YT YouTube2Text

YouTube2Text → Transcripts → [CS61C FA20] Weekly Lecture 01.LIVE - Great Ideas in Computer Architecture, Intro

[CS61C FA20] Weekly Lecture 01.LIVE - Great Ideas in Computer Architecture, Intro — Transcript

by CS 61C Departmental · 10,553 words · 1,738 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00recording on this oh i guess because it
0:01is a webinar it's a little bit different
0:02okay here we go
0:03and i'm starting it three two one
0:06ah it's live
0:09let's see what happens i almost feel
0:11like pulling this up on my phone to see
0:12like what would
0:13you know oh my gosh we have 93 a
0:21this is like the yeah we released the
0:24app how many people are downloading our
0:25app
0:26we're making money we're in the money
0:29this is great
0:31welcome everybody dropping in for 61c oh
0:34this is
0:35wonderful look at this number 612 you
0:37have no can you can everybody see the
0:39numbers too
0:40this is unbelievable i don't think they
0:42can see the numbers only they can't see
0:44numbers 680.
0:46no and they can't see the questions they
0:49i know they can't see the questions
0:50they're flying
0:51yeah yeah oh this is amazing great to
0:54see everybody
0:56hey folks right in the chat where you're
0:58from that'd be fun
1:00that'd be so fun to see all the places
1:03oh my god wow it's so fast
1:05it's like it's like twitch it's like a
1:08twitch channel
1:09it's like is amazing
1:13oh my goodness by the way that's the
1:15reason why we are
1:17not pushing them pushing the the
1:20the questions to the audience before
1:22they are like yeah you can't read it
1:24you can't read it yeah unbelievable
1:28this is great great to see you all holy
1:31moly
1:3361c this is amazing
1:37this is great what's the equivalent of
1:40having you know
1:41in the olden days you'd you'd ask your
1:42favorite students to sit in the front
1:44to make sure you have like your tas hey
1:45ta sit in the front so i get it that's
1:47that was historically back at mit to
1:48have all the taste in the front what's
1:50the equivalent of zoom like
1:51to make do something how do you get your
1:54special folks in i just love this is
1:56great
1:57camera's on actually can they what's
1:59that what's the view of students in the
2:00webinar can they see themselves they can
2:01only see us right no they can only see
2:03the panelists
2:04yeah okay so they can see all the tas
2:06there as well all right that's fine
2:07wonderful
2:08wonderful that's just great this is so
2:12fun i hope you all are relaxing steven
2:15just posted you share
2:16feel free to unmute stephen and share we
2:18did post the slides that pdf for the
2:21slides
2:22will be going so just push those to the
2:23website so it might take a second for
2:25the website to refresh
2:26and yep oh you want to see me all right
2:28that's good to see they only see me for
2:29now because i'm spotlight i guess
2:30that's yeah it's 4 10. ladies and
2:32gentlemen 410
2:34welcome to cs61c
2:37[Applause]
2:40so happy to have you here uh i'm
2:44teaching professor dan garcia
2:46my colleague i am
2:49professor boranie college and it's a
2:52great pleasure to
2:53see all of you around here this is so
2:56wonderful and we're going to break i
2:57hope we break a thousand we're at 8 25
2:59if you can't see it this is a delightful
3:02group i've got my ta's got a wonderful
3:03staff
3:04here uh that we're going to jump right
3:06in we've got a wonderful staff for
3:07supporting us
3:08thank you for being here and joining uh
3:11i'm looking here on the cypher q a okay
3:13wonderful let's jump in
3:15so this is the introduction slide this
3:17is attraction lecture 2621c
3:20great ideas in computer architecture
3:21otherwise known as machine
3:23structures that's rock and roll so
3:26here's the agenda for today
3:27we're thinking about machine structures
3:29we'll talk about the great ideas in
3:30computer architecture and then we'll
3:32talk about some red tape this is a very
3:34different world
3:35uh we'll talk about you know how we're
3:36moving this class online before i begin
3:38i do want to
3:39uh reflect on the fact that many folks
3:42around the country around the world
3:43i know we saw the list of where people
3:45were from are going through a hardship
3:47right now
3:48covensiting communities coverts
3:49certainly creating communities of color
3:51more than more than others we've got
3:53smoke in some communities we've got fire
3:55in some communities we've got
3:56uh hurricanes on the east coast so i do
3:59know people are going through different
4:00situations and i just my heart goes out
4:02to folks who are struggling for this
4:03we do have ways in this class if you
4:05find that
4:06something changes in your situation you
4:08need some help please reach out to us
4:10uh reach out to staff uh on piazza with
4:12private posts is that probably the
4:13cleanest way because we've got 1200
4:14students it's hard to handle all the
4:15email
4:16but piazza private post is what is that
4:17please folks
4:19uh reach out if you need some support if
4:21you need some extra delays so that we
4:22can work with you guys
4:23this is a semester we're definitely
4:24going to be lenient and flexible that
4:26way
4:27also i want to make sure you all know
4:28vote v-o-t-e please vote
4:31vote like your life depends on it folks
4:32there's another thing i want to make
4:33sure before we jump in
4:34all right here's the agenda let's go to
4:37the next one
4:39so you're taking this classics to nc
4:41we're so happy to see this is the by the
4:42way the largest 61c
4:44in the history of boards thinks it's
4:47actually
4:47bigger than just berkeley i'm going to
4:48say this is the biggest experience in
4:49the history of berkeley do you think
4:50it's bigger than that borah
4:52i am pretty sure this is the biggest
4:54offering of
4:5661c in berkeley but highly likely this
4:59is the largest computer architecture
5:01class ever taught anywhere ever
5:04taught anywhere ever taught anywhere so
5:08i mean this is incredible
5:09to be i'm just so happy to have my
5:11friend borah and i and this great staff
5:13of tas
5:14and tutors uh uh on our team and ais
5:17there's a lot of wonderful new people
5:19almost all of whom have gone through six
5:20to one three themselves so we've got a
5:21lot of people who just love this class
5:23i want to give back leslie do you think
5:25yeah good job
5:26jump in can you say two sentences about
5:28yourself you know how did you show up
5:30over here
5:32thank you for that wonderful and i'll
5:33ask you for the same thing so i
5:36uh was a grad student at berkeley and i
5:39fell in love with computer graphics did
5:41a phd computer graphics fell in love
5:42with teaching
5:43joined the faculty and then also when i
5:45was at mit i was
5:46my favorite class was the equivalent of
5:48of 61c was called 6004. 6004
5:51it was the digital logic class and i
5:53loved it so i said
5:54i can't believe i'm at berkeley getting
5:56to teach the class that i fell in love
5:57with when i was at undergrad i actually
5:58got two majors i got a double e major
6:00and a cs major at mit
6:01and so to teach this class actually does
6:03make that alignment what brings you to
6:05this class borah
6:06oh i i grew up in a country that does
6:09not exist anymore it used to be called
6:10yugoslavia now is serbia i
6:12grew up in belgrade serbia and i came
6:15you know
6:16actually to do a research project in the
6:18us at uc davis and i
6:20you know that was supposed to be like
6:21five months research project i
6:23it has been 24 years since
6:27you got bit by the bug i love it and
6:30yeah i i've been at berkeley for 21
6:33years
6:34teaching all kinds of things this is the
6:36second time that i'm doing 61c with dan
6:39and although i am probably not as
6:42expressive as that
6:43i'm as enthusiastic about the subject
6:46matter as he is
