every department you know nails
35:21it
35:21in every sub-area and so you know
35:24finding the faculty
35:25in that area that you're interested to
35:27be able to tell you oh
35:28okay here's here's the list of schools
35:30that are are particularly good in my
35:32area that you should be targeting
35:34um because it's it's just like
35:35underground you're applying to undergrad
35:37there's just so many places and like how
35:39do you make sense of of all that
35:40information
35:42um the applying to grad school is is a
35:44little different um
35:46in that uh it's who you want to work
35:48with really
35:49it's yeah it's a lot of who you want to
35:51work with but i think one of the neat
35:53things is
35:53that if once you get admitted a lot of
35:56the schools will invite you out
35:58and you'll have an opportunity to visit
36:00and that's just sort of
36:01a super unique it's almost the time
36:03where you'll get more
36:04interaction with faculty than you know
36:07per unit time than
36:09almost any other time in graduate school
36:11so it's true so taking advantage of
36:13of that opportunity and you know when
36:15faculty that are trying to recruit you
36:17to come there
36:17you know ask them questions like what do
36:18you what do you think are the important
36:20problems
36:21in in this discipline you know what what
36:23should i as a as a new graduate student
36:25be
36:26thinking about definitely definitely
36:29wonderful
36:29all right thank you craig thank you so
36:32much for taking the time it's always a
36:33pleasure it's always great to see you
36:35and thank you also for what you've done
36:37for prairie learning all of us doing
36:38these online exams
36:39thanks to your work and your team's work
36:41outstanding let's give folks a hand
36:42thank you very much
36:45good stuff all right take care thanks
36:46again please thanks so much
36:48my pleasure wonderful thanks for the
36:50invitation all right yes indeed
36:51all right back to the main program back
36:53to the main program
36:54all right we're back to here and i'll
36:56remove craig's spotlight
36:58perfect okay all right we're back so
37:01bor you want to take us through this
37:02upcoming schedule it's a fun week
37:06yep um so what we got here i i need my
37:10uh glasses over here because it's more
37:13fun
37:13computers computer is misbehaving so it
37:17doesn't want to project on a big screen
37:18or i have to disable that and
37:20yeah if you've noticed on the
37:24recent videos i'm not scribbling because
37:26i can't and it
37:27you know it it's refusing to do that
37:29it's connected to another screen
37:31uh yeah it is totally annoying so
37:35um where are we now i released one
37:37lecture on pipelining
37:39i am hoping to get the other one out
37:42tonight and the other one later the the
37:44mondays lecture so the wednesday lecture
37:46is has been released i think on saturday
37:49the friday's lecture will go out today
37:51and in the next two days we're gonna
37:53have also the
37:55next monday's lecture and that
37:57essentially finishes all the material
37:59till
37:59you know for uh for the midterm so you
38:03have almost two weeks to
38:05to sit on that material and digest it
38:08and make sure that um
38:11you're comfortable with it uh we have
38:14one more
38:14lab that is pipelining that is preparing
38:17you for a second part of the project
38:18tree
38:19and that's the pipelining lab it does
38:23take a bit of a time
38:24and get ready for that and and use it to
38:26your benefit
38:28there is no lab next week
38:31just prepare you know use that time to
38:33get ready for the midterm
38:35and that's all yep and there is homework
38:38by the way the homework that we have out
38:39now
38:40is gonna take we shortened it a bit we
38:42took out one question is left as an
38:44optional that was like a 20 minute half
38:46an hour question
38:47but the homework is still gonna take two
38:49to two and a half hours the budget for
38:51that next
38:52between this week and next week the
38:54questions there are good the questions
38:56are
38:56uh totally make sense um
39:00from the uh
39:03as a preparation for the midterm yeah i
39:05was just saying i can't have anything
39:06any better way to study for the midterms
39:07they do the homework
39:08we always suggest that great great stuff
39:11all right
39:12so we've got some announcements uh let
39:15me click on that
39:16all right so quest regrades finally done
39:19holy moly i have never seen a regrade
39:23that was as painful as this one we had
39:24virtually one in four students
39:26needed to talk to us with a lot of it
39:28wasn't just like i want points on this
39:30but it was like
39:31there was some technical or some this or
39:33that and all those things sometimes
39:34actually
39:34require us to rerun we would make a
39:36change and then rerun it on everybody's
39:38and so all those things
39:39anyway it was painful but thanks to a
39:41great set of tas who helped coordinate
39:43all that to make it work
39:45so about a one out of four four students
