YT YouTube2Text

YouTube2Text → Transcripts → [CS61C FA20] Weekly Lecture 04.LIVE - RISC-V

[CS61C FA20] Weekly Lecture 04.LIVE - RISC-V — Transcript

by CS 61C Departmental · 9,882 words · 1,623 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00panelists we could we we should mute
0:03and then borah will welcome us in and
0:04then we'll end up uh you know
0:06spotlighting you when we're talking to
0:07you that'd be good yeah and i'll read
0:09the the
0:11first the weekly news perfect wonderful
0:14and we should see some people up here
0:15look at this we already have
0:1750 folks here wonderful so great to see
0:20everybody
0:26hello hello everybody hello
0:32another minute yeah we'll start and here
0:35we go pushing record we remembered it
0:36this time
0:38recorded this computer
0:43all right all right ladies and gentlemen
0:47welcome to our fourth once a week series
0:50live lectures with more of myself ladies
0:52and gentlemen 621c
0:55[Music]
0:56great to see everybody all right four
1:00take us home
1:00all right agenda for the day where
1:03are we uh are my slides working
1:07and almost so what are we going to do
1:11we are in week four now so we're gonna
1:13cover interesting stuff
1:14that we have for this week um
1:18first we are going to cover things that
1:22are computing in the news there are two
1:24very interesting and very relevant
1:26to this class announcements um in the
1:29press
1:29this morning or overnight um and then
1:33you know incidentally i did i don't know
1:35how this was timed i think this
1:37is just incredible we have two
1:40you know rare experience for everybody
1:43in the class
1:44um two guests uh andrew waterman and
1:47janssep lee
1:48from sci-fi who if you did not know
1:51they created or at least
1:54heavily participated in creation of this
1:57five
1:58um and so they'll and they founded a
2:01startup company called
2:02sci-five that is making these five
2:06processors so after that we're going to
2:09talk about
2:10quest due dates and open mic
2:14but for now the news if you haven't seen
2:18last night finally um nvidia and
2:23arm enter the deal in which
2:26nvidia will be acquiring arm if you
2:30are not following things we mentioned
2:31this early on we had that in discussion
2:33in that open session with
2:35apple arm is a dominant architecture at
2:38the moment in
2:39the commercial world in embedded
2:41processors
2:43they have been acquired a few years ago
2:45by softbank they've been a
2:47public company incorporated in uk
2:50back in 1990 and
2:53they are essentially in pretty much
2:56every
2:58mobile phone that is out there so they
3:00are now
3:02gone to uh nvidias nvidia purchased them
3:06nvidia is a
3:08big graphics or a gpu company and
3:12there is a lot of speculations out there
3:14why did then be they buy
3:16arm and what is going through the future
3:17of arm are they going to continue
3:19licensing immediately immediately
3:21immediately commented about three hours
3:24after acquisition that
3:25the licensing stays unchanged so the
3:28world can
3:29be relieved but that also puts a huge
3:33spotlight on
3:34risk five now
3:38you know we are all at least five
3:39experts by now right we have
3:41seen two lectures already hopefully two
3:43of them are out
3:45the other piece of the news that came
3:47out it was in the shadow of this
3:49is that sci 5 is to introduce a new risk
3:525
3:53processor architecture and risk by pc
3:57at the conference in october
4:00i you know this is like kind of a press
4:02release
4:03came down the business fire and so on so
4:05it's kind of not that easy to understand
4:07it so
4:08we have uh jensep and andrew over here
4:12they may be able to tell us something
4:14may not you know they're now in
4:16in the corporate world so um
4:20andrew and janssep both graduated from
4:23cal about
4:24five years ago with their phds they made
4:28risk 5 possible
4:29they invented risk 5. and
4:33they after their graduation they went to
4:35found
4:36a hardware company which was sci-fi that
4:39made risk 5 it makes nowadays five
4:41processor
4:42processors they've been quite a bit in
4:44the news it's an exciting life
4:46uh sometimes uh they may tell you that
4:48maybe too exciting
4:50um and um
4:54so maybe i i don't need to labor too
4:57much
4:57i'll just ask them perhaps to each one
5:00of them tell us how
5:01they what were their career paths how
5:04did they end up
5:05at cal in grad school and how did they
5:08end up in the startup
5:10after that so maybe start with andrew
5:13okay yeah so uh thanks for having me
5:15borah so uh
5:17i guess uh my going to cal and starting
