Tác giả giữ toàn bộ bản quyền (All Rights Reserved) — không phải giấy phép mở. Bạn vẫn xem và học miễn phí trên OpenER; muốn tái sử dụng cần xin phép tác giả.
DAN GARCIA
Should not forget that. And I'm going to then also grab — I forgot to change my name — but I'm gonna do this. I'm going to spotlight you, Bora.
And I'm going to spotlight myself. And then the rest of the folks won't be spotlit on that and add that spotlight. That's just two of us.
And then we'll spotlight our speakers — our guests — when we share that. Welcome, students. Welcome, everybody. I see the numbers are rolling in here.
Perfect. And let me figure out how to change my name.
BORA NIKOLIC
Yeah. I'm going creator.
DAN GARCIA
All right.
All right. I think we're ready to start. Ladies and gentlemen, welcome. Welcome, everybody to — I'm getting a — let me — a little bit of — there.
Okay. Welcome, everybody to week — I think we're week 9 of CS61C live lecture! Woo! Good to see everybody.
Intro to caches. We're doing caches this week. A very exciting time. Lots of things are happening 61C.
We want to make sure to share that and just jump right in. We don't want to delay. So let me go to the next slide. Here's the agenda, what we're going to do.
We're past the halfway point. We're downhill from here. So that's just wonderful. It's nice to feel like Thanksgiving is not that many weeks away.
Looking forward to that. We're gonna talk about computing in the news. There's some exciting — I'm going to show a video for that. We have two wonderful guests from Intel here.
We'll update our schedule, as always. We'll do a sanity check to actually share with you how our students are doing. We have some good news for you all.
So stay for that. We hope that you're all — share this information with all the rest of the other 1,100 people who aren't part of this class. But we're recording this.
So we'll share with that. We have a very important relief package that we want to share. We were able to get this through Congress.
So we have the official 61C relief package we want to talk about. And make sure you stay for that 10-point relief package.
We want to talk about the Quest retake and update on the midterm, which is coming up this coming weekend. And we'll have an Ask Us Anything if we have time at the end.
Okay. And that's always what we try to do. All right. So I have a video I want to show you.
This is on neuromorphic computing. This is one of the Intel's new videos. And I'm gonna see if I can share this with sound if I get this done right.
Here we go. So Google Chrome with sound. And give me the thumbs-up, Bora, if this is working for you. Ready?
VIDEO NARRATOR: Organic. Intuitive. Fluid. Neuromorphic computing is designed to work less like today's microprocessors and more like the human mind.
Because brains are so different than transistor circuits, we have to rethink everything we know about computer architecture. Traditional microprocessors are synchronous, meaning every function is governed by a clock.
The clock says when instructions are executed and when functions happen. Everything works in unison according to the clock. Our brains are different. They're asynchronous, meaning no internal clock.
Neurons operate with no prescribed order. We just have thoughts, thoughts that build on each other or split off into different ideas. That's the way neuromorphic computing works, too.
It's completely asynchronous. So functions and instructions happen when they happen in parallel, which means neuromorphic systems think more like brains do.
They can adapt, recognize nuances, and teach themselves as they work. It's an elegant, organic, and beautiful way to think.
BORA NIKOLIC
All right.
DAN GARCIA
All right. Be sure to like and subscribe. So let me get out of that now.
BORA NIKOLIC
All right.
DAN GARCIA
And get back to this. And let you talk about that, Bora. Take us to some of that work.
BORA NIKOLIC
Yeah.
So one thing that we wanted to — that we wanted to talk about today is, in the past few lectures, we have talked about traditional microprocessor that is operating under the principles of a synchronous data system.
But there is a lot of stuff that is happening now that is labeled as neuromorphic computing. And this is a little bit different in how we do things.
Neuromorphic doesn't necessarily mean that everything is asynchronous. But people have been building asynchronous chips that they look a little bit more like human brains.
Process [INAUDIBLE] that are in these chips have perhaps all the complexity of a 61C project. So if you jump into this, you might be able to build one of these.
So info published a couple of years ago, this 128-core neuromorphic chip in 14-nanometer CMOS, each of these cores is not much more complicated than — when we look at the logic, not much more complicated than 61c project.
I mean, at first glance, there is a lot of stuff that happens behind it. There's more memory and things like those.
But then there has been a lot of stuff now when they're trying to put neuromorphic computing system out of that. So they have all these systems. The boards have Hawaiian means.
So there is Loihi, Nahuku, and Pohoki. Somebody has inspired the Hawaiian names here. But in the end, this is a pretty big machine that they put in this building.
So we have our great privilege today to have two guests from Intel. Both of them probably didn't have exactly the most linear paths to their careers.
And we do recognize that 2/3 of this class that not necessarily be CS majors. Many of you are not even LNS CS majors. Come from all kinds of different backgrounds.
So I am going to ask Farhana — Dr. Farhana Sheikh and Dr. Sriniva to introduce themselves to tell us how they ended up doing what they do now.
Farhana, I don't think is working on neuromorphic now. But she can tell us how she ended up at the [INAUDIBLE] for research.
And then we can chat a little bit about neuromorphic and other interests. So Farhana, tell us.
FARHANA SHEIKH
Okay. So I did my bachelor's in system and computer engineering a long time ago. So this was a kind of a broad program with a small emphasis on circuits, but not too much.
It was mostly at the system level and the software level and an architecture level. And then I ended up coming to Berkeley for a little bit, did my master's in with Edward Lee.
And we kind of talked about more on the software side of things. And I worked on Ptolemy. And then I went back to work. I worked in a CAD company.
And I worked there for five years. I became a senior manager. I was actually managing managers. And then I realized that I wanted to come back to school.
So I came back to Berkeley again. And I started off in the CAD field. And then I joined Bora's team. And I took some undergrad classes in circuits.
And I also took some CS classes, as well, in the undergrad. And I actually changed my field to circuits. So that's how I ended up in circuits. And then I went to Intel.
BORA NIKOLIC
Yeah. One day I may be known as Farhana's former advisor. That was my biggest accomplishment.
And Narayan, you also had a — probably not the most direct path to being an architect [INAUDIBLE].. You're muted.
NARAYAN SRINIVASA
My path is actually even more strange than you'd normally imagine. So I started as a mechanical engineer. My undergrad in India and then subsequently my PhD from here.
And mostly focused on robotics and adaptive control as the areas.
But as you know, these areas have emerged tremendously to the point where — and of course, at the heart of all this is the computer — and so being a curious person, I've always wanted to learn just more than just the robot and the associated tasks, but what drives them, what is the underlying hardware that
is implemented that allows us to actually do what we do. And so I started learning a lot more about it through courses first during my PhD endeavor.
And then I actually had some hands-on experience at Urbana-Champaign when I did a postdoctoral fellowship there working on visual processing and in a signal processing kind of areas.
We used to call it human computer intelligent interaction in those days. There was a big army-funded research lab there where we started working on some of those problems.
And it became apparent that more and more the issues are more challenging and from the perspective of energy efficiency, that is, how do you make these hardware sort of do what humans do from the perspective of processing information, that is, the efficiency with which they process information.
And so that stint in UIUC then led me to Hughes Research Labs, which is in Malibu, California. You may not know about it.
But it's a corporate lab funded by Boeing and General Motors as owners. But they do a lot of research for the government.
And there was a lot more challenging problems when I went there because now the government, as you know, is always interested in the leading edge in terms of efficiency because many of the assets, as they call it, are deployed where the power is a big question.
Energy efficiency is a huge problem. And so that led my experience to sort of drift towards designing architectures that would be more efficient. And guess what?
Our brains are a great example of trying to understand what's going on there. And so I got interested in that area and have ever since sort of worked heavily on that.
I started — my first opportunity came 2004 when I worked on a DARPA-funded program called Biologically Inspired Cognitive Architectures.
So that allowed us to sort of figure out various computer elements that go into an architecture for doing efficient processing.
And from then on, I had the fortune of working on several other DARPA programs at HRL, including one called Synapse, which is a very famous neuromorphic computing program that was basically the outcome of which was a TrueNorth chip from IBM.
And as part of HRL we did a small chip that did some learning, as well. And then Intel came calling. And they said, well, we like what you're doing, the HRL chip.
Would you come to Intel and help us design the first generation of Intel neuromorphic hardware? And that was quite exciting for me.
And as you can see, my transition from a mechanical engineer to a hardware architect and looking at the spectrum of problems along the way helped me sort of transition from one to the other.
And today, I'm more known for my work in the neuromorphic area than any of my past work. So that's how my career shaped up to become more aligned with computer architecture, if you will.
Yeah.
BORA NIKOLIC
Thank you. That sounds [INAUDIBLE]. I have an echo. Can you tell us a little bit about what this neuromorphic computing — just a little bit more of a resolution that what I said?
You're muted.
DAN GARCIA
You have an echo when you're not speaking. So I had to mute you. Sorry.
NARAYAN SRINIVASA
Oh, sorry. Okay. So basically, a neuromorphic computer, you can think of it as — like the video says, there's no clock in the system.
Furthermore, the instruction set is mostly based on what the neuron and the synapse do in terms of compute.