6:48i am i'm so delighted bored to be
6:50partnered with you again this is such a
6:51fun we had such a great time
6:53two years ago and i learned so much from
6:55you and i hope to continue to learn from
6:56you you are such a resource
6:57for us for the class so we're really
6:59lucky to have you on team thank you for
7:00that
7:01so 621c isn't about c programming if you
7:03came here thinking
7:0461b is about learning java you're wrong
7:061562 is about learning python you're
7:07wrong learning computer science here
7:09this is about understanding
7:10the hardware software interface what you
7:12need to know as a programmer
7:13when you're actually going into the to
7:15going into your career at the lowest
7:17level
7:18uh how does the whole thing work how do
7:19you optimize your car do you make it
7:20fast how do you make it parallel all
7:22those things you'll learn in this class
7:23in a way
7:23that really was hidden from you from an
7:25abstraction point of view in 621b and
7:27621a
7:28so we'll talk about hardware you'll
7:30learn how the entire thing works from
7:31the top to the bottom this is the first
7:33class well actually you'll know from the
7:34all the way at the top the top the
7:36highest most abstraction layer all the
7:37way down to current flowing
7:39and electrons filling holes all the way
7:41down in hardware this is the only class
7:42that makes a connection that's why it's
7:43so special and i'm happy to be
7:45uh leading that code lead in the group
7:47to make sure you understand that so it's
7:49really great
7:49you'll also understand how to peek below
7:51the hood and say okay this is the amount
7:52of hardware i got okay
7:53yep that that that great here's i'm
7:55gonna adjust from the software point of
7:56view to make this hard make this program
7:57screams you gotta learn how to make this
7:59stuff work in parallel in a way and
8:00that's again thanks to a lot of folks
8:02who've taken this course before me
8:03so i want to thank thank nick weaver who
8:05contributes a lot of the slides we're
8:06going to have
8:06i want to thank a lot of folks who have
8:08this uh dave patterson
8:10kirsty osanovich john morzinek a lot of
8:12those wonderful randy cats a lot of our
8:14kuby a lot of our systems faculty
8:16contribute to where this course is so
8:18again i want to honor them
8:19we're standing on the shoulders of
8:20giants we've got an amazing systems
8:21faculty and architecture faculty
8:23uh and doubly faculty in berkeley this
8:26is a
8:27this is a picture of old school 621c so
8:30we made a transformation of 621c at some
8:32at some stage and this is like this is
8:35old school 621c and by the way this
8:36photo is fake
8:38and the reason we showed this is because
8:40we only focus on a single computer
8:42a single computer like that's all it is
8:44as if a single computer with a single
8:46core like
8:46what how are you still thinking about
8:48computer science and architecture
8:50and systems as a single computer with a
8:52single core crazy this
8:54is new school 621c
8:57where you have more power in your hand
9:00and your watch
9:01and in your car and on your drone than
9:03people
9:04took to go to the moon years ago right
9:06it's unbelievable what you have at
9:08access
9:09and the model by the way is that
9:12the front end the interface is now a
9:14personal mobile device it's a car
9:16it's a it's a it's a drone that's the
9:17front end that's the usable and that's
9:19if you touch
9:20and see but actually that's the back end
9:25warehouse scale computing is where all
9:27of the
9:28stuff is happening you type something on
9:29google search it's a little teeny line
9:31in your window you think i said
9:32no all of a sudden something fires up on
9:35some data center
9:36somewhere in another country maybe and
9:38then feed you the answer you know
9:39whether you'd even know you say
9:40siri you say alexa all that stuff is
9:43being driven
9:44by these warehouse scale scale
9:47uh computer systems warehouse scale
9:49computing
9:50one of the things you'll notice in this
9:51picture there's there's wind farms
9:54there's water we'll talk about all the
9:55details of how to actually make this
9:56work with a whole lecture on warehouse
9:58scale computing just a little taste of
9:59that
10:00this is what happens you look inside
10:01look how cool that is it's like the
10:02future the future is here
10:04folks unbelievable amount of energy and
10:06amount of engineering to make this work
10:08um so much work to get the coolant
10:11to get this cool temperature is going to
10:13be a big
10:14theme here so to keep this cool and keep
10:16it the right temperature regulated
10:17there's a lot of work there's almost
10:19arguably more work there than it is in
10:21getting the bits to go around and
10:23do the right thing it's really
10:24interesting so this is the old school
10:26model and
10:27this is a class i've taught this class
10:28since 2004 and i happen to work with
10:30dave patterson in the first couple of
10:31years
10:32i did this i learned so much from i've
10:33learned so much from all my
10:34co-instructors i love
10:36co-teaching courses so this 61c goes
10:38from the top level
10:40through c through a compiler through
10:42assembler to machine code through the
10:43instruction set architecture
10:45big abstraction we make that all these
10:47are layers of abstraction although all
10:48the horizontal lines but we called this
10:50one
10:50really big to say this is a big one
10:52that's the software hardware boundary
10:54then the way we attack 621c is it's
10:57almost like a byte the way you take it
10:58in and out by
10:59it from the top down and now you hit the
11:00middle layer and now we're from the
11:02bottom up
11:02so now we're going to actually show you
11:04current and transistors and how to make
11:06a transistor and how what you can build
11:07out of transistors how to build basic
11:09circuits how to make visual digitals
11:10how to design these things how to then
11:12take those those big block
11:14elements and then make a data path and
11:15control that makes a computer work and
11:17the computer you're going to make work
11:18is a computer that will interpret the
11:20machine code you compiled and assembled
11:22down to
11:23so it's like the fingers will touch in a
11:25beautiful way this is an amazing
11:27this class is is eeks actually this
11:29class connects them in a way that no
11:31other class does
11:32and i'm so pleased to be part of that so
11:33it's great
11:35so this is the new school
11:38machine structures in which your
11:41hardware is on the right
11:43you've got parallelism at so many levels
11:46you've got parallelism parallel gates
11:48that's the hardware description but from
11:49bottom up
11:49i'll move my my body here hardware
11:51description all gates were comparable
11:52the same time
11:53now you have parallel data where you
11:55might chunk on more than one data unit
11:57at the same time
11:59then you have parallel instructions
12:00where you might actually be able to
12:02borrow we'll talk about this in a second
12:03have up to five instructions you could
12:05even have more than but what's a teacher
12:07machine that has five instructions at
12:08the same time
12:09to have parallel threads very little
12:10higher level software now you parallel
12:12threads you've probably heard of threads
12:13before
12:14now you've got multiple cores these
12:15threads might get assigned to the
12:16hardware course and you'll run with that
12:18then you might have parallel requests
12:19where you do a higher level where you
12:21actually send off
12:22different programs doing things and that
12:23might be parallel quests to do something
12:25and you might have multiple machines
12:26wake up
12:26multiple machines are waking up serving
12:28that request all of that is driven by a
12:30hardware and all that's driven by