5:19working on instructions and
5:20architectures is
5:21actually kind of serendipitous uh when i
5:23was a first semester
5:25senior in undergrad at duke um i was
5:28taking
5:28a graduate computer architecture course
5:30because it sounded interesting and the
5:32professor dan sworn who was teaching it
5:34brought me into
5:35his office and asked me like you know so
5:37what are you doing have you applied for
5:38any jobs
5:39no have you applied to grad school no
5:41you're an idiot andrew uh you should
5:43become a computer architect i'll write
5:44you a letter of recommendation
5:46so uh so um you know he recommended a
5:48few places to go
5:49uh calcium pretty awesome uh berkeley
5:52the city is being pretty awesome so i
5:53ended up coming here
5:55um and uh you know over the course of uh
5:58of grad school like you know youngstock
6:00me and some other folks were
6:02uh building processors and simulators
6:04you know to conduct our research
6:06experiments
6:06and uh we you know got kind of tired of
6:09the vagaries of the
6:10more popular uh computer architectures
6:12at the time there's like a lot of
6:13complexity in them
6:15and you know some complexity has its
6:16purpose you know it can lead to good
6:18performance but a lot of the complexity
6:20and other instruction sets
6:21is not there for some technical reasons
6:24so
6:24you know we ended up you know starting
6:26on this project to simplify our research
6:28lives and also to make it easier to
6:29teach
6:30some of the undergrad courses and grad
6:33as a cow
6:34because it's also easier to teach
6:35simpler isa and
6:37you know i don't think that we had uh
6:39any expectations that things are going
6:40to take off quite the way they did
6:42but um you know over the course of time
6:44it became apparent that it was a niche
6:45for an o
6:46for a high quality open instruction set
6:48architecture
6:50um and you know towards the end of my
6:52grad school career
6:53uh we you know had the you know the
6:55great opportunity to start a company to
6:56commercialize some of the ideas uh you
6:58know to build microprocessors
7:07all right how did you end up here oh
7:10yeah
7:11hello everybody my name is joseph lee uh
7:14i actually did my undergraduate in korea
7:16a university called curry advanced
7:18institute of science and technology or
7:20kaist
7:21uh you know i always wanted to learn how
7:25to build processors
7:27and at the time i have applied to a
7:30couple of grad schools in in the states
7:32uh berkeley was gracious enough to admit
7:36me
7:36to to the program and i came
7:39and worked with krista sanovich
7:43and dave patterson and borah and andrew
7:48we and uh
7:51you know as i said i wanted to learn how
7:53to build computers the best way to learn
7:55is to build
7:56we actually ended up building a spark
7:59machine
7:59first and then a mips machine and then
8:02we had a couple things we didn't like
8:04about it
8:05um and we kind of ended up
8:08doing our own thing and as you all know
8:11uh
8:12the term risk has been coined by dave
8:14patterson in the 80s
8:16and so we actually
8:19asked for advice on what we should name
8:22the isa
8:23um and dave kind of had a brilliant idea
8:26to call it
8:27risk five the fifth version of risk
8:28coming out of berkeley
8:30and i think that really helped kind of
8:32pushing the name out
8:33because everybody had learned from the
8:35same textbook that you know
8:37um i think you're you're learning from
8:40and risk was not a foreign term
8:41um so risk five started in 2010
8:45actually we had a 10 year virtual
8:48reunion
8:50of the risk by project where we had a
8:52bunch of interviews and things like that
8:55talking about how we got here uh and
8:59in 2015 as andrew said we started sci-fi
9:02and the risk five foundation to
9:04commercially
9:05kind of make risk five happen and it's
9:08been five years
9:09since uh we were at it
9:13and i'm very disappointed in that
9:15picture and you know that picture
9:17of the first blue t-shirts of this five
9:19were taken at hot chips
9:21and you know what i was doing on that
9:22day i was giving an ic prelim
9:24and berkeley that's a phd program so i
9:27did not go
9:28to i did not go to hardships
9:31that here and i'm missing in the picture
9:33i need to photoshop myself in because i
9:35do have a blue
9:36shirt it's the first five shirt
9:39um anyway um can you guys
9:42either one of you can share something
9:45about how a risk fight
9:46got started i mean we have a lot of
9:50young students here who are learning
9:51about risk five and i think they would
9:53love to hear something about that is not