39:46which i think is the highest fraction of
39:48regrades we've ever had and so that's
39:49why it took us so long so
39:50thank you for your patience with us
39:53we've got a final date we've got a quest
39:54retake this sunday
39:5624-hour period it's going to be exactly
39:58the same we can't make it
40:00you know as much as we want to give you
40:01test cases we can't
40:03it has to be exactly the same to be fair
40:05to anybody who just had trouble with
40:07technically with it
40:08and needs to redo it so that's the idea
40:10um i was just looking here i've got my
40:12there's a whole thread on slack where my
40:13staff is right now working through all
40:15the students who had challenges and who
40:17what the scores were we're gonna
40:18we won't make everybody do it but we're
40:19gonna try to figure out the final list
40:21by tuesday so you let you know by
40:23tuesday just don't
40:24send us mail and by tuesday we'll send
40:26it out we'll send a big piazza box
40:27saying here are people who need to take
40:28it again
40:29and then you'll just do this but it's
40:30essentially the same you're already past
40:31the material you are it's all hopefully
40:33the sea is already deep within you
40:35you know all the number representation
40:37and all that by the way is great
40:38preparation for the midterm so
40:39it doesn't you know does that hopefully
40:41you don't have to do any extra studying
40:42you're just basically studying for the
40:43midterm and kind of refreshing yourself
40:44that material is also fair game on the
40:46midterm so you're studying for this in
40:48general with this stuff but we'll just
40:49guys let you guys know
40:50and that between tuesday and sunday
40:51hopefully there's enough days for people
40:52to get some time to
40:53do whatever prep and also by the way
40:54it's open notes an open book i mean
40:57by the way i'm gloria i would tell you
40:58the story i i my first semester teaching
41:01at berkeley fall 2000.
41:03i said i'm going to shake it up here i'm
41:05gonna have an open book exam
41:07untimed like literally untimed i started
41:10at 7 p.m
41:11in wheeler okay and
41:15this is ready for this is the crazy
41:17thing so i look around about 7 p.m
41:20by 11 p.m there are people still taking
41:22it four hours in
41:23i look for a one-hour exam okay but i
41:25gave them infinite time
41:26and it's 11 p.m and the the janitor
41:30comes in and says
41:31i gotta kick you out to clean it so i
41:33took the remaining this is a midnight
41:35almost 11 students or 20 students a day
41:38and we walked i collected their exams
41:39and they walked silently single filed
41:41back to soda i found an empty room
41:43and they took the exam for two more
41:45hours now that's
41:47funny story number one first number two
41:48is after i headed the exam out i saw
41:50somebody in the front
41:51and they took the book out and they went
41:54page one and they started reading the
41:56book like they didn't
41:56actually they just figured they come out
41:59i got infinite time you said infinite
42:00time i'm going to read the book from
42:01page one
42:02page welcome to computer science
42:09don't do that don't do that that's my
42:10advice don't do that all right
42:12so we also have information about the
42:15midterm
42:17whoop all right the midterm is scheduled
42:19thank you everybody for voting what
42:22what days work for you um sunday and
42:25monday are the days people say we need a
42:27day
42:27by the way this exam we think the exam
42:29should be doable in about two
42:31ish hours that's our current thing and
42:33in fact we're so confident about it
42:35that we are 14 it's two
42:39we are so confident about it we are
42:40going to have every ta who beta tested
42:43and if the t
42:44we think i want the scale factor to be
42:46four but like maybe you'll argue that
42:48scale back to be two
42:49but if i tell you two hours then every
42:50ta should finish the entire exam in an
42:52hour
42:53and i don't think that was true for the
42:54last quest that they could do it in half
42:55an hour so
42:56i wanted to be at least two if i could
42:57have finished it even earlier than that
42:59so
42:59we're gonna say whatever we have for the
43:01tas we're gonna end up multiplying that
43:02by some
43:03factor to give you more time here's the
43:05idea
43:06you've got five questions many people
43:09said
43:09don't give us infinite time we will take
43:1148 hours
43:12limit it in some way but we also know
43:14you talk yourself in the conversation
43:16from craig
43:17it's appropriate to give people enough
43:18time to show what they learn i don't
43:20have people rushing through
43:21and not be able to finish that's not
43:22appropriate either so we're thinking
43:24what's the middle ground roughly we're
43:25going to write it so that's about 20
43:27minutes a question
43:28so it's about a three time scale factor
43:30for five different questions so each
43:32question can be done in 20 minutes if
43:33not if our tas can't do it in 10
43:35at least a factor of two if our tias