12:31so if you see the top picture here
12:33you'll see that that's the
12:35the inner workings of a data center and
12:37all of those are based on a computer
12:38which might have multiple cores how does
12:40that work whether execution units
12:41functional blocks and logic gates all
12:43that work
12:43on the hardware side pretty cool stuff
12:46i'm going to pass it over to my
12:47colleague board who's going to talk
12:48about the great ideas in computer
12:49architecture
12:50go thank you then let's see if
12:53oh look at this they can control dan's
12:57uh screen and hand slides we picked
13:01uh a a selection of six ideas
13:04that are great ideas but but there is
13:07many
13:08there are many other ideas in computer
13:09architecture that are great as well
13:12um here is a list its abstraction
13:16how we abstract layers of computing
13:19and represent them such that the other
13:22layers don't have to think about
13:23gory details underneath them the second
13:26idea
13:27is the moore's law
13:30and the principle of scaling the third
13:33one is principle of locality
13:35and memory hierarchy let me try to move
13:37myself uh
13:38this then crash this is great um the
13:41fourth one is parallelism the fifth one
13:43is performance measurement
13:45and uh this feedback of the you know of
13:47improvements based on the measurements
13:48and dependability
13:50and redundancy the first one
13:54dan already touched on this is slightly
13:55different picture of the same thing
13:57it's the idea that we would like to
14:00abstract every layer
14:01as we go through them such that we don't
14:04have to
14:04think about how do gates switch every
14:07time we execute a new c
14:08command or a c instruction so
14:12the idea here is to maintain all these
14:16layers of abstraction that
14:20that pass relevant things up to the
14:22upper layer while obfuscating things
14:24that we really don't care about for
14:26example
14:27how the wires work inside a
14:29microprocessor
14:31second great idea is the moore's law
14:35moore made this project prediction long
14:38long time ago in 1965 based on
14:41really if you look at this slide there
14:42are only uh
14:45four data points oh on that way uh
14:48on that slide five data points and he
14:51noticed that over the first
14:53five years of semiconductor industry the
14:56number of
14:57components per chip essentially
14:59components or transistors doubled in
15:01each technology generation
15:03and he made the bulk prediction that
15:06that is going to continue
15:08for the next 10 years well
15:11um he also as a side note he was a cow
15:14alumnus
15:15and graduated not from
15:18ecs he graduated from chemistry in 1950
15:21he went on to
15:22fund intel corporation after that but
15:25look at that what happened with moore's
15:27law over time
15:28it didn't hold up only for 10 years it
15:30held up
15:31for five decades after that and it is
15:34still
15:36continuing although it is slowing down
15:38you have seen a lot of that
15:40in the news lately we'll touch on that
15:42in a second
15:44and there is a corollary over there well
15:46not directly
15:48related to moore's law this is the a's
15:51law
15:52this is showing statistics in the u.s
15:54what is happening with the grade
15:55inflation
15:56basically showing that the number of
15:59ways is increasing
16:00in every year with a small trend
16:04um well if you continue doing this going
16:07this way
16:08by 2070 in classes like 61c and
16:11everything else everybody will have
16:13ace that's great news your grandchildren
16:17will all get a's by the way i mean we
16:19can we are okay to give everybody
16:22an a this year um you know help us out
16:24with that we would find out
16:26uh experimentally what does the college
16:28of engineering
16:29or lyrics and science do to us if we do
16:31that but we would
16:32love to go through that experiment with
16:34you i think they'll celebrate us if we
16:36get to some stage where mastery learning
16:37happens and we take them all across the
16:40finish line i think that i think they'd
16:41love it i think they'd love it i think
16:43rather than being vilified i think it
16:44will be celebrated how they'll say how
16:46did you do it how did you do it tell us
16:47how you
16:48you know rather than just it for free if
16:49they all got a level material if you
16:51have to if they all do well in their
16:52final
16:53we give a same final the same difficulty
16:54every year if all the class gets above
16:56something and
16:57all the grades above we give all these i
16:59love it let's do it let's make it happen
17:00the first year ever
17:02know wouldn't that be wonderful to break
17:04every goal in
17:06in this college about the grading uh dan
17:08we'll talk a little bit about that
17:09towards the end but help us out we are
17:11we're open this is a
17:12weird semester let's make it even
17:14weirder
17:17the next thing next principle is based
17:19on this you know beautiful picture that
17:21jim gray um uh touring award winner um
17:24during award is like a nobel prize in
17:26computer science and uh
17:28who jim gray uh got the bachelors from
17:31cal and
17:32a phd i think it was the first phd in cs
17:35in cal
17:35in 1969 he made this analogy of
17:38trying to visualize how far away
17:42is the data and how long that how much
17:45how long does it take us to retrieve the
17:46data from different parts of our
17:48computing system
17:49so um on the left hand side what we see
17:51is uh
17:53delay or a latency of how does long does
17:55it take us to get the data from a
17:57different place in a compute system
17:59uh in nanoseconds from one to
18:02ten to the sixth and the ninth and then
18:04on the right hand side
18:05we have analogy of how would it look
18:08like
18:09to retrieve a piece of paper or a thing
18:12from
18:12somewhere around us so getting something
18:15from the registers that are on the
18:17microprocessor chip
18:18um takes approximately another second on
18:21the right hand side if that is an
18:22equivalent
18:23of a room or over under uh um
18:26if it takes me to retrieve something
18:28from my thinking about heart about
18:29something
18:30it takes me to get something uh from
18:32this room take me would take me a minute
18:34if we well i mean when we get back on
18:36the campus if we need to get something
18:38forgot i don't know um a handout
18:42on the other end of the campus and went
18:43to the class it would have taken us 10
18:46minutes
18:47um if we left it somehow in sacramento
18:49would have taken us
18:50hour and a half sacramento is an
18:53equivalent of going to the memory going
18:56from the microprocessor chip to the dram
18:58chip which is a separate chip on a on a
19:00computer
19:01motherboard if we need to go to a disk
19:04on
19:04to a hard drive or even a solid-state
19:06drive that
19:07is like going to another planet going to
19:10a magnetic this is like going to pluto
19:12or going to a solid state disk is
19:13you know kind of going to mars not a big
19:16difference
19:17from that standpoint and then finally if
19:20you need to
19:21[Music]
19:23somehow get to the
19:27to to to a tape or
19:30an optical uh robot that is doing
19:32backups
19:34well that's like kind of going to
19:35another galaxy
19:37all right so that's what is important
19:39order
19:41what is important there is understanding
19:44this
19:45principle of locality and memory
19:47hierarchy
19:49if we look at this if you look at this
19:53what we have is this principle of
19:56locality where
19:58there is a lot extremely fast
20:01memory that is available on a
20:02microprocessor chip that is extremely
20:04expensive and has tiny capacity
20:07and as we go out away into the
20:10the caches into the physical memory on
20:13the board which is
20:14dram and down to solid-state drives and
20:18hard disk drives
20:20in every layer the amount of memory
20:22there