9:56you know that that that may not be
9:58that well known i mean risk five is well
10:02known now but it has kind of started out
10:04of a
10:05blue i kind of gave my version uh
10:09briefly
10:10uh young stuff you want to go yeah
10:13um i think it was probably summer 2010
10:17i think it was about may uh
10:20i think it was andrew who kind of came
10:24um into the office
10:28saying you know why are we fixing
10:31or discovering other people's problems
10:34in the isa and trying to replicate them
10:37so i guess a bit of backstory on the
10:39spark machine that we
10:40originally built it was a spark v8
10:43machine
10:45um the whole idea of using a
10:47commercially
10:49you know available isa was we thought we
10:52could reuse
10:52all the software ecosystem you know like
10:56the compilers the operating system ports
10:58that were already
10:59already built for the isa and it turned
11:03out
11:03you know we have read the sparkv manual
11:06in and out
11:07and we have implemented the processor
11:09correctly
11:10and we thought after we did that we
11:13could just put
11:14linux um you know on from the main
11:17source tree on it
11:19and it turns out it didn't boot so
11:20obviously we thought it was our problem
11:22so we kept debugging it to the point
11:25where we found out that
11:26actually the linux kernel of spark b
11:28worked around a hardware bug
11:30uh in the silicon and we were just
11:33finding it
11:34and we were replicating the same bug so
11:36that we could use the same software and
11:37i think that day
11:38in summer 2010 andrew comes in and says
11:40why are we doing all this
11:42there are probably some curse words in
11:44the middle of that sentence yeah we're
11:45being recorded so
11:47um you know andrew was saying why are we
11:48doing this
11:50and it was like why don't we just do
11:51something that we like because
11:53you know the myths i say um i don't know
11:55whether
11:56some folks know about this but there are
11:58some kind of weird things about it
11:59there's like
12:00something called the branch delay slot
12:03which um the shadow
12:04instruction of a branch would be
12:06executed and it was really tied to a
12:08five-stage pipeline that they
12:10initially developed so they kind of
12:11couple the hardware to the software
12:14and we obviously didn't like about that
12:16so when we're building our second
12:17machine on mips we kind of got rid of
12:19that and
12:21at the point we were like you know let's
12:23just go do something we like and
12:24we started doing some stuff we changed
12:27um some of it um i think we actually
12:30used a very initial version in cs 152
12:34so if you go look back in like
12:372012 cs 152 um you'll probably see the
12:41initial versions of wrist five
12:43which um basically is not the same isa
12:47actually the one that's available today
12:49but
12:50you know we kind of incrementally made
12:52first five
12:53better and better and better um to the
12:56point where
12:57in 2013 i think we put everything online
13:03risk5.eke.berkeley.edu
13:04and we didn't really know who would be
13:06interested in
13:08using it and but but andrew kept
13:12changing clients saying to the point
13:15where some people started emailing us
13:17asking stop changing the isa we're
13:20trying to do something with it
13:21so so this game was quite a surprise to
13:23us uh so um you know researchers at
13:25other universities and
13:26and actually even a couple of people at
13:29uh
13:30for-profit corporations were using this
13:32thing that we had just put on the
13:33internet that we were using
13:34you know for our only for our own
13:35research and teaching purposes
13:37at the time or so we thought so we
13:40weren't being uh
13:41you know a bunch of meanies by uh by
13:44like
13:44keep continuing to tweak and make
13:46backwards and compatible changes to the
13:47design we were doing so because
13:49we were trying to improve it and didn't
13:50realize that other people were relying
13:51on what we were doing yet
13:54yeah that stuff happens when you put
13:55stuff on something open in the internet
13:57somebody may be watching you
13:59indeed actually absolutely true
14:04all right a few so i'm assuming folks
14:06who are dialing in are the ones that uh
14:09are keeping up with the lectures and so
14:11on so
14:12so far i mean if they watch i recorded
14:14videos and uh
14:16up to the stuff we haven't posted quite
14:18the stuff for the for wednesday
14:21but up to wednesday we have covered the
14:23base isa
14:24integer is a 32
14:27bit version so can you share some
14:30insights you know