43:37can't do it in 10 then i'm not going to
43:39have
43:40an hour i will give you more time for
43:41that particular question so there's five
43:43different sections i should say five
43:45different sections
43:46i should have edited this let's say
43:47sections just like before we had three
43:49sections
43:50there are now five sections each section
43:52will have a time limit
43:53and we're going to basically have all
43:54the tas take it figure out what they can
43:56take it and then
43:57triple that time or something and that's
43:58how much you can take it so we'll make
43:59sure that you have enough time
44:01so the time isn't an issue and then
44:03that'll be the other so we'll kind of
44:04make both sides of the of the room happy
44:06the side that says don't give us 48
44:08hours we
44:08will use it all up okay we'll limit it
44:10but the other side says i need i've got
44:12slow test taker which i
44:13am myself give you more time i want to
44:15make sure that you guys can finish it
44:16and there so that's the idea okay
44:19so we should not be a five hour exam but
44:21we're going to give you one and you know
44:22we'll as i said
44:23we'll make that more per question if
44:24that particular question is too hard
44:26you take it any time in a 48-hour period
44:29you can start one question the first
44:30hour and then take a 12-hour break and
44:32take the next question
44:33although whatever you need to do same as
44:35in terms of style
44:36and the topics are these five topics
44:38it's going to be a quest redo
44:40basically summarizing all that you saw
44:41on the quest everything after kind of a
44:43week you know three and a half or so
44:44when the exam
44:45ended freeze that that's one question
44:48there's gonna be a float question
44:49there's a question
44:50is there is there is there c yes in the
44:53quest
44:53is there c's yes there is c so get ready
44:55to do that
44:56make sure you're prepared for this there
44:57is c there's a risk five assembly
45:00question which we're going to be able to
45:01do some coding and we're going to have a
45:02practice midterm so you'll see what kind
45:05of thing it is but you're going to be in
45:06venus we're going to ask you to do
45:07something and you're coded in in risk
45:09five
45:09um that's that's the plan there there's
45:11also going to be a data path control
45:12pipeline that's this material
45:14and there's sds logic which was last
45:16week's material okay
45:18that's what i got i'm with five minutes
45:20left and that's the that's what i have
45:21for the plan but i hopefully that
45:23addresses
45:23a lot of the concerns that the students
45:25had in terms of um
45:27what's on the midterm what's the scope
45:28what it is there's no more it should be
45:29any more surprises okay
45:31and by the way we mentioned last week
45:33that we will be giving you test cases
45:35for your c
45:36and test cases for your risk five so
45:38we're going to say do this thing and by
45:40the way here's some test cases and i
45:41told you
45:42last week we're going to reserve a
45:43couple test cases so you don't just hard
45:45code those those things in there okay so
45:47we're trying to make it so that the
45:48timing isn't an issue but we got to make
45:50sure you guys can get above the bar in
45:51code risk five
45:52you're above the bar and code some c and
45:54again we hope
45:55we want borah and i and the entire staff
45:58want you to totally crush
46:00the quest one the quest is gonna be a
46:02full redo
46:03plus a curve thing so whichever one's
46:05better for you will do that but we want
46:06you to crush that quest we do
46:08and it's all in one question there's
46:10gonna be a lot of partial there
46:11but again folks it's got to work you
46:13can't give a c that doesn't work if you
46:15give a c that doesn't even compile
46:17that's when that's that's not going to
46:18give us we're not going to fix your code
46:19for you that's never we've never done
46:20that for projects
46:21we don't do that for exams it's not
46:22different than we used to ask a quest
46:24the quest used to be write this line of
46:25code and
46:26mostly they give you it's mostly right
46:27no you have the compiler you have all
46:29the time
46:30get it working okay so hopefully you'll
46:32have enough time for the c again we're
46:33gonna take all the tis running through
46:34that question
46:35and whatever they take we'll triple that
46:37or whatever and that'll be the time you
46:38get so we'll give you
46:39enough time to get that one and make
46:40sure we get away keep these questions
46:42going
46:42keep these questions going i mean we got
46:44the two questions is the grading
46:47testing the same um i
46:52not sure exactly what that means but we
46:53will provide the full test coverage
46:56right right if that is what this meant
46:58yeah right yeah we're going to still do
46:59the same thing
47:00we're not going to grade it live we'll
47:01pull it out of it drop it in grade scope
47:02write some autograders to try to grade