So basically, you have this multi-core architecture where each core houses a bunch of neurons and synapses that are basic compute elements and the operations of which are sort of dictated by instruction sets that are not exactly — does not really map to what you see in a typical computer.
And then like you pointed out, there's this a multi-core system.
So there's an asynchronous compute element here where each neuron integrates information and sends out its own sort of a pulse that dictates how strong the information of input was to that particular neuron.
And then there's a stream of pulses, and they're all acting in parallel.
So it's about handshaking between neurons and when they receive information to process what they get and then pooling that information across other neurons to sort of make inferences about what's going on.
So thereby sort of avoiding this need for a global plug that you typically see in a computer architecture — synchronous architecture today.
So notwithstanding that, how do you design algorithms and software that allows you to run applications that are useful, given this paradigm is the other side of the story that we worked on?
And what it turns out, it's sort of — the class of algorithms that it actually works well for is sort of complementary we find to the ones that we normally are used to from the Intel side of things.
So general purpose computing versus this one, it's things like graph search or things like sparse coding, or things like constraint satisfaction.
These are all problems, domains that are sort of up and coming, but they're sort of orthogonal to your normal machine learning Mac operations kind of things that you see today.
And they complement each other pretty well.
And so Intel is very interested in figuring out that space of compute domain that neuromorphic can offer its energy efficiency for, and then see how we can go build chips that people would be interested in using.
So that's where things are now. Yeah.
DAN GARCIA
Can I ask you just a very brief question? Or, I didn't mean to short circuit you. You are doing the interview today.
BORA NIKOLIC
No. Go ahead.
DAN GARCIA
The students just finished a project called The Game of Life. You have a grid of black and white cells, and they live or die.
And then you have the next — it's very much a heartbeat kind of problem. It's this. And it's an array. And then there's the next day and the next day.
How would that kind of a problem be encoded in your system? How do you — I mean, you can't just have them all random because they depend on each other.
And it's like, you have to take this day minus 1 affects day I and the next day, then plus 1. So how do you do that in an asynchronous way?
Can you help us with that?
NARAYAN SRINIVASA
Yeah. That's a good question. So think of it as a Sudoku is one example.
A constraint satisfaction of the Sudoku would be one where you have a state, and then you do — to make a move. And then it transitions. Right?
So the way our system is architected is, you don't really have to do — so each neuron, let's say, represents one particular node in the Sudoku board or in a Game of Life thing.
And they sort of locally integrate information from everybody else that they communicate with.
For example, the 3 by 3 square in Sudoku, a neuron would sort of look at the states locally there and pool, and to sort of infer what it needs to do based on the constraints it needs to satisfy.
So that can happen sort of asynchronously to another square next to it or the one after that as long as there is a communication that allows you to do a global concern at some point, some checks at some point.
So you can sort of see a sort of a demarcation of asynchrony versus synchrony even in that problem.
You don't have to all compute it with a clock, but you can sort of break it down.
And especially if it's more parallel, then you can sort of see a combination of the two in that case. So yeah. I don't know if I've answered it or not.
But yeah.
DAN GARCIA
No. That makes sense. I can almost imagine a Game of Life where you have two different areas, where there's like ones — the alive creatures are different areas.
And this, because — NARAYAN SRINIVASA: They don't talk to each other.
DAN GARCIA
They don't talk to each other, and there's a gap. And so this might actually be going faster and being able, like — this might be a generation 1,000.
This is still a generation 5 because there's more stuff to compute. So it's an asynchronous model.
And if they ever were to bleed together, the guys in the middle would say, look, I'm not ready to have my final state until you're both left and right are at the same day.
NARAYAN SRINIVASA
That's correct.
DAN GARCIA
And if I'm day 1,000, I need both guys day 1,000 to compute day 1,000 plus 1. And if you're day 100, I'll wait for you to get till you get to day 1,000.
And now that you're at the same place, now I can do 1,000 plus 1.
NARAYAN SRINIVASA
Exactly.
DAN GARCIA
I think I — that makes a lot of sense. Really interesting. Boy, that's really, really interesting.
BORA NIKOLIC
All right. I know what Farhana is doing. I mean, she does a lot of stuff. But can you tell us what you're working on?
So others know what are you working on, Farhana?
You need to unmute yourself.
FARHANA SHEIKH
Yeah, yeah. Sorry. So yeah. So I'm actually come full circle.
So I'm actually working on heterogeneous systems integration. So I was a system engineer a long time ago. And now what I'm doing is I'm actually working from the circuits level.
And I'm actually a little bit deeper into the technology side of it and trying to figure out how to advance our technology to support heterogeneous integration.
So 3D stacking is really something that I'm looking at right now in my current job.
And so I'm trying to stack RFICs, so wireless chips with digital architectures and digital processors and FPGA. So that's my current work.
So it actually requires you to understand software. It requires you to understand architecture. It requires you to understand technology and circuits.
BORA NIKOLIC
And can you say, like, in a few sentences, what is heterogeneous integration so a typical sophomore can understand?
FARHANA SHEIKH
Okay. So heterogeneous integration is really taking things that are disparate, like, things that are different and trying to put them into a single package. For example, you could have a processor.
But at the same time, you would want your processor to say, interact with a wireless chip. So let's say a modem or something. And you'd want to bring those two things very close together.
Typically, we would have processors located on a separate package and a modem on a separate package. But you want to bring them together in the same package. So that's what we call heterogeneous integration.
And actually, if you open up your Apple iPhone, and you deconstruct it, you'll actually see one of the biggest examples of heterogeneous integration in there.
BORA NIKOLIC
Is that related a little bit to this thing about Moore's law may not be with us forever? So I think some students may be worried about the health of Moore's law these days.
And is this a relief, I think, to not have to keep pounding on Moore's law forever?
FARHANA SHEIKH
Yeah. [INTERPOSING VOICES] FARHANA SHEIKH: So a lot of people are concerned that Moore's law is running out of steam and we're not going to get to smaller nodes.
So what's going to happen?
So really, Moore's law actually mentions this in his 1967 paper where he says that at some point, what we're going to be doing is we're going to be working on putting different chips together, chips that do different things.
And we're going to be putting them together in single packages. And so this is kind of extending Moore's law beyond just technology scaling.
BORA NIKOLIC
Perfect. Thank you. So there is one question. But I'm going to ask some questions on behalf of students.
One question is, where do you work? Are you in headquarters in Santa Clara or somewhere else? I think I know where Farhana is. I mean, you can see the hills behind her.
[INTERPOSING VOICES] NARAYAN SRINIVASA: Yeah. I'm also based out of Santa Clara.
FARHANA SHEIKH
Yeah. So I'm in Portland, Oregon. And the interesting thing is that Portland, Oregon has the majority of Intel labs there. But Intel labs is spread across Santa Clara and China and India.
And I was at Intel labs for 11 years. I recently moved over to the Altera side of the business, the FPGA side of the business. And I work in the CTO office there.
And so all my team is in San Jose, but I'm in Portland. And I get to when — post-COVID, I'll get to go back on the plane every week to visit my team.
BORA NIKOLIC
Uh oh. She said CTO's office. You know, she's not — FARHANA SHEIKH: It's called the — it's called the FPGA.
BORA NIKOLIC
Yeah — FARHANA SHEIKH: Yeah. It's called the FPGA CTO office.
BORA NIKOLIC
Yeah. A question that we typically ask — do you guys hire interns?
And what would it take to — if you're a sophomore in Berkeley in computer science or electrical engineering and computer science, to get an internship at Intel?
I mean, is there a — how do we go about that?
FARHANA SHEIKH
Narayan, do you want to go first?
NARAYAN SRINIVASA
Yeah, sure. Yeah. So actually, Bora, we have what is called the Intel neuromorphic research community, or INRC.
So this is basically set up to work with the academic community and students to sort of help look into problems, this problem space of neuromorphic computing, and see they can encourage students to sort of learn about it as well as to apply their own ideas to create new possible applications given
the hardware we develop. So there are opportunities every year that we look into for hiring some students as interns to sort of explore some space of interest that Intel may have.
So if there's interest, I can send you a link to share with the students so they can definitely reach out to us. And we'll be happy to look into it.
FARHANA SHEIKH
Yep. And I can — BORA NIKOLIC: Please, please, send us a — send us a link, and we're going to put it on Piazza right away.
FARHANA SHEIKH
Okay. Yeah. And I just want to follow up with that is that, yes, we have opportunities at all levels. We have opportunities at the undergraduate level.
We are interested in folks that are doing their bachelor's and coming on board and working at Intel. And we also have opportunities to do something graduate [INAUDIBLE].
BORA NIKOLIC
Yeah. And one thing that I wanted to add, last fall I was teaching 151.
There was a myth at some point that you need to have an advanced degree in order to be hired by these hardware companies.
I can tell you, the largest number of people with bachelors level were hired by Apple. The second largest went to Intel. So it's totally possible.
I mean it's very doable. But we need to get work on these internships a bit more. They're a big stepping stone.
So if you guys have an internship program that you can share, send us over.
NARAYAN SRINIVASA
Definitely.
BORA NIKOLIC
With that, all right. It doesn't look like there any other questions. I think we have sorted out most of the stuff. Let's thank Farhana and Narayan.