20:24gets a by about an order of magnitude
20:28bigger by about an order of magnitude
20:30cheaper
20:31and about an order of magnitude slower
20:34the
20:34what we'll see these compute systems are
20:37built what we're going to be
20:38doing with building these uh these
20:40compute systems as we go through them
20:42to make user see memory as infinitely
20:47big infinitely fast
20:50the third abstraction abstraction just
20:53got to say it over and over and over the
20:54most important idea of this class is
20:56abstraction that whole slide was about
20:57abstraction
20:58you live at the speed of the small fast
21:00guy but you actually get the abstract
21:02view with the sizes of the bigger guy i
21:03love it thank you for it perfect
21:05thank you dan yeah then there the fourth
21:09idea is that we'll be drilling in
21:11um through various views of abstraction
21:15and else is the the view of parallelism
21:18dan mentioned
21:19uh parallelism a bit already that many
21:22things in compute systems are
21:23happening concurrently so this is the
21:25first one uh we're gonna drill into this
21:27this is a complicated picture showing
21:29execution of a pipeline processor
21:31where three instructions are
21:33concurrently being
21:34executed on a processor in time slot
21:36three and
21:38we don't need to understand that now we
21:40are going to go into a glory detail of
21:41that
21:42there are different ways how this
21:43concurrency can happen um
21:45another one is this idea of thread level
21:48parallelism
21:49for example you know we can utilize we
21:51can run multiple software threads
21:53through um our compute hardware through
21:55a computer
21:56um even if we do not have multiple
22:00execution units for example you know
22:02what where does that come from
22:04while we are busy going to for example
22:06to
22:07andromeda to get our next batch of data
22:10we can run
22:11a complete program through it or
22:13multiples of them
22:14that's that's the principle of running
22:16multiple threads on the microprocessor
22:19you know the third way how we can see
22:21this parallelism is
22:22you know this is an old illustration
22:24from the times and people still use
22:26paper folders so if you have
22:29a bunch of folders that showed up on
22:31somebody's desk that need to be
22:33urgently processed you give them to
22:35multiple people
22:37and these multiple people and ben and
22:40chen
22:41can sort them in parallel if it is the
22:44same algorithm that they need to to
22:46perform
22:47this is a principle of data
22:50level parallelism there is a caveat in
22:53there
22:54that a stands behind is a principle it's
22:58it's the
22:59basic message of andol's law handle says
23:01this that
23:02the longest pole on this tent will
23:04always stick out
23:06so if we can if we have a large amount
23:11of parallel code for example in our
23:13program
23:14say you know 50 of parallel code
23:18that means that the other 50 percent of
23:21a code is serial code
23:22if we try to parallelize to infinity
23:26this uh
23:27this parallel code that means that the
23:30total speed up
23:31is only going to be up to 50 percent
23:33because
23:34now the serial code is going to be the
23:36bottleneck
23:37we'll be seeing this over and over in in
23:40compute systems
23:41the principle of the longest ball we
23:43will always work
23:45on squishing something that we can
23:47accelerate and make fast but then
23:50we have to worry about the other things
23:52that pop up
23:55now the fifth idea is the idea of
23:58performance measurement and improvement
24:01that is one of the things that
24:03dave patterson our emeritus
24:06has
24:12spent
24:17spent a lot of time in his early career
24:19in trying to quantify
24:21the metrics of compute systems so the
24:24idea here is we would always like
24:26to match the application hardware
24:30to match the application to the
24:32underlying hardware to do things
24:34that uh that we care about so we would
24:37like to exploit
24:38locality would like to exploit
24:39parallelism we would like to
24:41uh build some special hardware features
24:43but this is something incredibly
24:45important
24:46going forward from now on now important
24:49thing about performance measurements is
24:50we do want to metricize what we care
24:53about and that is often latency entropy
24:56so when we talk about latency that is
24:59how long does it take to complete the
25:01task
25:02what is the triplet how many tasks can
25:04we complete
25:05in a given amount of time
25:09but things that matter over there are
25:11also how long does it take us
25:13for these highly parallel tasks how does
25:15it take us to wind up how does long does
25:17it take us to fill up the cues
25:18and then how quickly we run through the
25:20full cues
25:21and remember latency is all about the
25:23time to finish
25:25the tasks the sixth
25:29important idea is about dependability
25:32and redundancy
25:33the picture over here is something that
25:36is going to get us
25:37even if we take care of everything else
25:40this is a picture of a cloud chamber uh
25:42we had a uh a video of a cloud chain but
25:44now we can't run it
25:46through this kind of a zoom application
25:48that we are running
25:49the cloud chamber basically records uh
25:52hits of
25:53um cosmic rays onto the earth into this
25:56so we see some ionization happening
25:58there these little traces that we see on
26:00the
26:01on the slide are essentially charged
26:03particles that are hitting
26:05a chamber over there they
26:08hit us all the time they hit our chips
26:12so even if we build everything redundant
26:14these things are going to rain on us and
26:17they've been raining
26:18uh since the big bang on everything that
26:21is out there in the space
26:23they have a tendency or ability to flip
26:26bits
26:26and bits you know that used to be zeros
26:29may become ones and those that have
26:31ones may become zeros and that's not
26:33going to look good on for example on
26:35somebody's bank account
26:36especially if it if one gets flipped to
26:39a zero
26:40so many of these compute systems need to
26:43be made
26:44very dependable in that way
26:48this is now really popular topic when
26:50you talk about self-driving
26:51because there's certain standards but
26:53that these systems need to meet there
26:56one of the basic principles there is the
26:59redundancy and that principle of
27:02redundancy
27:03says that a failing
27:07piece doesn't mean that the whole system
27:10will fail
27:11in this case you know if we ask three
27:13processing elements what is one plus one
27:16and if two of them think that it is the
27:18answer is two one of them thinks it's
27:20one
27:21well uh we can vote what is the right
27:23answer even if it did not know what is
27:25one plus one
27:26and we can use this majority vote to
27:30determine what is the answer
27:32this is a one of the basic principles
27:34how compute systems have added
27:36uh reliability to them over time
27:40one thing uh consequence of mars laws is
27:42that we have
27:43very many transistors nowadays um one
27:46thing
27:47that we do with that is we can use them
27:49to do redundant computation
27:52here are a few examples of how does this
27:54is being done
27:55one uh thing that many someone
27:59did not know that there are redundant
28:02data centers in the world
28:04dan has picture uh shown a picture of a
28:07data center out there
28:10and amazon runs data centers in
28:13different parts of the world if there is
28:15a power outage
28:16amazon doesn't go dark
28:19you if there is a storm if there if
28:22there is
28:23a fire amazon doesn't go dark
28:26the next thing you might have heard
28:28there are redundant
28:30disks and raid systems which basically
28:33means redundant arrays of independent
28:35disks
28:36if you basically have a system of five
28:38disks