47:04it so very similar way to do this
47:06there's a question will the practice
47:07exam reflective of the actual exam in
47:09terms of length and difficulty the
47:10answer is
47:11no we're not we don't have time to write
47:13a second exam and give that away
47:15so we're just going to give you like
47:16here's the kind of question we write
47:18risk five might be much easier
47:19so it's not at all i'm telling you no no
47:21the practice exam will not be an
47:23assessment of that
47:24it'll be you're going to write something
47:25but we're going to make sure the
47:26practice when it gets you practicing
47:27with you know with venus and doing
47:29something so ask you to do something
47:30it may be harder maybe easier but it's
47:32not going to be the same we're not going
47:33to be able to write
47:33the full i can't write all the float
47:35questions i can't write all the things
47:36all the things you saw
47:37it's just too hard look at our old exams
47:38for that kind of thing we're just going
47:40to give you something to give us a basic
47:41a rough test to that
47:43but the the point here is it will test
47:45exactly
47:46the the mechan mechanics of uh
47:52mechanics exactly right yeah yeah and by
47:55the way
47:56if you had any trouble technically with
47:59your setup i'm letting you know you have
48:01two weeks to get that debugged okay
48:04whatever you need to do
48:05practice with the zoom plug your i can't
48:07tell you people didn't plug their phone
48:09in
48:09what are you kidding me plug your phone
48:11in whatever you do get an extension cord
48:13plug your phone and get it set
48:14so don't whatever you had trouble
48:16whatever their cause was for the redo
48:18you got to get that working we i don't
48:19know if we have time to we had to write
48:21a sec we have to write a quest
48:22and then a quest retake and like we
48:24don't have time to write a midterm
48:25retake we may not be able to do that
48:27so you got to get this working we
48:28haven't yet figured out what to do if
48:30people have technical problems
48:31but we we don't you have to get the
48:33debugged you know we had enough time
48:35quest crest retake midterm you've got
48:37enough time to practice practice
48:38practice on that
48:39to get that stuff working so get your av
48:42whatever you're doing whether you're
48:43zooming or whether you're time lapsing
48:44or you're moving whatever you're doing
48:45get it debugged
48:46make sure it works make sure to work for
48:48whatever the maximum time is that's by
48:49the way another advantage of a timeline
48:51per question
48:52is you know i only have room on my phone
48:53for a time lapse that holds an hour and
48:55a half well
48:56if no questions more than a half that
48:57you're fine that's the kind of thing
48:59so make sure you get that stuff done
49:01exactly this was one of the reasons we
49:03also
49:03why we limited the time um there's a
49:06question
49:07if we get 100 on the quest redo coding
49:11question
49:11does our quest score become 100
49:14yes it does 100 first time ever
49:18by the way first time ever we've never
49:19done that before we've always done a
49:20curved thing where
49:22we compare you to your neighbors and if
49:23you moved up relative to your neighbors
49:25it was a curvy way of doing it
49:26not this semester that quest was so low
49:29it's a full clobber 100
49:31map so we want you to ace that quest
49:33question just rock on that
49:34beautiful yeah and then you know that is
49:37due to us and due to the semester
49:39being uh different than any other
49:42semester there they've been
49:43yep love it so the second question will
49:46the quest redo c have test cases or just
49:48the post quest
49:50c the quest redo c the quest redo does
49:52not the quest we do has to be exactly as
49:54before
49:55so unfortunately no test cases for the
49:56quest redo it has to be the same if
49:58people
49:58can't get free test question if you get
50:00a redo but you didn't get it when you
50:01took the real test
50:02that has to be the same but the
50:05quest read how to say this the quest
50:07redo c on the midterm
50:10so the question we do in the midterm
50:11will have test cases if that makes sense
50:13so not the one that's happening this
50:14weekend that's the retake for
50:16for the video problems that will not
50:18have test cases
50:20but on the midterm the quest redo coding
50:22question will have
50:24and the risk five we'll have so all the
50:25coding we're going to give you test
50:27cases
50:27and we're also going as i said keep a
50:29couple ourselves so you don't hard-code
50:30any answers there
50:31got that all right beautiful answer live
50:39uh i um know recording policies will be
50:42the same on the quest
50:43basically the exactly the same thing
50:44you'll get the same options
50:46and by the way i i really encourage you
50:48have two weeks i encourage you to try
50:50every option first try it time lapse
50:52then try
50:53the video then try it the the
50:56zoom proctoring whatever that thing