Thank you very much.
FARHANA SHEIKH
Thank you.
NARAYAN SRINIVASA
Absolutely. Our pleasure. Thank you for inviting us. FARHANA SHEIKH: yeah.
Thank you very much.
BORA NIKOLIC
Thank you.
FARHANA SHEIKH
It was great to be here.
BORA NIKOLIC
Thank you.
DAN GARCIA
Thank you for joining us. Wonderful.
BORA NIKOLIC
Thank you.
NARAYAN SRINIVASA
Thank you.
BORA NIKOLIC
All right. And we are going to move on. And feel free to — DAN GARCIA: All right. Schedule.
Here we go. So Bora, you want to cover that while I'm doing the spotlight thing real fast? Yeah. Jump on in, if you want.
BORA NIKOLIC
Yeah. Let me see. Hang on, hang on.
DAN GARCIA
Today — I'll just say it. We are at Week 9, Monday, Week 9, 10/19. It's the last lecture on pipelining. You're closing off the idea that we're making a computer faster.
No difference in the output, just a little bit more efficient in terms of being able to do multiple things at once at the pipeline stages. Then my set of lectures comes in.
And we're going to have four lectures on caches. And very important — this cache information is not on the midterm.
So you can stop worrying about the midterm after Monday's material and then just focus for four days on the midterm. And again, the midterm is Sunday and Monday.
You have 48 hours to do that. So you've got kind of like, a week, a full five or six days to work on the stuff that you need to practice on before then.
There's no lab this week.
BORA NIKOLIC
Yeah. So I wanted to — DAN GARCIA: Go ahead. Yeah.
BORA NIKOLIC
So what I wanted to add about that, today's lecture is — it was posted last week. But it is only about 35 minutes. I mean, it just basically ties the loose ends in pipelining.
And it is all what you need to complete both project 3D and the last piece of a puzzle for the midterm. And also, there is no lab this week.
So try to use that time to fully digest this material.
DAN GARCIA
Sounds good. Let's see. And the midterm material goes up to Monday here. And keep queuing the questions up.
We got a lot. I want to make sure to get to the next piece. This is a really important piece.
Bora and I did a — BORA NIKOLIC: --sanity check — [INTERPOSING VOICES] DAN GARCIA: --check. You want to talk about it a bit?
BORA NIKOLIC
Go ahead.
DAN GARCIA
No. Well, we did a sanity check to see how you were doing. We just wanted to say in the health and wellness — workload and wellness — that's one element of it.
But the other is performance. How are you performing in this class? So I just wanted to give a sanity check.
And then Bora can give — and somehow I have — this is — it's green. So my green is showing Berkeley. This is like a Berkeley — I love this.
This is like a basking of the Berkeley picture. There's a lot of purple in that picture. Folks are doing really, really well.
I think if I turn off my — I adjust this, that will adjust it. But folks are doing really, really well on project 1 in terms of that, just really, really doing well.
And very, very pleased with project 1A. Just really, really pleased with 1A.
BORA NIKOLIC
So what I wanted to say — I mean, what does it mean when the mean is — when the median is 100 points out of 100 points?
It means a little over than half the class is getting A pluses.
DAN GARCIA
For this assignment. For this piece. For this particular assignment.
BORA NIKOLIC
For this assignment.
DAN GARCIA
Yeah, Yeah.
BORA NIKOLIC
We can take a look at the next one.
DAN GARCIA
Yeah. Here we go. Next one. Let's see how we're doing.
How is the class doing? Here's 1B. Wait. It didn't change.
You're right. It didn't change. Look, if even got better. Okay.
Okay, maybe the mean is a little lower. But I mean, out of — what do we have total students? 1,100, maybe students? 970 are acing it.
I mean, just remarkable how proud we are about folks putting the work in and getting it. So you're doing a great job.
BORA NIKOLIC
Let's move on to the next one.
DAN GARCIA
All right. Let's — yeah.
BORA NIKOLIC
2A.
DAN GARCIA
That's 2A.
BORA NIKOLIC
Still the median is 100%. The median is still 100.
DAN GARCIA
By the way, I don't think this is usual, by the way. This might seem usual to you. This isn't usual. Normally, there's a distribution.
People are just crushing it. I mean, maybe we're adjusting it, pulling it back. But people are crushing it this year in a way that I haven't seen before.
If I pull up — I want to do some deep data to look at like, last year and a year before, year before. I don't remember seeing these numbers.
I really don't remember seeing these numbers.
BORA NIKOLIC
Yeah.
DAN GARCIA
It's remarkable seeing how much success students are doing here.
BORA NIKOLIC
So we are very pleased with this. I mean, we can move on to project 2B, where there was some discomfort about how things were graded. We have a new distribution.
The distribution has changed a lot in the next slide.
DAN GARCIA
Here we go. Here's the 2B. And this changed a lot. This changed a lot after we took a look at the auto-grader, and Stephen put a lot of effort into thinking about the points.
And I think we were initially too harsh. And I thank you for that feedback. I really want to thank folks for the feedback about 2B. And I think we've corrected it this way.
This is what you — by the way, this kind of graph is what you normally see for 1A, 1B, and 2A.
We normally see this kind of shapey thing where it's normally — it's heavily weighted towards the higher hump. But you don't see people acing it.
So I love the fact that people are just really crushing their first three parts of those things. Just great job, folks.
BORA NIKOLIC
Yeah.
And so the big news here is before Stephen spent many hours last night in trying to reassign partial credits to the parts of a project that we're failing on — DAN GARCIA: Failing the test.
Failing the hidden test.
BORA NIKOLIC
It's fine, compliance — Yeah. Hidden tests on the registered conventions.
The big thing here was we had 150-something students with zeros or close to zeros. And that definitely did not feel good.
So we tried to assign partial credits to that. Dramatically, they didn't change much our mean.
Hang on here. Yeah. It didn't dramatically change our mean, but it shifted the median by a lot, by 20 points or so. So this looks more normal.
Now if we exclude the Quest, because there are still two more opportunities to clobber the Quest. So this is an — first interim score on Quest that we have now.
At the moment, based on what has happened, or the grades that have been assigned so far, we have about 2/3 of our class that are getting As.
DAN GARCIA
2/3. That's never happened before. That has never happened before at Week 8 that 2/3 of students are getting As. And by the way, you know Bora and my spirit of this.
We want to give everybody an A. We think if you earn an A, you will get an A, and we will celebrate that. There's no curve in this class.
This is an absolute graded class. And folks are doing really, really well. And the question on the thing was getting As.
And getting As in all — if we just look at the projects, if we ignore the Quest for now, the class so far — because the Quest really was not a typical Quest.
We think the midterm will do much better, and people will then overwrite the Quest. It will just be a bad memory. Or we'll all go back and laugh together.
We'll share a sparkling water, and we'll shake our glasses together and that. But I think the students are doing remarkably well right now, really remarkably well.
Although that said, there's a chunk of students that aren't. And. let's bring that up. Let's actually talk about that.
BORA NIKOLIC
So when we look at this, so an average student in the class or a median student in the class is doing spectacularly well.
DAN GARCIA
Right.
BORA NIKOLIC
However, we have a long tail. Every time we teach 61C, there is a tail. There are people who are one week, two weeks, three weeks behind. And that is evident this semester.
I mean, this semester is more difficult than any other semester that we have. And we do want to help.
So we don't want to — the way how we designed the following 61C relief package — DAN GARCIA: Here we go! The relief package. Let me queue it up so you can see.
BORA NIKOLIC
Exactly.
DAN GARCIA
The relief package!
BORA NIKOLIC
To help those — DAN GARCIA: Go ahead. Go ahead.
BORA NIKOLIC
To help those that are not doing well, to give them some relief to be able to catch up while providing an equal opportunity to those that are doing great to continue doing great.
So let's get into it.
DAN GARCIA
We have a 10-point relief package. 10 point with this package. And I want to spend some time. We spend 20 minutes talking about this relief package to help everybody.
By the way, it doesn't help to help the — just like normal relief packages, don't just help the folks in the bottom, they help everybody, in theory.
So this is going to help everybody in theory. But particularly, we're focusing this on students who are struggling.
BORA NIKOLIC
Yeah, exactly. So the first thing, what they wanted to — also we wanted to say, we designed this course to be a little bit heavier in front.
And this naturally, there was a drop-off in the amount of things that needed to be done every week.
But we are making that dramatically easier so people can catch up way before the semester is over.
The first thing that we have to deal with, there was an issue with free speech and deletion of some posts because things got a little bit out of control on Piazza.
What we did is, we talked to the TAs, we talked to the students. And we do understand that there are some posts that were inappropriate.
But even though they were not completely — they're a little bit heated, put it that way. They should not have been deleted. And we are not going to delete any posts going forward.
We talked to our TAs about trying to be better at moderating these exchanges before they become heated. So let's try to keep it that way.
We also do want to ask students to work with us. And we are going to be very supportive.
We are going to work with our TAs to make sure that these discussions never get out of hand.
And we are going to be in here to step in like we have stepped in with the grading or regrading of the project 2B.
There is no reason for things to get out of control and get very heated and to have inappropriate posts from anybody. And so let's not do that.