28:39it allows you for one disk to fail and
28:42you continue working like all of them
28:44are there and
28:46another example on the bottom of this uh
28:48is a dim uh which is a
28:50dram device that gets plugged into a
28:52computer
28:53um it is there to represent eight bit
28:56words but look there are nine chips one
28:58of them
28:58is redundant i should point out that our
29:01colleague randy katz was one of the
29:03leads of the raid project so
29:04a lot of the things that you're learning
29:06are invented at berkeley
29:08in fact we're going to see in two slides
29:10how people have been honored for things
29:11invented
29:12as well yeah we are a little bit uh you
29:16know we do tend to
29:17to um invent a lot of things and um
29:21reach to take credits for those now
29:24a couple of slides here um is
29:28something that is very interesting here
29:31um the computer architecture has become
29:34the most exciting thing um out there in
29:38the computing
29:39at least in my opinion but i'm not the
29:40only one aforementioned
29:42dave patterson claimed that this is the
29:44golden age of computer architecture
29:47why is that so well uh compute systems
29:50went through the waves and these waves
29:53you know we have seen
29:54three waves so far the wave of
29:56mainframes that was
29:57long time you know like when dan and i
29:59were little kids
30:01um there were main frames that were used
30:03for scientific computing
30:04and big data processing made by ibm and
30:07they would take a whole room
30:08there was a revolution of compute of
30:10personal computers
30:12that came in the 90s and exploded with
30:16the development of the world wide web
30:18in the 90s and was replaced by the era
30:21of smartphones in the killer app there
30:23was the app itself
30:25we are on the tail end of that wave
30:29it is being replaced by a new wave that
30:31is very amorphous and we don't know
30:33exactly how it looks like it there is a
30:35whole bunch of other things that are
30:36that are important there and they're
30:38happening
30:40that involve healthcare and telemedicine
30:43and education autonomous driving and
30:44gaming entertainment and vr and ar
30:47and all kinds of other things
30:50that require in order to be efficient
30:53require
30:54to be implemented specialized systems
30:58now because of moore's law
31:01slowing down this becomes an
31:04architectural issue
31:06what we find out machine learning is a
31:08common for most domains nowadays
31:11but it is not the killer app the killer
31:13app is to make the car
31:14drive use the machine learning to make
31:16that happen
31:18uh this is another chip that show
31:20another slide that shows
31:22how things have changed over time we
31:25have
31:26gone through a lot of moore's law um
31:29five decades of mars law
31:30and now if you look at the latest points
31:32on this chart and the top curve that is
31:34showing the number of transistors
31:36um essentially doubling every
31:3918 months or every two years
31:42depends on which part of a slope are we
31:44in
31:46it is slowing down it is still you know
31:48we are still getting more transistors
31:50but they are not
31:51coming that easily easily and that
31:54abundantly as they did before
31:58so the other thing power has been
32:00limiting many of the chips
32:03so in order to limit the power increase
32:05otherwise we will be running low
32:07furnaces in front of us
32:09the frequency had to be limited single
32:12thread performance was tied to the
32:14frequency
32:15in order to get better performance we
32:17went parallel so we saw the number of
32:19cores increasing
32:20but what has been observed now there is
32:23a limited amount of
32:24parallelism that is out there because of
32:26amdahl's law
32:28so what do we do now to get more compute
32:32performance
32:33that's the idea of specialization that
32:36is why we are seeing
32:37you know a lot of development of these
32:39heterogeneous
32:40differentiated systems for different
32:42compute domains
32:44an example here is
32:47apple a13 bionic that has you know
32:51multiple microprocessor cores on it
32:53and then has specialized units like gpus
32:56that will find graphics processing units
32:58and now has a
32:59neural processing unit and a whole bunch
33:01of things that are unlabeled here
33:03um apple does have an event tomorrow
33:05they
33:06are interested in hiring a lot of people
33:08because they want to capture
33:10a lot of domains out there so check it
33:12out
33:14they may tell you something about cs61c
33:16and how much they care about that class
33:20now trying to wrap these ideas up where
33:22where are we now and where are we going
33:24to
33:26all commercial wisdom involved
33:29basically this moore's law giving us a
33:31free ride
33:33every year or every other year we would
33:36get a better computer
33:38so no matter how bad our software was it
33:40would just
33:42naturally run faster the year after so
33:45some people say that that was the uh
33:48that led to the education or development
33:52of lazy programmers
33:54people who just basically you know it
33:56did not make sense to
33:58work a lot in your application that
33:59makes sense to develop custom chips
34:01because it
34:02um you know you can just wait for a year
34:05while you're instead of you're sweating
34:07and developing something new
34:08a new custom chip well your general
34:11purpose processor is just going to get
34:12better and everything is going to get
34:14to to run faster the next year but with
34:16moore's law slowing down
34:18this is not the case anymore so we need
34:21to build
34:22specialized computers this is
34:25uh google's tpu version 3
34:28which was publicly released only last
34:31week
34:32interestingly that was at hardships
34:34conference it was showing
34:36the details of what you have there so
34:37this is showing a top view what you see
34:39on the top is a top view of a tpu3 board
34:43that has three uh that has four
34:46liquid cooled chips uh
34:49that thing when put in the racks that
34:52can deliver
34:53over a 100 beta floating point
34:56operations per second
34:58that's a very large number that's
35:00something that was completely
35:01unthinkable
35:02and will not be reached by general
35:04purpose computers for a long time
35:07you need a specialist computer as a
35:09specialist computer for that
35:11now to wrap up what dan has started
35:14talking about
35:16uh dave patterson and john henderson two
35:18years ago
35:19won the touring award the the biggest
35:22award that exists in
35:23in computer science for
35:26essentially developing many of these
35:28principles and writing them in the
35:30textbook that we are using
35:32um that you're covering in this class so
35:34with that
35:35i'm gonna hand it over back to dan
35:38[Music]
35:39wonderful thank you so much for thank
35:41you so much the
35:42the i mean we one of the things that
35:44patterson and hennessey did was they
35:46they went against conventional wisdom if
35:49the world said this thing
35:50the world said we have to have complex
35:52instruction sets with every
35:53we'll talk to that means and they said
35:55no no i want to reduce instructions have
35:57much more simpler instructions the
35:58hardware is easier to build
35:59they said ah they would laugh at they
36:01were laughed at
36:02at conferences they'd be like they would
36:04be like a panel and they'd be on one
36:05side and the
36:06the sisk people the complex would be on
36:07the other side and everyone was on the
36:09other side
36:09but they held their own and they were
36:11right all along we go back you know they
36:13were right it's like
36:14some people were right all along so this
36:16is that that that was certainly well
36:18deserved and certainly honors
36:20some of their foresight that they had
36:21with regard to the path of this
36:23let's talk about what you need let's