14:31were there some instructions there that
14:33didn't quite make it
14:34that for early on in and then they
14:38uh get dropped well i guess there's kind
14:41of two answers to this question
14:42one thing is like we we developed this
14:44philosophy of having a very small base
14:46isa and then having a bunch of
14:48optional extensions uh idea being that
14:51the the base isa is
14:52suitable for like very simple embedded
14:54computers and is
14:56also suitable for you know for
14:58educational purposes
14:59um and then there's a whole bunch of
15:01instructions in these other extensions
15:02which are not
15:03bad ideas they just are more than you
15:06need for certain kinds of systems like
15:07for example
15:08um floating point arithmetic
15:10instructions and virtual delivery
15:12but i think to uh but i guess the other
15:14answer is like was there stuff in there
15:15that got dropped i think you know one
15:17thing that comes to mind is
15:18some other architectures have uh fairly
15:21complex
15:22addressing modes for their loading store
15:23instructions like you know you can
15:25compute an effective address by like uh
15:27taking a register and adding it to a
15:30shifted copy of another register and
15:32adding a constant stuff like that like
15:33the intel architecture for example has
15:34this addressing mode
15:36and stuff like that we considered they
15:38didn't end up making the cut because it
15:39adds complexity to the simplest
15:41implementations makes the isa
15:43harder to teach and for a lot of
15:45applications it does it just doesn't end
15:46up helping that much
15:48and i think andrew yeah and andrew i
15:51think you got rid of zero
15:52extension right in the i say oh that's
15:54true so um
15:56yeah so one property of the uh
15:57arithmetic instructions in this five is
15:59that uh some of them have immediate
16:00operands
16:01and the immediate operands are always
16:03side extended even when they're
16:04logically
16:05describing unsigned quantities and this
16:07simplifies the hardware implementation
16:09and it actually
16:10didn't have any negative effect on
16:11performance despite being simpler
16:13so uh i think uh yeah the lesson is
16:15don't assume that you know the people
16:17who came before you necessarily got
16:18everything right
16:20and there were things that actually got
16:21more complicated as well
16:23um you probably all wonder why
16:26the complicated immediate kind of person
16:30there is i'll tell you the original
16:32version of risk five
16:34you can actually go back um in history
16:37on cs152 lectures
16:39it didn't start out like that and i
16:41actually for the record i was against it
16:44and this is kind of you know caution um
16:46uh cautioning you all
16:48just you know optimizing some stuff
16:53maybe andrew can talk a little bit about
16:55why it ended up that way
16:57yeah i don't want to go into too much
16:58detail since y'all haven't covered that
17:00yet in 61c but
17:01um but the basic uh premise here is that
17:03sometimes you can make trade-offs that
17:05slightly reduce harbor cost but
17:07significantly increase the logical
17:08complexity of the thing and
17:10that's a tough judgment call to make
17:13right uh yeah and it's it's a
17:16that rastness rat's nest that is going
17:19to show up in next week's lectures
17:21is a tough one to explain uh by the way
17:24i mean i
17:25i when i learned computer architecture
17:28as an undergrad i
17:29they spent about two weeks on x86
17:31addressing modes and i
17:32don't remember them all anymore because
17:34there are so many of them
17:38you know this is risk five i didn't
17:39mention that it's a simple load store
17:41machine
17:42the only thing you can do are loads of
17:43swords
17:45a few other questions um
17:48i was going to ask you about envionness
17:50but uh i'll skip on on that
17:53um well i'm going to answer that anyway
17:56let me ask you the question yeah i'm
17:58still about the envy on this
18:00and there is a lot of uh people i mean
18:02rhys 5 is a little indian but there was
18:04people who are pushing for this fight to
18:06be uh by
18:06indian andrew strong opinions
18:10so um okay so here's my strong opinion
18:13it's not that i'm a bigot for little
18:15indians
18:15so uh i make the analogy to
18:19which side of the road you drive a car
18:20on so you know it's the opposite in the
18:22uk as it is in north america
18:24and both systems are perfectly valid no
18:26one has neither has any significant
18:28advantage of the other
18:29the screw-up is that there are two
18:30systems and you have to learn two