50:58works make sure it works
50:59beta test it triple beta test it make
51:02sure that works but yes the same
51:04rough rough policy as before is each
51:07section equally weighted
51:08the answer is no we have to think about
51:10that and think about that placeholder
51:11usually
51:12what i'm saying is like float is one
51:14lecture but it's one question
51:15that can't be the same weighting as the
51:17quest we do which was
51:18like four weeks and it can't be the same
51:20as risk five which is probably
51:22four or five or six lectures if you had
51:23them all seven lectures six seven
51:25elections
51:25seven lectures right i mean that's great
51:27so we we can't right because it's above
51:29it's there it's the machine so we
51:30certainly are not going to have equal
51:32weighting but we have to think about the
51:33balance of all that and probably
51:35if you look at our old midterms it's
51:36probably the similar waiting that we did
51:38before our
51:38float questions were smaller you know
51:40and other questions were bigger so think
51:41about that
51:42but it's a very good question thank you
51:43for that um
51:47this section will be labeled very
51:48clearly the question is will the
51:49sections be labeled so we can review the
51:50material before taking that portion exam
51:52um yes i mean i'm telling you here look
51:54folks ready the topics are there
51:57the topics are right there see the
51:59bullet way where's my hand where's my
52:00hair
52:01right here there's the topics okay
52:05there's five questions five topics we're
52:07telling you what's on the exam it's
52:08basically the whole
52:09all the material up to now so so it's
52:11beautiful
52:12so one thing to keep in mind i mean
52:14there are things that leak over
52:15uh uh a bit like uh you know you
52:19you in digital systems you need to know
52:21digital systems in order to build a data
52:23path
52:24right but you know it's like it's
52:25nothing so it's not so isolated exactly
52:28yeah yeah
52:28but we yeah but we try to we tried as
52:30much as we can put them there but
52:31sometimes they leak over here
52:32is exactly right beautiful all right
52:36um question regrades i don't know if the
52:39question grades have been released
52:41the equestria grades are done i just
52:43told my ta that they just finished them
52:44but i don't know if they've actually
52:45pushed them back to release them back
52:46but uh
52:47but the questions from our point of view
52:48are done they probably have some a
52:49little bit of
52:50a job of churning it adjusting it and
52:52then pushing it back out but thank you
52:53for that question
52:54good good good good uh and by the way uh
52:56related to that if we
52:58mess up any of the the the if there is a
53:00surprise in your
53:01in your grade in the overall grade do
53:04let us know
53:05because there are some things that have
53:06been um
53:08regraded manually and we have had issues
53:11before
53:11with that so don't get freaked out if
53:15if some grade gets lower
53:18perfect all right so that one's all done
53:20um
53:22i don't know what the first way so it'll
53:23be hacker sprank
53:25style where we can run test cases and
53:27see how many non-hidden ones pass
53:29yeah the idea is we'll give you a you
53:30know we'll give you something that lets
53:32you check
53:33all some some basic maybe just maybe one
53:35test but it'll be a test at least it
53:36will give you that test to make sure
53:37it's there set it up
53:38you write your code and then run it at
53:39least through that one i would by the
53:41way i would run it on other ones because
53:42that might not test every case obviously
53:44but it's going to test enough because
53:46it's more of a sanity check there but we
53:47have some test cases and again we're
53:49going to have the real one that has a
53:50big problem
53:51does it do the big one that you can't
53:52see but but it'll be enough to give you
53:54a very good sanity check sense of that
53:55so that's a good question there
53:56we got that one all right keep on
53:59keeping coming this is great
54:00um uh
54:06for the question what if we only know
54:10we did wrong wrong but how to fix it fix
54:14it
54:14could we get auto grading test after
54:16quest redo i don't know what that's
54:17asking
54:19it's asking us to release the authority
54:21of this test the test cases yeah yeah um
54:24we may release that let's think about it
54:26we may release it
54:28after we do the quest redo um because it
54:30might be a similar thing so
54:32we we we'll talk about that it's a good
54:33question let's let's board let's try to
54:35remember to bring that up to see if we
54:36would release the test cases for that
54:38thank you for that question
54:39and um on the quest the sections are
54:43level one two blah blah blah on the exam
54:44section should be labeled c for example
54:46because we first get going to start
54:48um i think i i mean i don't mind i don't
54:51know borah if you have opinions i don't
54:52mind labeling them sections one through
54:53five even calling it