36:24talk about some red tape so
36:26this is i'm a huge star wars fan huge
36:29yoda fan
36:29baby yoda can't wait till the next
36:31season comes out uh and so this is my
36:33the worst yoda imitation ever but it's
36:36always emotion the future is so that is
36:39the idea that
36:40who knows what's going to happen who
36:42knows when storms and smoke and all
36:44these things
36:45our schedule may change slightly
36:47depending on factors and this includes
36:48everything
36:49lectures assignments and labs but we'll
36:51try to keep it as much as we have but it
36:53will also let you know as soon as we
36:54know something's going to change we've
36:55got that going and we're also trying a
36:56lot of new things
36:57and we say well we promised this but
36:58then we had to back it off a little bit
37:00because
37:00things might go down the server might go
37:02down if we need to practice on this and
37:03we need to
37:04hopefully get a stable system for you
37:05and we know it's not stable pull it off
37:07and
37:07go somewhere else so overall this is
37:10some course information just some basic
37:11red tape here the course is called 61c
37:14the website is called 621c.org thank you
37:16steven for doing an amazing job of
37:17making it
37:18look really really usable and accessible
37:20and dark mode and it's awesome so
37:22some wonderful web designers are part of
37:23that team you know born our instructors
37:25please call us dan and borah don't call
37:27us
37:27don't call us professor or something or
37:28something it's dan and borah never
37:31never yeah that's that's why mom calls
37:33me professor
37:35anybody calls me dan i'll even correct
37:37you if you call me anything other than
37:38that so call us that
37:39um we have an amazing group of teaching
37:41assistants and i was gonna
37:43get out of this but i'm stuck this was
37:44the c webpage was when i was gonna get
37:46up and show the thing
37:47there's a wonderful web page i wanna
37:49maybe stephen can someone add the link
37:50to the
37:51list of tas i can encourage people to
37:53look at the list of tas we have a
37:54wonderful staff
37:55tas tutors ais we're going to add as
37:58many pictures as we can to that
37:59a lot of people are here as part of our
38:00panelists to support us amazing group of
38:03folks i want to thank them all for their
38:05service to the department service to
38:06this course
38:07we would not be able to teach this
38:08course without them this is the two of
38:10us and there's 1200 of you
38:12i've got 911 participants here this is
38:14the biggest course i've ever taught
38:15uh ever and i we have an amazing staff
38:18this year we worked up really hard
38:19uh you know the head teams can know we
38:21worked really hard to pick the best the
38:23best the best folks to be part of this
38:24based on experience and passion and we
38:27we think we got we got all of them so
38:29we're excited about that uh just one
38:32thing to add
38:33a complete teaching crew whole teaching
38:35crew has approximately 100 people
38:37just think about that it's a typical
38:40computer architecture class
38:44yeah managing a small army it's
38:46unbelievable
38:48uh in fact we were so big that last year
38:49we had there was no room on campus that
38:51would hold us in fact we're already too
38:52big for any room on campus because we're
38:53my campus is 800 and wheeler
38:55already too big for the biggest room on
38:56campus so we actually allowed this the
38:58first time we ever have everybody in the
38:59first lecture it's pretty exciting
39:01so nine nine ten nine eleven and
39:02hopefully what more people will be able
39:04to come into other sessions
39:05our textbook we have three textbooks
39:06patterson and hennessy we showed you the
39:08picture before
39:09cod computer organization and design
39:11make sure you get the risk five edition
39:13there are other editions there and you
39:14can get some cheaper there uh but make
39:16sure you get the risk five edition
39:18k r this is the classic every c
39:20programmer
39:21ever i'm gonna say ever has the k n r
39:23second edition on their shelf
39:25there's a big cyan c on the front page
39:27make sure you get that one not the older
39:28one
39:29but ralph stonehalsey i'm not sure if
39:30i'm pronouncing that right uh it has the
39:32data center as a computer and there's a
39:33new third edition so we're going to have
39:35the new version up there i think we
39:36might have the old ones still linked but
39:37we'll get the new one
39:38we'll leave this like week 12 you don't
39:39need this until now piazza is where all
39:41the action happens that our community is
39:43getting together our 1200 people
39:44happening on piazza every announcement
39:46discussion clarification happens there
39:47even
39:48pinned threaded posts where you can
39:49respect we've really our teams have
39:51thought about how to actually manage
39:52piazza really wanted a great job with
39:53that
39:54let me do kind of an old school new
39:56school this is what we used to do okay
39:58many of you i'm not i don't think
39:59anybody here is a freshman but if you
40:01are welcome
40:01maybe i'll take 61c your first time
40:03freshman year that's impressive maybe
40:04that's one or two of you in the group
40:06um i think there was somebody years ago
40:08took six one a b and c first semester i
40:09think that
40:10actually happened once and they got all
40:12a's it blew my mind um
40:14this is the old school in the left
40:16column the new schools on the right
40:17um so lecture obviously monday wednesday
40:19friday for an hour if you miss it you
40:21can watch a video that and many people
40:23by the way over time decided
40:24you know what i'll just watch a video
40:26and so you know 62 a used to have 1200
40:28now it's over 2000
40:30and they tried in one classroom and they
40:31said you know it doesn't matter the
40:32classroom only holds 800
40:33because not everybody's coming anymore
40:35and it'd be 200 and then 150 and then
40:37there's the stalwart hundred
40:38and they're all spread out nobody even
40:39sits in the front they all sit spread
40:41out
40:41even before you had to six foot
40:42distancing um so we used to be able to
40:44fit in
40:45into wheeler nevertheless day but in
40:47general but anyway the point is we had
40:49that the new model is
40:51borah and i want to bring community
40:53together we want to share what the week
40:55is like we want to talk about computing
40:56in the news we want to talk about
40:58how to set you up for success you know
41:00give you little announcements like
41:01announcements we want to do that we want
41:03to be able to share the john de niro
41:04announcements line so
41:05all that happens during these live
41:07lectures they're going to happen
41:09one time a week usually it's monday if
41:12monday's a holiday
41:13it's going to be wednesday okay so we
41:15want to make sure just to bring the
41:16community together
41:17there's our tas will be here our staff
41:19is here to ask any questions this is a
41:20chance for us to get us live live action
41:22okay that's lecture we all
41:25have recorded and we're in the process
41:27of recording all of our content lectures
41:30into little snippets into little like 8
41:3210
41:3312 maybe if it's a little this one has
41:35longer 20 minute snippets
41:36after each one of those is going to be a
41:38clicker question those clicker questions
41:40are
41:40credit you you will try it if you get it
41:43wrong there's no penalty just keep
41:45trying it to get it right it's like a
41:46lock you have to get it all right until
41:47you get it right
41:48once you get it right you get credit for
41:50watching the video okay so that's the
41:51idea
41:52these click questions are kind of a way
41:53to force you to do this and you're going
41:54to have a couple days after lecture