If there is an issue, bring it to — put it on Piazza. If you don't get an appropriate response, you can escalate it to us. No need to get very heated about it.
We are very, very, very, very reasonable about that.
DAN GARCIA
And we appreciate the feedback. We also appreciate the feedback.
I mean, every time you — there was a concern that — if I'm critical of the class, I'm going to have my EPA points docked. Actually, it's the opposite.
And we'll talk about this in a moment. We want to hear that feedback. That's why we give you so many surveys. We want to hear from you.
And we're going to Institute some more policies.
As part of this relief package we'll actually talk about how we're going to more open channels of communication to help and get a pulse of how the students are doing.
That's what the purpose of all these surveys were.
BORA NIKOLIC
Exactly.
DAN GARCIA
Yeah.
BORA NIKOLIC
Exactly. So let's move on. But again, we are going to sanction. If somebody behaves inappropriately, don't test what's going to happen.
DAN GARCIA
Right. Yeah. I mean, we will bring in the Center for Student Conduct. I mean, that's what happens.
Right.
If you're violating in some sense Student Conduct, if you're making the place uncomfortable for people to be there because you're doxing people, you're using expletives, you're doing other stuff in a space that is a learning space — it's just like if you were doing it on campus, we'll pull the Center for Student Conduct in.
We'll delete that post. So we're not going to not delete posts. We'll delete that one.
But when we delete a post, it's because we're pulling you into the Center for Student Conduct for something that's way inappropriate. By the way, this happened once in my 20 years here.
So I don't think this is going to happen. And criticism is not that category. Criticism, we welcome that. That's not what that is.
I'm talking about someone doing something way above where it's clear to everybody in class that was inappropriate. Don't do this. Okay. So we'll try to be very clear when that happens.
BORA NIKOLIC
Yeah. And we are not going to delete any posts. Dan, to you.
DAN GARCIA
Yeah. So this is a big one. We had a wonderful meeting with two students — actually, three or four students — over the weekend.
And my eyes were open to how much folks are struggling more so — and in fact, it was really useful not just to hear another data point about that, but to actually have a working path forward.
And we were kind of throwing out how would this work? What do you think? So it was like a feedback space. It was really wonderful.
And I thank those students for being there.
One of the things that came up — and a lot of things came out of this meeting, so I thank for them — is, they were very thankful as a representation — a face-to-face representation — that we made project 2 a partner project.
We mentioned before, we had a fear that by doing that, that could be a solution where one student doesn't do the work and just buys a coffee for the other friend, they do the work.
We're just hoping that doesn't happen. We — please don't do that because you'll miss the opportunity to learn it yourself. We decided that it was better to do that.
It was better, both sides all being equal, it's better to have that person to ask questions to and support and work together positively.
We've decided to take every homework that was never a partner — never homework partners ever in 61C as far as I remember to make every homework going forward a partner, if you want to.
We won't force anybody there. But we encourage everyone to have partnerships in your homework to do together to talk about it. You'll see that homework is like, online.
You click, kind of like it's like a EdX kind of thing. But it's on grade scope. But we encourage you to do it live. Get on a Zoom call, talk your way through it.
Grab a piece of paper, work on the problems. And learn together. So we want people to learn together, form learning communities. So boom.
Number 2 is, reduce your workload because you're working together in homeworks. Hopefully that's a big way to help with that. So that's the second one. Over to you, Bora.
BORA NIKOLIC
The next one — yeah. The next one is homeworks 9 and 11 are now not graded. Or actually, you get 100% whether you do them or not.
That's supposed to free up some time. It is not a huge amount of time. But it frees up a few hours for each one of those.
So you can use that time to catch up. If you're caught up, please go ahead and do these homeworks. Next one is in the same spirit.
DAN GARCIA
So on 3, the homeworks 9-11 are still out there. So we encourage you all to do them. But if you can't do them, no problem.
We'll have — TAs are there to answer any questions about them. But again, you get the free points. We're just giving free points away to support folks with the workload.
We had a really big conversation about labs. And labs could be, oh, it's a half hour lab, it's in hour lab. It's a five-hour lab. That's a really big deal.
That's too many hours in that.
So we decide — and by the way, the spirit of a lab check-in was, it's the single place where every single student has to have a face-to-face, high touch experience with a staff member in 61C.
We spent all our TA budget in that, really. And the idea — face to face, it would be a lab and people — our AIs and lab TAs are all walking around answering questions.
It's a very healthy Q&A. You raise your hand, two seconds later, people come over. They swarm, and they answer questions. It's really healthy. You get a check-in.
And they verify you know it. We decided to have these lab check-ins here.
And the idea was, for every 15 minutes or so, a certain amount of time, maybe three or four or five minutes, is a health check-in. Social emotional health.
How are you doing? How are you doing with your life? How you doing with your health? How are you doing with 61C?
Are you caught up? Are you confused about this or this? It's like a check-in. And then we start to the technical Q&A of like, well, what happens here?
Well, what if you do this, and what if you — Okay. So that's the last part of it. And we were hearing this isn't working. We were hearing this isn't working.
A, there's no check-in. There's no health check-in. And B, there's almost no time to do that.
And students who are acing it and don't need that support, stop using up TA support to prove that the student knows it when they already know it.
So we decided that this is really an experiment. I've never done this before. We're trying this. We believe it's more important to do the health check-in than anything.
We want to know how are you doing, to have a chance for every student to have a very small meeting — you and your partner — with a TA, with a staff member.
So what we're going to do is, we're going to change the lab check-ins. You're still going to get — by the way, for the homeworks, we're just giving you free points.
That's how we deal with that one. But for the labs, what we're going to do is just show up. Talk to your TA. Tell them how you're doing.
Talk about how your class is going. And then ask — the rest of it is yours. Technical check-in driven by student. We encourage everyone to keep doing the labs.
Do the lab, then ask the questions of the lab of the TA. So rather than the TA probing you, you are probing the TA. You know, I was stuck over here.
What would happen in this case? And the TA is helpful, happy to help. So it's driven by the students. It's whatever you need.
If you need to just talk and vent, use the time to talk and vent. You will get points. So come to those sessions to have a check-in. You will get points for showing up.
That's the idea. Because of this, now we're not allowing any more late check-offs because there's no more content in those things. It's about a health check-in.
So it's not like, well, I can just show up to at the end and do this. No. You have to every week check in.
And you get the free points if you just show up and have a conversation with our TAs. But again, if you want to make them technical, make it technical.
Use the whole time to get support, one-on-one support. You want to use the whole time talking about your project, a bug you got, use that time to talk about the bug.
The point is, it's for you. We're still testing the materials. So it's still out there.
But if you're behind and you're, oh, I can't breathe, I can't do this — labs is at — Fine. Do it at the end. Do it during RR week, if you need to.
But let's be flexible for yourself. That's the idea. Okay? All right.
BORA NIKOLIC
1.5. Fairness. Project 2B. We already said put a lot of effort in regrading it.
And we reshaped the distribution by assigning more partial credits. Keep in mind that register calling conventions are really, really important.
And even though if you do make a function that works on its own, if it cannot be called by something else, then you have a problem.
I mean, it's going to be actually a nasty, ugly problem that you're going to have a real hard time debugging in your practice as a practicing engineer. So hopefully, this was a good lesson.
But it was an expensive lesson. Again, in the grand scheme of things in a number of points that people have lost, it's not a lot. But it was still felt like a lot.
So we went and reshaped this distribution. So hopefully, people are feeling better about that, about the number of points that they've lost. So it's just a small penalty on the calling convention issues.
Hopefully, people are feeling a lot better about that now.
DAN GARCIA
And by the way, we thank every single student who brought it up to our attention that this was a little unfair. And I got some emails. And I want to thank you all.
And by the way, you're going to earn EPA points for being able to speak up. That's — EPA is Effort, Participation, Altruism.
I count participation and altruism is what you — by speaking up for the class. I speak up on behalf of the trees. That's the Lorax. I speak on behalf of the students.
That is what EPA is about, being a good student. So bring that up, talk to us, improve the health of the class. EPA points for people who brought it up to me.
Everybody sent me mail gets EPA points for doing that. Thank you for that. Okay. Relief Package point 6.
Again, this is — I just got to mention this a second ago. And by the way, number 14, 15, don't worry about that. That's our internal numbers for that.
So you know what that means. There's never points off for complaining to us about feedback. Again, we're going to — in fact, we're going to include students in our staff meeting.
We're going to have a spec — it's never happened before. All these things are brand new in this case. So we're going to have a couple of students as representative of the class.
And I'm going to ask the students to actually coordinate yourself. And those folks will attend our staff meeting and give us a pulse.
So rather than hearing, waiting until this meeting, which is so great over the weekend, it was Week 9 I heard about these things. I want to hear them every week.
So we're going to have a chance every week to hear from the students and really support them and say like, how did this go? How does our relief package going?
And should we make other changes? We have the power of the control of things. Should we drop something else? Do we need to drop another homework?
How's the work there? We want to support students in that way. Bora.
BORA NIKOLIC
Yeah.
The other thing, so what we are realizing from Piazza posts and also from the quiet probably not majority, minority in this case, there are students who are behind by a week, by two weeks.