54:55quest reader question float question
54:57risk five assembly questions so you'll
54:58know what you're getting
54:59i don't need i don't think we need to
55:00surprise people so i i i'm okay
55:04we can we we usually label them um one
55:06thing though right i mean there is this
55:08leakage so you know don't don't expect
55:10oh we can you know you
55:11you can't ask us any digital systems in
55:13data pack
55:14it is a digital system so so
55:18there will be a little bit of that but
55:20yes the main
55:21theme is going to be labeled yeah i i
55:24don't have trouble with labeling the
55:25main themes yeah that'll be easier for
55:26that
55:27there's a question and again yeah okay a
55:30handful of students actually didn't
55:32realize that there were three sections
55:33so we're gonna make it clear there are
55:34five sections
55:35each label make sure you finish all five
55:36sections you're not done with the ques
55:37with the midterm until you finish all
55:39five sections thank you
55:40yeah and then we need to to wrap up in a
55:42minute or so because
55:43there is another class i think after us
55:45there's a question do we have any
55:47extra credits for this class and we do
55:51um it's called the epa um then do you
55:55want to
55:55explain sure we mentioned in the first
55:57lecture epa is effort participation
55:59altruism
56:00there are ways for you to to show
56:02excellence and show that you're helping
56:03other students in the altruism case
56:05you're participating you're coming to
56:06these sections you're coming to these
56:07things
56:08um you're coming to office hours people
56:09getting to know you uh
56:11and and you um e is for effort so
56:15effort means you do every assignment you
56:16don't blow any labs off if you show that
56:18you really put your time into this class
56:19there are ways to get that
56:20um as well as if you take the
56:22performance if you try that there's a
56:24chance for the competitions if you
56:26submit a competition it means you're
56:27putting
56:28extra effort in so we'll see that and
56:29we'll give you some extra bonus so
56:30there's some extra bonus there
56:31um with that we'll play with that as
56:33well um
56:35exact timings for each section will be
56:37visible before starting a section yes i
56:39don't see a problem with labeling that
56:40so they'll know how much time to take
56:41for that because you know
56:42they set your timer blah blah blah boom
56:44got that there good question benjamin i
56:45love that that's a good question
56:47um discord might be counted epa if we
56:50can if there's a way for us to kind of
56:51get some numbers about it there's some
56:53way for our teas at the end by the way
56:54it's an informal
56:56internal number and if our tas can get
56:58to know you through discord then
56:59certainly discord is a useful as a
57:00mechanism to get to know you just kind
57:02of like office hours very much so i mean
57:04we're going to ask our
57:05tas to tell us what what they've uh um
57:08who's what they've learned about travel
57:10to the top exactly um
57:12he's visible and uh
57:15we'll i'll i'll save that one right riot
57:17for for later that's a little secret
57:18thing all right
57:19um good great thank you thank you so
57:22much folks
57:23thank you well it's great to see you all
57:24we'll see you next time there was a lot
57:26of questions
57:26this this is why we have these live
57:28videos exactly exactly
57:30we have that connection with other
57:31connections yeah all right thank you
57:34and we have exciting you know more
57:36exciting uh guests are coming
57:38so get your get your people from
57:40industry coming
57:42and there will be people from academia
57:44and by the way people have done very
57:46well in the project so everyone's done i
57:47mean
57:48i know the quest was a challenge we're
57:49hoping to replace with the midterm but i
57:51think folks are doing great in this
57:52class in terms of the scores i'm seeing
57:54on the project board have you seen those
57:54numbers in this room
57:55the median is a hundred for every single
57:57section
57:58for every single one project 1a project
58:001b project 2a median's 100. keep it up
58:03we want that folks we don't need to
58:04you know we want to give you all 8
58:06pluses it's not about making a you know
58:08weeding you out come on let's you gotta
58:10we'll put a bar if you all get above the
58:11bar we're happy we all did our job so
58:13keep it up keep on keep keep helping
58:15each other right that's why we have you
58:16work together
58:17help each other teach each other
58:18learning community we love it all right
58:20we'll see you next week folks take care
58:2161c thank you
58:24see you folks take care bye

About this transcript

This page contains the full transcript of [CS61C FA20] Weekly Lecture 08.LIVE - RISC-V Datapath by CS 61C Departmental, generated from the public captions YouTube serves with the video. The transcript has 12,233 words across 1,960 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.