41:56to do that every snippet is you know
41:58roughly as i said 4 to 20 minutes or so
42:00so you watch this and then have a have a
42:02question
42:04lab the way lab used to work was you
42:06know we had two hour lab you're done in
42:08pairs typically
42:09tas and ai's circling uh to answer any
42:11questions you get a check off at the end
42:13and those are also
42:14meant to be done for credit meant you
42:15get full credit for that just like your
42:16lecture
42:17clicker questions mention you for credit
42:19um there's no quicker questions for the
42:20live lectures only for the content
42:22lectures
42:23now here's the idea you and your partner
42:26will do them together
42:27you'll find a zoom room you'll find a
42:28partner in piazza you'll find a zoom
42:30room
42:30you can also come to lab office hours
42:32we'll have people we have staff ready to
42:33support that
42:34when you're all done to get checked off
42:36you either get checked off in the lab
42:37office hours
42:38or you'll get checked off in this
42:4013-minute check-off spot
42:42this is the only required face-to-face
42:44interaction so actually
42:46we try to be supportive and flexible for
42:49this online space if i can't make
42:50lecture we're filming every lecture
42:52we've already pre-filmed the content
42:53lectures
42:54and we'll be filming these live lectures
42:55done if you can't make the lab
42:57you only have to show up once a week
42:59meaning you can't just
43:01page out and asynchronously take this
43:02course the only one thing you have to do
43:04is show up for the checkoff spot 13
43:06minutes you and a partner it's the time
43:09for our students our staff will support
43:12you
43:12we're going to find out how are you
43:13doing how is the class going are you
43:15ahead behind
43:16it's a time to get some human
43:18face-to-face connection
43:19this online world with 1200 students
43:21it's just so hard to do this
43:22we have some discussions it's so hard to
43:24have some to face this is the time for
43:26that so
43:27you have to have this checkoff spot have
43:28to have a hazard going spot
43:30in the 13 minute checkout spot that's
43:31our chance to find out how you're doing
43:34discussion typically was an hour happen
43:36at the end of the week where we
43:37summarize the week and we maybe have a
43:39worksheet that drives
43:40everything the same thing is going to
43:41happen but we realized you know you
43:43could watch the videos to do that
43:45if i give you the worksheet and say work
43:46on it and then we have a video of our
43:47tas doing the worksheet giving the
43:49solutions
43:49you can get just as much benefit so
43:51we're gonna but if but we encourage you
43:53to come to the live discussion so we're
43:56going to have three of them only three
43:571200 people we would have had 40
43:59sections
43:59you know what we're going to take our
44:01staff and actually focus on those check
44:03offs that
44:04face-to-face interaction that's where
44:05we're kind of paying are taking the
44:07currency of all of our staff and pouring
44:09it into the face-to-face small group
44:10learning
44:11and the check-off in the in the how's it
44:13going space so discussions we've got
44:15three sessions
44:16spaced all across the day those are
44:18going to be video recorded
44:20so you can watch those if you want to
44:21they're all the same material but you
44:22can just see different people attack the
44:23material different ways
44:24and we're going to have our pre-recorded
44:26uh discussion videos you can watch all
44:28those are just resources we'll give this
44:29to you
44:29so that means if you work with if
44:31synchronous learning works for you
44:33you can do synchronous come to lecture
44:34live uh yours the content is still
44:36pre-recorded but you'll come to lab live
44:38you come to discussion live
44:39but if you actually because of some
44:40situation you can't do that no worries
44:43we got you covered
44:44watch the you know do it on your own and
44:46then get the check off in lab that's you
44:47have to at least we say
44:48face to face and loud check off and you
44:50can also just watch the discussion video
44:51and watch a lecture video so all those
44:53things will be great
44:53and yes discussion videos will be
44:55recorded we also have other parts of the
44:58course
44:59we have office hours drop in office
45:00hours get questions answered we're going
45:01to have that as well
45:02we typically have tas typically run the
45:05majority part of the office hours those
45:07are
45:07called conceptual office hours i'm
45:09confused about parallelism or
45:11caches or whatever the topic is and
45:13they'll kind of concept kind of like hkn
45:14office officers like that
45:16there's also office hours they're not
45:17conceptual but about help me get through
45:19this project i'm working on it i'm stuck
45:21what can you help how can you help me
45:22with and there might be some i got this
45:24bug and then my t
45:24the ta might be teaching you how to
45:26debug they're going to just hand like
45:28line 47 that'd be the worst ta ever says
45:30line 47. the best ta is
45:32teach me what you're thinking about tell
45:33me what you were asked to do how did you
45:35know it was wrong
45:36they're working with you to teach you
45:37how to debug it so that you find it
45:39yourself we don't want our tas to give
45:40the answer out that
45:41that doesn't teach you how to fish we
45:42want to teach you how to debug it so you
45:43can solve the next problem you have
45:46so the way we do this online is we're
45:48going to have the same
45:49kind of thing again lab office hours
45:51online on zoom
45:53project office hours for those and
45:54regular office hours which are great
45:56for anything else and that'll typically
45:58be um
46:00as i said either conceptual and they'll
46:01also be at a faculty office hour so
46:03you can i'm gonna have two times and
46:05borah can talk about what he's gonna
46:06have i'm gonna have
46:07a general office hours for everybody
46:09just drop in and i used to be able to
46:10have 13 shares in my office now i can
46:12have
46:121200 if you want to drop in it's going
46:14to be a time where i'll typically do
46:16kind of a summary of stuff i'll i'll say
46:18well look you know next week we've got
46:19an exam let's look at some exam
46:20questions i might pull out some examples
46:22do them together
46:23it's really great i also going to have
46:24if you want to talk to me personally
46:25privately 101
46:26i'll have some individual slots you can
46:28sign up for and obviously i don't have
46:29an infinite number of those but you can
46:30grab some infinite time you just want 10
46:32minutes to
46:32to talk about something in private
46:34that's always people used to do that
46:35back in the days
46:35hey dan can we walk outside i would walk
46:37outside my office and have a private
46:38conversation i'm going to have to have
46:39sign up times for that so we'll do that
46:41and those are all going to be during the
46:42same slot
46:44monday wednesday friday four to five is
46:4661c slide and my slots going to be
46:48group office hours on wednesday and
46:50signup office hours are going to be on
46:51friday
46:53exams in the olden days okay what are
46:55you doing for office
46:56my friend you want to talk about that
46:57i'm just going to do the mirror review
46:58uh yeah
46:59yeah the general one on friday
47:02um sign ups on wednesday
47:06perfect perfect we'll try it out that
47:07way if we need to modify we'll modify we
47:10are here to help um one thing i wanted
47:12to put a pitch for these mondays
47:14so we know what we're doing on
47:15wednesdays and and fridays i
47:17really would like everybody to
47:20get together for these monday meetings
47:22we're doing something special
47:24we are dan and i put an effort to bring
47:26guests from outside