And there is a huge appreciation for a session — a lost session — that Caroline is doing on Fridays.
We are going to add one more that is perhaps going to trail Caroline a bit. We are trying to find a resource for that. But I think we will get one.
So we'll have another catch up. So that you don't have to have a choice — am I going to study this on my own?
Or I'm going to try to attend the current sessions and try to somehow go through this?
We'll open up some time hopefully with a little bit reduced workload that will — or significantly reduced workload with the homeworks — and these catch-up sessions, people will be able to put time into that to catch up such that we will have everybody pass through the gates to the finish line together on the same time in the end.
Next point.
DAN GARCIA
Again, this is under the theme of support for struggling students, the folks who are not in that median of 100 for most of those folks. We're going to try to do two things.
One is, the goal of moving to Discord was to simulate what happens when you're in a normal face-to-face line for office hours. There's an office hours queue for project office hours.
And you got a bug to get fixed. And normally, there's a line outside the door. And then students are sitting with their laptop, and they're helping each other.
They're talking continuously. And often, people will get their bugs solved. And they leave before they even get to the TA. The purpose of Discord was to do that.
And I'm learning from students that they're not doing that. So idea — I don't know how to support that other than to say, that's the place. While you're waiting, don't wait for six hours.
Wait and be talking about things. We love that. Help each other. Don't give the answer away, but help people debug.
You know, help people through the process of that. So what we want to do is we want to create some videos. And.
I'm going to ask my people who are the most Discord savvy to create some videos of kind of a best practice of how do — how to use Discord to get those questions answered, how to join, how to do this, and how to be — while you're waiting — be connecting with other people.
In addition, we created a help video for the project, kind of like a, here's how to think — and we actually talked about it during this live session a couple of weeks ago.
So we're going to try to do the same thing where we have — hope to do is to prepare you for Project 4, which is a fun project.
But it's a big one. So we want to make sure you're ready for that. So help you is a kind of — make sure you pay attention to this.
That'll be in the Help video. Bora.
BORA NIKOLIC
Yeah. The other thing is, this is also related to that. In normal semesters when we are in person, people have — students have time to develop relationship with their TAs.
TAs get to know them. They know where they are, what are their strong points, what are their weak points. But we have a bit lost that this semester.
So what we do want everybody to sign up for the same slot with the same TA for the office hours such that they get to know you.
One of the challenges that we see here is that TAs are not exactly getting a handle on knowing exactly who you are. We do want you to develop a relationship with TAs.
TAs have a chance to keep their hands on your pulse and know how you are doing. If you are doing well, or you're suddenly struggling out there, they can tell us that.
And we can do something. We can intervene. So please, please do that. And we do want to have that in there.
We want you to be participating with the TAs.
DAN GARCIA
Great. And finally, this is the biggest one. I saved it till the last one.
BORA NIKOLIC
This is crazy.
DAN GARCIA
This is the biggest one. I talked to Bora about it. And he was on board. So I want to mention this.
This is a really big deal. I want to have a safety net for all of you. I want to make sure that none of you feel like it's just, I can't deal with it.
It's just not worth it. This course is too much. I'm going to offer right now to say that I will not fail anybody this semester. That's not to say I'll give you all passes.
What it says is, you say, well, the course got off the rails at some point for me, and I couldn't keep up with it.
All you need to do is, mark that date for me. I'll have a form that people can fill in. And I will give you an incomplete. I've never offered this before.
I offered it to CS10 in the spring, and it worked. Never offered at this scale to 61C before. What it means is, you'll get an incomplete for the grade.
Choose this if it's just not going well for you. And you say, Dan, yes, I even did get it to Week 15, but really I was really not doing well since Week 10.
We'll go back, and we'll backdate your incomplete to Week 10. And your grades after Week 10 won't be counted. And you'll be able to redo those again.
So it's a chance for you, if you were going to end up in the failing area, to — first of all, you can always reach out and say, I just want an incomplete.
And I wlll say yes to every incomplete request, period. And then everybody who would have normally failed — which is a small percentage, maybe like 5% or something.
Very small. But I hate it anyway — I'm not going to give them Fs or NPs. I will give them I's.
And then you will have next semester or the summer, and I'll always extend these incompletes as long as they take. Okay? So no fails this semester.
Only incompletes for those folks who would have failed. And reach out if you need some help. That's out there as a safety net. So don't think that's the end of the world.
We're here for you.
BORA NIKOLIC
But do reach out. We have seen people who have been dropping the class. And I said in Intervene, hey, I mean, you don't have to drop it. We can give you an incomplete.
I mean, you don't have to redo the whole thing and restart from this point.
DAN GARCIA
Yeah. You put work in. Keep the work that you've kept. Don't throw that all away.
Exactly. If you drop it, you don't get to keep the work you've done. You get incomplete, you do. That's a really big deal.
I don't know if any other class — tell me if another CS class is doing this. I'm doing this. I did it in the spring. It worked.
I'm doing it again here to support everybody at the bottom — at the bottom level here.
BORA NIKOLIC
I mean, these are extraordinary circumstances, right. I mean, there's justification for an incomplete. It's default this semester.
DAN GARCIA
Out of your control. And COVID is out of your control. And this whole situation is out of your control. This whole system is out of your control.
All right. We got some more things quick. It's at 5:00. But I want to make sure I cover these things.
53 people took the Quest, a revision of the Quest last Sunday. We did not shut down Piazza. There was so much things happening this weekend, we didn't want to shut that down.
But we let it happen. People were very, very good. We let every student know you need to ask a private question on Piazza, 100% of you did the right thing.
So thank you for all walking the line and keeping that, because otherwise it was going to be an issue where you're giving the answer away. And all of a sudden, students are [INAUDIBLE]..
So when we shut it down, people were very respectful and did this. We will shut Piazza down for the big midterm, because that's for everybody.
But we didn't want to shut it down for this small group of students. And it was similar to the Quest in terms of style.
Not everybody had to do all the parts, only the parts that we didn't have a video for you for. So those folks had to do them again.
We're going to have to write custom auto-graders. You saw how long the regrade took. It took three weeks for that. Now it's a smaller load because it's mostly the same as those auto-graders.
But we have to write them now. And so hopefully you'll hear back from us this week about your score. Finally, the big deal. We've got a midterm.
The midterm is scheduled. You know about this. This is last week's information. It's this coming Sunday and Monday.
The material covers — Bora already just mentions — up to and including pipelining, which is today's lecture. There's going to be exactly five questions. The topics are on the bottom.
You've got a Quest redo, which is coding in C. You've got a float question, very small lightweight question there. RISC V assembly, which has some coding. So practice that.
Data path control and pipeline, and an SDS logic. We will give you tests when appropriate.
There's one small part where I can't give the test because really, the test is about, can you think about how to test it. But I'll give you a structure for the test around it.
So be a structure for two tests there ready to go there. Time limit roughly an hour per question. We're going to share our TA test times. Our TAs are beta testing it now.
Or they'll be testing, I think this evening they start. And all those beta testing times we're going to share with you.
We'll say, a TA took this many minutes, this many minutes, and this many minutes, the range. And then we're going to take that time and multiply it by a factor like 3 or 4.
And that's how much time we'll give you. So we'll take the maximum time any TA took to take it, multiply it by a factor, and that's what you should take.
Okay? You should expect to take at least two hours, but I would say 2 to 3 is the safe space in terms of reserving yourself.
But leave yourself more of that because if it's an hour per question, that's at least maximum 5. But we're not going to say it's 48 hours.
But we're going to say, we think we can get everybody done at like, a multiplying factor of about 4 times the longest TA time ever. So that's what we're going to do.
We didn't do that with the Quest. We're going to do that here to give you enough time, but also constrain it so It's not a 48 hour just taking up the whole time.
Okay. Lots of questions in Q&A. I'm going to stay until those all get answered. So let's hear.
BORA NIKOLIC
Yeah. I did some. How to give you a Quest. So they're asking for Quest.
Now since the retake is done, when can we put — DAN GARCIA: Well, I think we already shared the answers, I believe. We already shared the answers for the first version of the Quest.
And if not, I don't have a problem with giving those — giving that away. I think that's a little piece of code, and we can give that away.
So we'll do that if that hasn't done already. So I'll answer it live.
There's an interesting question that has come up in our conversation, which we have — it's not part of our 10-point relief package, which is, could you use your slip days to resubmit project 2B?
Make the fix. Our hidden tests are still hidden. So what if you could use a slip day to actually fix your code and then resubmit for an updated grade?
I'm actually OK with that. I want to check in with Bora. I didn't get a chance to talk about this. But it happened why were we giving live.
BORA NIKOLIC
We'll discuss this.
DAN GARCIA
Yeah. We have to discuss it. We might have good news for you next week if we decide that that's OK. So we'll think about it.
But that would be amazing to be able to play with that. I love it. So I'll argue in your support for that. But other people may say, well, it's not fair, blah, blah, blah.
So let's hear all perspectives before we let you do that one. Okay.
So — BORA NIKOLIC: Well they can — obviously be effective for Lab 6 if we were planning to get checked off late.
DAN GARCIA
I think the check-in policy is now. If you have content, if you're ready to be checked off, go ask those technical questions.