18:32systems in order to operate in both
18:33contexts so little indian isn't better
18:36than big indian
18:36or vice versa really but uh but you know
18:39different distance
18:40different like systems designers back in
18:42the middle of the 20th century made
18:43their choices there unfortunately um so
18:46we chose little and even initially
18:47because that was most popular in
18:48applications and server processors at
18:50the time like the intel architecture for
18:52example
18:54um
19:00so the thing is the pendulum might be
19:02swinging back a bit in the sense that
19:04we probably will have to make it a buy
19:05indian isa ultimately because
19:08um some markets are big indian like uh
19:11like pretty much everything in japan
19:12that's what is big indian um so uh in
19:15order to make risk five applicable there
19:17we don't wanna say
19:18you know you know japan shouldn't build
19:20just five processors because we prefer
19:22little india
19:23so it's going to end up being
19:26a big deal all right so shifting gears a
19:29bit
19:30so i before sci-fi they were
19:33really for about five years there are no
19:35hardware startups
19:37you guys went into what was considered
19:39to be in completely uncharted territory
19:41so
19:42how is life in a hardware startup i mean
19:44this is you guys are not a time
19:46startup anymore i i don't know can you
19:47share how many people are in
19:49sci-fi but a lot more than uh
19:52three or four of you that were early on
19:56yeah so how's the light in in our
19:57startup uh
20:00is a cto and and uh andrew is a chief
20:03architect and you know they
20:05are still maybe in their 20s or
20:08you know just left their 20s you can
20:12go for it yeah um we're currently 500
20:16people so
20:17life is pretty different compared to
20:19when we started the company when we were
20:21three people
20:22but yeah um five years ago when we
20:24started the company hardware startups
20:26were not a thing it was actually two
20:29years ahead of the ai
20:30kind of wave of hardware startups
20:34uh i've been kind of talking with lots
20:36of
20:38investors on you know starting a
20:41free and open processor based you know
20:45startup and just think about that sound
20:49when there's no harm virtually no
20:51hardware startups and you go meet an
20:52investor and say hi i'm jensep lee i
20:55am interested in building a you know
20:57open source processor based startup
20:59the running joke was the more the
21:02investor knew about hardware they would
21:03cut me off earlier and the world's
21:05record is two sentences actually it was
21:07it was an investor who had a stanford ee
21:11phd perhaps that was the problem
21:13but that uh he he i went in two
21:16sentences i said hi i'm joseph lee i'm
21:18i'm interested in building a
21:20open source processor-based startup he
21:22just cuts me off there and says
21:24not interested good luck you know i
21:28later you know to thank him he saved my
21:30time he saved his time but
21:32you know eventually um we found
21:35surrey hill ventures who were interested
21:37in building infrastructure-based
21:40startups and they were they thought
21:43verse 5 would become a big thing and i
21:46think
21:47obviously we can't predict a lot of the
21:50news like the one that came out
21:51yesterday where
21:53nvidia acquires arm and all of a sudden
21:56the
21:56entire semiconductor industry is kind of
22:00you know just trying to cope with what's
22:02what's going to happen and
22:04i think it's a it's it's a it's a big
22:07opportunity for risk buy but
22:08obviously you can't kind of predict
22:11those events
22:12when we're doing a startup but you know
22:15startup life is interesting in that
22:17sense where there's a lot of twists and
22:18turns
22:19um on a lot of external factors but you
22:22know i think
22:23the parts that we can control is what we
22:26do
22:27internally at sci-fi where you know we
22:30get
22:31we get at it every day and we make
22:34incremental progress every day
22:37thank you this is it's pretty crazy i
22:40mean
22:41what you guys went through is pretty
22:42crazy um
22:44and i mean i i can share that one of the
22:48um of the venture capitalists before
22:52funding you you know comes to me and
22:54says have you heard of
22:55open source hardware i mean is this is
22:57this
22:58make any sense totally
23:01totally yeah it's amazing
23:05how much things have changed since then
23:06uh you know i think a lot of it has to
23:08do with uh
23:09you know with the these revolutions
23:10happening in machine learning and
23:11artificial intelligence
23:13but uh you know venture capitalists seem