47:29don't complain
47:36do
47:40they're all surprised people so you
47:42gotta come and you won't know until you
47:43get there because we want to keep it a
47:44surprise
47:45yeah we you know we can't publicize that
47:48we might break
47:49zom right
47:54good stuff uh i'm going to record my
47:56general office hours borah there was a
47:57question on the on the q a for that i'm
47:59going to do mine
48:00there's no reason i should do that too
48:01yeah we're just going over problems
48:03that's a little bit i'll say anybody
48:04have any questions i'll try to summarize
48:05some stuff and i'll try to do
48:06re-teaching of that which would be great
48:08but we only have five minutes i'll make
48:09sure we get out because we have i think
48:10chemistry is right after us
48:12um so the way we're going to do exams is
48:15a little different you're going to be
48:16given
48:1624 hours and we'll give you an exam it's
48:18going to be a randomized exam you know
48:20that that's kind of the new software
48:21we're using now and you'll get a new
48:22copy and you'll be able to use
48:24all open book never done that before
48:26open book exams love that
48:28but close internet you can't search it
48:30you can't talk to your friends and we're
48:31going to ask you to take your cell phone
48:33here it is and make it can't see myself
48:35you're going to take and make
48:36a time lapse video you put it so your
48:38cell phone can see over your shoulder
48:40see your hands see your screen and you
48:42say go make it a time lapse
48:43time-lapse means it's just much shorter
48:45video then you take the exam when you're
48:47done you stop
48:48it you submit the exam so that's the way
48:49we make sure that cheating isn't
48:50happening we keep it all honest
48:52that'll be i got i think seven slides to
48:54go through and five minutes so let me go
48:56faster dan
48:56pin yourself right you can't see yeah oh
48:59it's been
49:00there we go no way thanks okay course
49:02grading we got
49:03four minutes here we go go to dan mode
49:05here we go uh lecture lab
49:07homework projects quest there's some
49:09exams in there
49:10there's some sprinkle points for being a
49:12good citizen for apa
49:13on top of that and there's also going to
49:15be a performance competition so there is
49:16some place you have to give some extra
49:18credit for that
49:18um four projects awesome the first two
49:21are individual the second to a partner
49:23and you've you've seen you know labs and
49:24homework so labs to check off lecture
49:26that's the clicker quiz we talked about
49:27homework is there
49:28and projects are there the grading
49:31policy we're absolutely great we're not
49:32curved we want you to work together
49:34collaborate teach each other this is the
49:36space this is an amazing space
49:38curve i believe curving is evil where
49:40you can't you know 100 people all
49:42the students get a's no you can't
49:44because you're not allowed to get all
49:45these why that's crazy you can't give
49:47all your students a's if they've done it
49:48so
49:48i think curving is not the right thing
49:50to do we've been absolute grading for a
49:52long time
49:53job internship interviews the material
49:56whatever grades you get the material you
49:58learn in this class is the most
49:59important you know the most important
50:00classes for your interview
50:02huh cs70 when you're the most second
50:04most important class for your interview
50:0661b but the third most important class
50:08for interview is 621c and we're sticking
50:10by and we think it's great so
50:12we love this class we think you're going
50:13to learn a lot from it it's awesome all
50:14right
50:15our goal as instructors is to make the
50:17greatest course ever we want to be 7.0
50:19hca instructors at the end of the year
50:20we're trying to give all these
50:22pro-student policies humor enthusiasm
50:24technology guests
50:25that you're on the news guests fun and
50:27challenging programs hopefully you
50:28really love these projects and you
50:29really learn from them as well
50:31pro student we have exam clobbering
50:33other classes don't do it i do
50:35you learn it later you can override an
50:37earlier exam we love that
50:38we want to be that give us feedback let
50:40us know how to improve this class all
50:42right two minutes
50:44apa if you're a good student show come
50:46up to office hours complete all the
50:47assignments turn on your homework zero
50:49um ask great questions and do make it an
50:51interaction do the performance
50:52competition help others in lab and
50:54piazza write some software do some other
50:55thing that's
50:56out of the box that teaches people boom
50:58all those things are ways that you can
50:59get altruism credit help people on
51:00piazza is another way and ask questions
51:02on piazza that's kind of an easy way to
51:04kind of measure those things all right
51:05a minute 30. we have slip days you know
51:07that we up typically have slip days we
51:08have three slip tokens not slip hours
51:10not sip minutes you can slip an
51:11assignment and you can use these three
51:13tokens
51:13either all at once or save them until
51:15the end or one at a time
51:16when you need them this is an important
51:19thing
51:20please read this slide all projects
51:22aside from actually
51:23this is updated besides the first and
51:25second i'm sorry for not obtaining that
51:26page we'll be done with the partner
51:28so second and third third and fourth is
51:31where the partner first and second are
51:32solo
51:33you cannot obviously you can't cheat
51:35don't don't share material don't have a
51:37public github repository anybody who
51:38gives the material
51:39is equally responsible someone who gets
51:41the material all those things are bad
51:43we have caught in the last several
51:44semesters 50 to 100 people
51:46and we prosecute all of them and we will
51:48do it again this year
51:50so don't do it take a step back if
51:52you're thinking you're doing this ask
51:53for a septic
51:53talk to us we would love to say you know
51:55dan something's going wrong i got some
51:56code smoke whatever we will give you
51:58some special things if you need this
51:59semester because we know that online is
52:00weird
52:01do not cheat folks if you get that point
52:02please do not take it
52:04take a step back we made it in summary
52:0720 seconds left
52:0861c is about six great the ideas of
52:10architecture borah
52:11burst through it a great job of covering
52:13these six the whole semester is
52:15revisiting this and hitting these six
52:17we'll see it over and over and over
52:18borah take us out with some thoughts one
52:21thing
52:22you said everything great one thing that
52:24you then put on powerpoints
52:25there will be prizes for the best
52:27performers in projects
52:29yes um apple will tell us tomorrow
52:33uh that cs61c is one of the classes that
52:36is on the
52:37price list
52:40they're excited so please please do that
52:43prize police performance competition
52:45there's chances for actual physical
52:46prizes we'd love to have that
52:47folks we are so happy that you came i
52:49don't want to go too much over but we're
52:50sober from 61c ladies and gentlemen
52:52621c
52:54[Applause]
52:57so happy you're here we're going to have
52:58a great year we got a great staff for
53:00you
53:00we will see you on monday have a great
53:02weekend folks
53:03go whoo good job everybody
53:10and stop the live thing how do we stop
53:12the life the emitting for all
53:14let's see stephen how

About this transcript

This page contains the full transcript of [CS61C FA20] Weekly Lecture 01.LIVE - Great Ideas in Computer Architecture, Intro by CS 61C Departmental, generated from the public captions YouTube serves with the video. The transcript has 10,553 words across 1,738 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.