I don't see why we couldn't start that as soon as we can, to be honest. I think — meaning you can get full points — BORA NIKOLIC: That's last week's lab.
DAN GARCIA
Oh. For last week's lab.
BORA NIKOLIC
Yeah.
DAN GARCIA
No. You should finish last week's lab. It starts from now forward.
BORA NIKOLIC
Let's keep it equal for everybody. Yeah.
DAN GARCIA
Starting making it forward. So check you off late. You can still get checked off late, but just use the time during your session to check off last week's one before we're done.
That's fair.
BORA NIKOLIC
It's a fairness thing. Yeah.
DAN GARCIA
Exactly.
BORA NIKOLIC
Can we check in for more than one day a week? And when do revamp check-in start? Revamp check-ins start — they can start this week. We don't have a lab this week.
DAN GARCIA
Yeah. Yeah.
BORA NIKOLIC
Can you do it more than once a week?
DAN GARCIA
I mean, we have to look to see if more — if you wanted to have two hours with our TAs, if we have the time for it. I don't know.
I think we have to continue to make sure that every TA can cover every student. So I'm not sure I'm going to guarantee that.
But I think it should lighten the load in terms of meaning. Some people may be using their check-in to get their question answered for their project.
And all of a sudden now, project office hours are lighter. And normal office hours are lighter.
So I think we're going to lighten the load when people are making use of these check-ins to just give more face-to-face time with your TAs that isn't about the lab.
You say, I learned the lab during RR week. We encourage you to learn it because sometimes the projects need it. But I'll learn that particular lab that isn't on a project later.
I'm going to use the time now when I really need it. That's fine. I think that'd be great. And that might lighten the load on other times.
So I would say right now, no. One check-in a week for you. But come to other — there's a ton of office hours all throughout the week.
Hit those, and do that one if you can.
There's let's see, 2B score was not supposed to decrease. That's a question about this. 2B was not supposed to decrease after these changes. That was actually one of the things we talked about.
But it may have. [INTERPOSING VOICES] BORA NIKOLIC: I mean, it could happen. I mean, in some corner cases, it could have happened. Please let us know.
We'll give you the higher of the two.
DAN GARCIA
Yeah. I think that's right. I think you get the higher of the two options. So there we go.
There's the max of either the old one or the new one. We can easily do that in software. So let's do that one. I think that's fair.
BORA NIKOLIC
Yeah.
DAN GARCIA
I believe — I don't know. EPA points are just a small fraction. They're not a big deal. But a lot of people are asking about EPA.
But I don't know if you would comment anonymously, whether at the end of the year we download the Piazza participation.
And we use that to inform — not to completely override — but to inform our EPA. And I don't know if when you comment anonymously, whether it gives you credit for that.
It's hard to test. You have to test it with a blank Piazza and make anonymous comment and see if you get a note for it.
I know who — Bora and I can see who wrote the comment. So the staff can see. I don't know if the software sees. I can ask that question.
I'll try to look into that. I'll just say that. I don't know the answer to that one. I would like it.
I would like to be able to be anonymous so you don't get ire from your neighbors. But you should be able to get credit, I would say, for at least participating in that way.
Go ahead. Bora, you want to grab one?
BORA NIKOLIC
Hey. We are at 5:06. So we — DAN GARCIA: Yeah, yeah. I know.
Yeah. Okay. I'll do easy ones. Midterm is open notes, just like the Quest.
The midterm is almost identical. Midterm, yes on PrairieLearn. Yes, almost identical. There.
That's easy.
BORA NIKOLIC
Think logistics.
DAN GARCIA
Yeah. Would the Quest coding contain number representation questions? The answer is, everything. Everything up till now.
So the answer is, everything's live. Who knows what's going to be on it? But yes, it would potentially contain number representative questions. Know that stuff too.
Important.
BORA NIKOLIC
How should we reach? Private Piazza? That's the best.
DAN GARCIA
Private Piazza is one way.
BORA NIKOLIC
[INAUDIBLE] DAN GARCIA: But the second way I think is — BORA NIKOLIC: There is email if needed.
DAN GARCIA
Yeah. For me, email is not the best way because I have too many.
But I would say probably reaching out, the best way to do this is, from at least my point of view, talk to the student group that's going to be the representative students' group talking to us once a week.
And that will be a chance for you to join that group or at least get your voices shared.
We want to try to have a small group of students talking to us representing the students. So talk to your representative who comes to meet with us once a week.
So we'll try to organize that and coordinate that. And I'll make with a student who volunteered to kind of maybe be part of that. So that's a great way to do that.
Boy, a lot of questions.
BORA NIKOLIC
Labs are — there is an auto-grader that is going to — there is still a lab auto-grader that you — but it just gives you 100%.
So that's — DAN GARCIA: Well, I mean at the end of the day — BORA NIKOLIC: [INAUDIBLE] DAN GARCIA: At the end of the day, the points in the grade file will be 100 if you show up.
But the auto-grader might give you feedback. So play with the auto-graders if there are lab auto-graders to learn whether you got it right or wrong.
You got to know whether you're right or wrong or not. But at the end of the day, if you showed up live, you get the points on the grade book.
Can I use the C reference guide in the midterm for syntax purpose? Yes. Download any reference guide you have in advance. You just can't search for it live.
You can't say, oh, the problem asks you for a, make this kind of shapey thing a snakey thing. You can't say, Google snakey thing. You can't do that.
But if you download anything in advance, you can have it. So a great idea. Get any C guide you need to do that. As well as bring up your C — every book.
It's open book everything. You just can't search anything live. So you don't do that. That's perfect.
How crowded are professor's office hours? That's easy. Bora, how crowded are your office hours?
BORA NIKOLIC
They were not in the past. I had about 10 people on Friday. But I did my office hours for one hour. I spent 2 and 1/2 hours on Friday.
And I intend to run the two-hour office hour again on Friday.
DAN GARCIA
My office hours, I have two of them. I have a one-on-one for particular special problems for s to talk about maybe a 47C thing.
But I have my general office hours are mostly what I do normally in the face-to-face year, which is, I go over mid-term questions, and I try to help people with exams.
But I also — we talk about anything that's coming up. But I mostly go over content in mine. So I'm happy to have both. The first one, the content is — I don't know.
What did we have? The people who have been there in that group, about 10? So I welcome people to come to my office — the same time on Wednesday is my content office hours.
And one on one, you can make it a one-on-one appointment if you have something special. But again, many one-on-ones can come to that representative thing.
Ryota asks a very good questions — BORA NIKOLIC: [INAUDIBLE] DAN GARCIA: Go ahead, Bora. Please.
BORA NIKOLIC
Yeah. We have two types of office hours. If you have something that is bothering you, come to the individual, and there is no one else.
You just come and talk to us for 10 minutes. If there is nobody in the queue with me — and typically after 5:20, 5:25 there is nobody.
If somebody shows up at — or sorry. 4:20. Usually there is a few people who are queued up with immediate issues. I clear that by 4:25.
And then somebody shows up at 4:50. So you can grab that time. Again, if I advertise that, it's going to get crowded. So those are my Wednesday spots.
Then the Friday ones are content ones. And I go over the issues or whatever the issues are. And just come in and ask questions.
Last time we went for 2 and 1/2 hours until we were done with the questions.
I can't guarantee I can do every Friday 2 and 1/2 office hour — 5 and a half hour office hours, but, I'll try.
DAN GARCIA
Right. Perfect. I've got two easy questions I can hit. Can you download websites for reference?
The answer is, yes. Pre-download websites. The point is, you just can't search for it when I give you a problem about the snakey thing, you can't search for the snakey thing.
But if you want to download all the questions you have, download them all. That's perfect. And that's Monisha's question. Good question.
And the second is, can you use local computer search? I've delved into the big directory. And I'm searching through my directories for where I put that silly thing.
Where did I put that website that had about — You can do that. You just can't ask Google. You can't ask Bing. You can't ask the web.
So yes, 100% local computer search. Very, very good. That's in the spirit of that exam. You want to jump on one, Bora?
Bora read more. Those are the two of the easy ones that I got here. Discussion — BORA NIKOLIC: How do I sign up to be a representative.
I think they're going to ask somebody to be a representative.
DAN GARCIA
We might be [INAUDIBLE] to a Piazza page, or who wants to be. And we'll have a couple of folks. And then what that — here's the point.
You'll coordinate — the representatives will then coordinate a student meeting. And they'll check in, what do we want to share with this? And then those representatives will come to our meeting.
We want to have a couple of people who are really on point with a bulleted set of things. But we want to have that conversation outside there.
So we'll try to let it be self-organizing Rio, do you want to organize it? Organize it yourself. And then get the group folks together.
And then you and the person who I met with earlier will then be representing. So we'll figure out who those representatives are.
You guys can choose among yourselves who wants to be the representative to come to talk to us. Okay? Love that. Thank you for organizing that or thinking about that.
We'll set up a Piazza thread for that. Good stuff.
Let's see. Saving slip days to the end. That's been an interesting conversation, a design decision. I think I've never allowed slip days to count for anything.
If you don't use any slip days, you can use them for points. We've never done that in the past.
I prefer you use your slip days to just give yourself — the point is, it's flexibility.