23:15to be letting like willing to let a
23:16thousand flowers view
23:17and they're funding a whole bunch of
23:18parker startups including some that are
23:20fairly similar to each other it's
23:22very different than it was even just a
23:23few years ago
23:25yeah even though even though it sounds
23:28like a
23:29pretty big change i think a lot of the
23:31external factors were there
23:34i think perhaps the biggest one is the
23:36slowdown of moore's law
23:38where you know when it was the golden
23:42era of moore's law was every 18 months
23:44you would get smaller bigger i mean
23:47smaller transistors and
23:49exponential amount of them 2x of them
23:52so in that era um there was kind of no
23:56need to customize
23:57the hardware per se where you get out of
24:00your way to go
24:01you know build a custom architecture for
24:04a certain domain
24:05you really didn't need to do that the
24:07the job of a computer architect was turn
24:09those
24:10extra transistors into better faster
24:13computers right
24:14and there's obvious places to use those
24:17transistors you know
24:18make the caches bigger you know
24:20integrate a floating point unit
24:22those type of ideas but now like
24:25once moore's law slows down there's
24:28really no
24:28golden formula to make your processor or
24:31your compute chip better
24:32and you know the need for customization
24:36was rising and
24:37there was just it was not really
24:39possible
24:40and you kind of see how linux changed
24:43the entire world
24:45so i think there was those trends i
24:47think the betting part the gambling part
24:49was really how fast it would all unfold
24:51and i think
24:52that part was kind of to our surprises
24:55going really really fast we're still
24:57going fast
24:58so it's it's a really exciting time to
24:59be a computer architect uh even
25:01because moore's law is ending uh you
25:03know it's it's pushing a lot more uh
25:05effort into customization apart
25:08therefore you know specific
25:09complications and algorithms
25:11which uh as you have said you know
25:13people hardly considered doing
25:14in general purpose computing even 20
25:17years ago
25:20that sounds um all really exciting so
25:23i mean our students are definitely now
25:25confused i mean here are a lot of things
25:26i mean
25:27believe me dan and i pitched in like the
25:29first lecture this
25:30need for customization it's not that
25:33easy
25:34to do that it's a really good marketing
25:37pitch
25:38for now it really sells well at all
25:40levels but it
25:41will need a few villages to implement
25:43that
25:44um the question do you have career
25:46advice for our students
25:48you know and then related to that um do
25:50you hire undergrad interns by any chance
25:52uh you know how does that happen i mean
25:55uh do we
25:56you know should they just apply on the
25:57website is there a magical way how
26:00how does that happen um do you hire is
26:02there a secret handshake that's what i
26:04wanted is there a secret handshake they
26:05could the berkeley
26:07you know secret handshake to get in
26:08that's so that's what we're right now
26:10yeah what's the password 61c
26:14well you can email younsep at scifi.com
26:16or android scifi.com
26:18while i give our 150 of those
26:21well you know bring it on uh yeah i
26:24think um
26:25i think curry advice um you know i i had
26:28great great mentors throughout
26:30i guess my you know limited career so
26:33far but
26:34i think one thing that really resonated
26:35with me was um
26:38uh when i
26:41say something you know an idea to
26:43somebody else
26:44and you know i get like a murky kind of
26:48response
26:49uh well i you know don't get discouraged
26:52by that there there could be a couple
26:53reasons behind that
26:54um uh or when people say well that
26:58sounds impossible
26:59you know it's kind of a moment to stop
27:02and think about it
27:03um obviously not all ideas are great but
27:05at least don't get
27:06discouraged by some people saying you
27:08know like maybe that's not a good idea
27:10or something you know because
27:12if i heard all those things so far we
27:14probably wouldn't have
27:15you know gotten to risk five but you
27:18know
27:18i i ended up you know kind of doing
27:21interesting things exciting things that
27:23really was fun to me with
27:25you know great friends and colleagues
27:27like andrew and i think you know it's
27:28it's it's really it's really great yeah
27:32so i find it kind of difficult to give
27:33career advice because i think a lot of
27:35uh the success stories that we've had is