Let's say if you get sick or something, you get the sniffles, and you need to take a break, or you're overloaded. It's about that space. And here's the problem.
The reason I don't like people saving at the end is, if you get sick, I want you to use them.
I don't want them to become currency to become free points later because then when you do get sick, you'll just bury through it. I don't want that.
I want you to be taking your breath, rest, drink fluids and rest, and take the break when you need it. So I actually prefer them to be worth nothing at the end.
So that's my policy in the past. And I want to stay with that, which means you can still use them up for the last project. That's fine. So save them for the last project.
If you don't use them for the last project, you won't use them at all.
But I don't like the idea of saving them up for something else because I want you to be using them if you need them.
So don't — I think the answer is, I'm not going to let you do use slip days for points. It's an interesting idea. BORA NIKOLIC: it looks like there is no class.
There is no class in this slot. But that — DAN GARCIA: Oh, yeah. I think we're good with a class overlapping. That's — yeah.
More slip days? I think for now, no. I think we always have the possibility to give you more later. That was just a question here.
But we gave you three more. And now you have six. Six is a lot. We've never had more than three in the history of 61C.
And now we're doubling it. So I think that's enough to be able to be flexible, especially with no more labs you have to do, project, teamwork homeworks. I think that's a little bit.
A 10-point relief package was enough. So we'll think about it. It's in our back pocket to give you more. But for now, I think the answer is no for that.
We'll see if we can do that. Okay?
So the lab points, do they only come from health checkups plus technical questions if we want them? Standard auto-grader check off component. No, there's no standard auto-grader. There's no checkoff component.
You just — showing up and using the time in whatever way you want the time to be used gives you the points. That's it. That's basically free points now.
The lab — usually lab people, most people ace all the lab points. But now if you want to ask the questions about that, it's on you now.
It's actually more ownership of you to understand that material, less about, we need to push you because we're the big brother making sure you get — no.
If you need the questions asked, at the end of the day, we've got an exam, it's going to be testing that material.
If you want to learn it just in time, or you want to learn it during the lab, I would learn it when it's happening. Don't fall behind it.
So I encourage you to do it then, which I love. So that's great.
BORA NIKOLIC
It is important. Will the midterm have similar proctoring to the Quest? Yes, it's identical.
DAN GARCIA
Identical. It's identical, which means, plug in your phone! Oh, my gosh. Plug in your phone.
I can't tell people say, oh, the phone went out. And turn off notifications.
We can have a whole checklist — we're actually editing a whole document with, turn off notifications, plug in your phone to practice. And then try calling yourself. Try getting a notification.
Try to create a notification to see if it stops the Zoom recording. And by the way, of the two options, I actually found it much easier for students who created a time lapse.
So there's any way for you to download software to create a time lapse, I encourage you to do that one because I actually found that was much easier to see the screen and much easier — cause Zoom crashed for a lot of people.
But the time lapse didn't. I had far fewer problems in the time lapse than I had with Zoom. So think about the time lapse and get your phone to do that if you can.
Or download a free app or something like that. Great job. Five question means five midterm sections, like the Quest had three. Exactly.
Five questions. We already talked about them. That's on the slide. It's right here.
Look. It's right here. Here are the five — look at my fingers. See my fingers?
Here's my fingers. Those are the five questions. Each of those — sorry. Five sections.
I should edit this. It says 5 question. Each section might have multiple questions. Okay.
I'm sorry for not being clear on that.
BORA NIKOLIC
A few questions per section.
DAN GARCIA
Make sure you've finished the whole exam. People got confused on that. Don't do that. Okay.
So that's easy.
There won't be a guide for writing tests. Essentially, we're not going to grade you on the test. We're just going to give you an easier framework to be able to have tests.
Almost like, for example, in your Quest many of you had to do a split. You had a linked list of things. You had to split it in some ways.
There were different versions, but you just split this. Right? I didn't give you anything. Where now, I would give you something that if you did it, it would split a simple thing.
If you did it the right way, it would split a simple thing. It would test whether the split worked. We're going to give you that thing to test whether the split works.
That's the idea. Or at least print it out. We'll have something that I might even print a linked list that will even let you see it, whether it's split work.
So if you split it, and now I have two of them, I can print both of them. Did it work? So we'll give you that kind of framework to do that.
It's not a formalized framework. But it'll be an easier way for you to test if you start from 0 to do that. Okay.
And by the way, remember, it's four times — three or four times the time of my longest, my slowest TA. So hopefully, you'll have enough time. And they don't get it either.
So we'll make sure that they can do it in a reasonable time and then multiply that by a factor to give you that. So that's a good question Okay.
Keep going. I'm still here.
SSHing into exams to something like Val Grinder CGB. Yes, indeed.
If you want to use — if you want to SSH into the hive, if you want to code on the hive because the hive has all the software, please do.
We actually answered that on the Quest. SSH in. So practice that. Practice — okay.
Everybody, practice to make sure you know how to code in C, print stuff out, debug, use Val Grind, use Memcheck, all those things, and then practice coding in RISC V. If you haven't done that yet, make sure you do that.
Okay? And we're going to try to have a practice midterm that'll have a very simple question. Won't be the same difficulty. Maybe a question that'll be like, do something.
Do something. I think I'm suggesting to say, right Fibonacci in RISC V, which is something we do in like lab or homework or discussion.
Just say, right Fibonacci RISC V. And you'll be able to say, okay. Wow. Pretend this is an actual exam. Could I write Fibonacci in RISC V and print it and make sure it works?
If you can do that, you're probably 90% of the way there in terms of getting the RISC V stuff to work. And the C is the same way for the Quest.
Good question.
BORA NIKOLIC
Yeah. Just make sure that it's visible that we can see what you're using in your recording.
DAN GARCIA
Exactly. Exactly.
BORA NIKOLIC
For time lapsing, are we making videos for each question? Yes. Five videos.
DAN GARCIA
Five videos. Exactly right. Yep.
BORA NIKOLIC
Five sections. Not question. They're each section.
DAN GARCIA
Yeah. Each section. Sorry. We've got to make people hear about this.
And take a break between it. Right? Don't do it straight. A lot of people actually did this straight.
I don't think it's healthy. Take a section. Go take a drink, hit the restroom, have lunch. Then do another section.
Jump back in. Some of them are really short. The float is going to have two questions for the one section. Two.
I used to have four or five. Two! So I'm like — we're trying to remove the workload but still testing the important stuff. So we're going to try to have fewer things per there.
And so that'll be hopefully there. Venus is OK to use. So Venus, certainly, yes. Venus is okay to use.
That's how we encourage people to do it. You're gonna code RISC V, test it, debug it in Venus. We're going to have a whole structure that still isn't written yet.
We're going to try to make the whole structure there. So that's how you test. Good question. You're doing really well with good questions today.
Logisim questions.
I don't think, Bora, if you want to talk about whether we're going to ask them to use Logisim or not or what the plan is for that or whether they can use it.
What do you think? We're gonna talk about where Logisim fit into the piece of this. We don't want to reveal the question. But in terms of that piece of that tool.
How do you talk about that one? What do you think?
BORA NIKOLIC
I think we can say that we were thinking about using Logisim, but we are not going to make you use Logisim.
DAN GARCIA
Right. So there's no forced Logisim right now is the current plan. And I think that's not going to change. But I'll ask from the student's point of view.
Could they use Logisim to kind of help with that broader question?
BORA NIKOLIC
That's fine. That's fine. Go for it.
DAN GARCIA
Okay. Okay.
BORA NIKOLIC
Any simulator you want.
DAN GARCIA
Yeah. Okay. Good stuff. Perfect.
All right. So I answered live. Wonderful. I saw it.
Can you have two videos, one for the physical workspace, the other for the screen? That's perfect. Yeah. If you want to have — a lot of people did a screen video and a workspace.
That's great. That helps us a lot. So I can see. By the way, whatever you choose, I should be able to read what's on your screen.
So if you give me a video — and we were really soft on this in the Quest, but we're going to up the level — I have to people to see what's on your screen.
So if you can't — if you're going to try to Zoom call it, and it's all fuzzy, and I can't see it, or the contrast is bad, you got to practice that.
And make sure we can read your screen. This is — now this midterm is worth more than just — With Quest we didn't care as much. It's only worth the 5%.
We don't care. We'll trust you on that. But we've got to be able to see your screen. So if you need to do a time lapse, time lapse, I can see everybody's screen.
Everybody should time lapse, I can read their screen. I can read the smallest font. But lot of Zooms I couldn't read the screen.
We're going to say, you have to be able to give me a way for me to read the screen.
So if you're going to try to Zoom, if you want to do that, that's fine. But you got to make a screen movie, too.
I've got to be able to — whatever you do, I need to be able to see what's on your screen. That's the key. Okay?
So you can't just have a little window and doing something, and I can't see it. So practice this. Adjust your light. Adjust the backlight.
Play with your phones. Play with all those things to be able to get that. And by the way, so I love the idea of two videos to do that.
And we'll have a way for you to upload those two videos.
I would just zip — we'll probably have a one link and put them in a zip file and then put them there is one thing. That's a great question, Ryota.
Again, you're doing really well for questions today.
Let's see.
You don't submit the labs to the auto-grader. Correct. The labs, you could submit to the auto-grader to get feedback for yourself. But you don't have to.
That's the idea. You will get full credit for that lab for showing up and talking to our TA. And it can be all a technical conversation or a health check-in.
But you don't need to submit things in the future for all labs auto-grader. The auto-grader is there to help you, but not going to be something we'll grade.
Okay. The grades at the end of the day for the lab is not going to be part of the auto-grader.
You can mess them all up and never get them right, you'll still get credit for your lab. Okay? It's more about checking in.
BORA NIKOLIC
But use the auto-grader. Because it is — DAN GARCIA: Yes. Definitely use it. That's great feedback.
Great feedback. Yeah. I love that. Love that.
Let's see. I guess there was — from the chat, I didn't see the chat here. They're saying Discord isn't working partly because the TAs aren't there enough for things TAs need to help with.
I mean, the whole point of Discord is, we have our TAs to help with questions. When you finally get to the TA, ask whatever question you want to have of the TA.
But while you're waiting, use Discord effectively to ask all the questions of there.
If you take the sum of knowledge of how to do stuff, like, how to set up repo or do these things. With the students, it way exceeds what TAs know.
So students can help each other in ways that can scale with students. TAs are limited resource and they're overworked this semester. I'll be very honest. So there might be a long line.
But help each other in Discord. We're going to make a video as the best practice. We're gonna make a video of the worst practice and the best practice.
Worst practice is, you're putting something in the queue, and you sit quietly upset for 6 hours. That's the worst practice. Best practice. I'm in Discord.
Hey, folks, anybody else working on something? And you're chatting, and you're talking, and the TA might pull somebody off and have a one-on-one thing and come back. But you're chatting.
You're talking. You're having a conversation. So try to make Discord as productive as possible and as simulated of the face-to-face talking to each other in the hallway as we can.
That's the whole idea. And hopefully that video will help in terms of doing that. But thank you for that. And I get it that there are things that TAs can help with.
BORA NIKOLIC
I want to emphasize. We are way over our TA hours. So we are burning a lot of TA hours. So we can't throw in any more TA resources.
We are trying to free them up, as well.
I mean, the second purpose of this reduced workload, we are trying to reduce a bit the number of TA hours that we owe to our TAs. But it's there.
We are trying to streamline this as much as we can.
DAN GARCIA
Let's see.
Can I send you a time lapse workspace on a regular Zoom video? All those things. If you want to send multiple things, that's great.
At the very least, the bottom line is, I got to be able to see your screen and see your desktop. So it's not just, well, here's the screen only.
No, I need to see what your hands are doing.
So if you do a time lapse from the side, I can almost universally see the screen and desktop and hands and phone and all those things. So do what you need to do.
And if it means you need to submit multiple videos, that's kind of answered before. Please do that. But it's not required if I can get to see everything there.
So that's the key. So the more the better, to be honest. The more the better is what really we say there.
BORA NIKOLIC
Make sure I understand this one. Can we use the 15 minutes like a pseudo 15 minutes mandatory office hours for us?
Ask questions about project work, especially 15 minutes per student is definitely a huge TA investment. And office hour queue has been long.
I think it would be huge to use this like office hours to lower the strain on the office hour TAs. Yeah. You can use those for 10 minutes.
[INAUDIBLE] DAN GARCIA: Anyway you want to. Yeah.
BORA NIKOLIC
Yeah. Yeah.
DAN GARCIA
Anyway you want to. Now do know that when we have project office hours, the TAs practice. They read the project. Not every TA knows every project.
So we've got to be very honest. All the staff is going to be supporting in this office hours. But you may get a TA who doesn't the project as well as you do.
So they'll do their best and you'll teach them what it is. But it might not be as effective as if you went to project office. I'll just be very frank.
Our TAs who run project office hours, they are rock stars about that project. They've spent time reading it, doing it, answering questions about it.
But a random TA doing random office hours may not know what you need. So there's a little bit of like, your mileage may vary.
BORA NIKOLIC
So the question — if you have personal questions, come to Dan and me. We have private slots every week.
DAN GARCIA
I have one-on-one time every week.
BORA NIKOLIC
[INAUDIBLE] And if we need more, if we need more, they have not been sold out yet, we can add more. So we can talk to more people per week.
DAN GARCIA
Amen. Amen on that. More stuff. Will midterm parts be labeled?
Yes, they will. Which I mentioned that to my students, to my folks building this exam. We're gonna try to label them very clearly. Here's the topic.
See this right here? It'll be like, these are the five labels you're going to see for the five sections. The midterm sections. The word is sections, not questions.
Sections. The sections will all be labeled. Great question. All right.
Good stuff.
I had thought that we gave the OK to release the Quest, Bora, the Quest solutions. Now — BORA NIKOLIC: Yeah. I think we're waiting for the Quest retake to be done.
So — DAN GARCIA: That's right. We were waiting for the Quest retake. Because we don't do that. That was the idea.
And I think that now that that's done, I think we can release that and release some solutions.
I'm not sure if we'll release every test we use because maybe we want to use those questions again. But we're basically release — here's that TA solution that would have gotten full credit.
I think that would be useful to you, right? I don't mind doing a TA solution. That'd be fine.
So this is a good question. Of the two — we need to have a conversation — how many slip days would you be using to fix 2B?
I don't know how long it takes you to do that. And how when we say to do this. So it's a little bit — we need to talk about the details of this one.
BORA NIKOLIC
Yeah. We already have apparently a Slack thread going on that.
DAN GARCIA
Yeah.
BORA NIKOLIC
So we'll get back to you.
DAN GARCIA
Yeah. We'll get back to you. Again, I still like it. So we need to talk about being fair for other students.
BORA NIKOLIC
I like it. But we need to make sure that it's fair. So there'll be some penalty in [INAUDIBLE] we need to decide how many.
DAN GARCIA
Yeah. But that's a minor thing. If you saved all of them up, you've got a lot of them for that. That's great.
Project office hours this week. Nothing on the calendar. That's interesting. I think this most week was mostly going to be a setup for — if I look at the calendar, setup for the exam.
And you're right. I don't see any project office hours here. I see a lost section, which is great.
BORA NIKOLIC
Because we are adding project mid-term reviews. That's where we use some of those.
DAN GARCIA
So we burned, we moved to time over. Right. I did see that. Yeah.
I'm going to say, I don't know if we can. We can maybe ask our TAs.
But if we end up not adding any more, I would say go to all the office hours that you have to try to get support in the project space for that.
That's a little harder for that. But I appreciate it. It doesn't answer all your questions, but I'd do that.
And if you use all your slip days because you were not feeling well, then you have no slip days.
And so that's — whenever you have no slip days, you can't do the slip, which means you cannot slip on a project to add more time for that.
And you can't slip on doing the redo of 2B. So that's that space of that.
BORA NIKOLIC
We have added — just for reference, if you were sick and have a doctor's note, send it to us. That gets you points.
DAN GARCIA
We can give you extra. If you were actually formally sick, that like, I went to the hospital. I was doing a thing, and I was out, then we can actually do that.
And Bora, I don't know if you want how to manage that, via form or via email to you. I certainly don't any more email. [INTERPOSING VOICES] BORA NIKOLIC: We did a platform.
We did it via platform.
DAN GARCIA
Let's do the form for that. I don't want to — I can't do it as it is.
This is about the spec being clear.
I think we — there's a question about the spec. Try to update the spec — well, not the spec, but try to update the auto-grader because the spec wasn't clear.
And I think that already happened. I mean, we can bring that particular question back to the TAs. I wasn't in the lower level loop of that conversation. I just said, hey, guys, fix this.
You shouldn't be able to lose 40 points on a small error. And so I think that already happened. But let's make sure.
I'm just copy and paste this into my channel just to make sure that this one is resolved because I don't have the answer for that one very easy here.
Okay. I'll say, a student asked. So you want to cross that first person off, Bora? And we'll try to get back to you on that also next week.
We're here to answer these questions. We'll try to be here again next week. I actually don't mind staying and answering all the questions that are there. This is like our face-to-face time with everybody.
This is great. For those who decided to come to the section, which I thank you, the folks. how many people came. So ends that live.
Yeah. I think that's the same kind of thing. If you have [INAUDIBLE] chance if we fail one of the tests in a group, do all the test group take it off?
I don't think that's true. I think the way that it's supposed to work is — BORA NIKOLIC: No, no. That used to be like that. But we have changed that.
So Stephen managed to break this up.
DAN GARCIA
Perfect. Perfect. And thanks to Stephen for that hard work to do that.
BORA NIKOLIC
Yeah. It was a little bit rough of a cut, but it seems to be working.
DAN GARCIA
Yeah. Perfect. All right, folks. Thank you so much for showing up.
5:30. We're 30 minutes over. We're here for you, okay? Hopefully, that relief package helps a lot.
We'll see you guys next week. Thank you, folks.
BORA NIKOLIC
All right.
DAN GARCIA
End of course. Thanks, folks. Bye.
Đăng nhập để đánh giá học liệu này.
Đăng nhập để bình luận — chưa ai bình luận về học liệu này.
Không có khóa học liên quan.