27:37serendipitous not to say that we haven't
27:39worked really hard for it as well but
27:41you know i think like a lot of things
27:42lined up in ways that that we didn't
27:44really anticipate
27:45but i guess like you know one bit of
27:46advice i have is find people you like
27:48working with
27:48and continue working with them that's
27:50been a consistent theme the whole time
27:52you've been doing this so it's worked
27:53really well
27:55we got plenty of those but i mean this
27:57is pretty crazy story right 2010
27:59junior graduate students
28:02started a new architecture 10 years
28:05later
28:05it's not just that sci-fi has 500
28:07employees
28:09doing their dream job but there is like
28:11a dozen
28:12risk-five hardware startups startup
28:15seller or more i i don't know how many
28:17um i i stopped counting when when we got
28:19to i don't know five or six or something
28:21like that
28:22um um i have
28:26two quick questions and then we need to
28:27go back to to our agenda
28:30one is
28:34in today's news in a shadow of the arm
28:37acquisition there is the announcement
28:38that
28:40sci-fi will in october present uh risk
28:445 pc you know that's a dream right at
28:46that moment that we have closed
28:47everything
28:48is there anything you can say i i know
28:50that you know that you're supposed to
28:52make an announcement and you know
28:54yonsept is the one that is supposed to
28:55do
28:56to do the speaking there is there
28:58something that you can say
28:59msbs or you know a couple of bits of
29:01that yeah
29:02um so so far when we started sci-fi
29:07perhaps the biggest problem or the
29:09biggest challenge rather
29:12was to build a software ecosystem around
29:14a new
29:15isa or new architecture because
29:18at the end of the day what sells
29:20hardware is software there needs to be a
29:22lot of interesting software that runs on
29:24top of
29:24you know the hardware that's the only
29:26reason why the hardware is interesting
29:28right
29:28so we're built we're trying to bootstrap
29:31a linux software ecosystem
29:34and it's kind of a well-known secret
29:36that software people only start working
29:38once there's like a piece of hardware on
29:39your
29:40desk and that's definitely true and at
29:42the time we really didn't have that
29:44you know that that piece of hardware on
29:46the desk so
29:47we ended up building a development board
29:49called high five unleashed
29:51that was our initial version we had very
29:54limited production of the board
29:56a lot of people were very excited um
29:59using that board to develop their
30:02favorite risk five software um since
30:05then we had a lot of requests to
30:06continue developing on that so you can
30:08kind of
30:09you know continue thinking that it's you
30:11know it's a next step up to that board
30:14and if you're going to build a
30:15development board why not kind of make
30:17it
30:17like a pc you know i think lioness said
30:21it
30:21you know in order for a architecture to
30:24thrive you need more developers
30:26actually using it for their daily kind
30:28of day-to-day um
30:29jobs and i think it's gonna be a very
30:33interesting milestone for risk five at
30:35sci five to market that board
30:37massively to the you know to the market
30:39for people to
30:41use it for their daily lives
30:45i mean you you guys can just imagine
30:48having hive machines built out of risk
30:50five
30:51such that you know just the whole thing
30:52makes total sense suddenly without
30:54having to learn
30:55x86 addressing modes for three weeks
30:59and yeah that's right and i mean
31:03the main kind of part of starting sci-fi
31:06for me was
31:07um when i was learning computer
31:09architecture i always
31:10kept learning something partially and
31:12then once i went to like
31:14to work at a company they would always
31:17tell me like
31:18here's the real isa it gave me like 10
31:20volumes of
31:21like books i think if we actually
31:25take risk five off the ground at the way
31:26we think we want to then
31:28whatever you learn in your 61c class
31:30would be exactly the same thing that's
31:32going to be used
31:33when you go get a job so there's no
31:36additional learning that you need
31:37you need to be doing i thought that
31:38would be really cool right and i think
31:40the moment
31:41for me um on kind of the linux
31:44thing was when their web browser came up
31:47on on risk
31:48on a on on on like a graphical screen
31:52you you know click the web browser and
31:53you could go to risk 5. and you can kind
31:55of see it
31:56actually working that was
32:00yeah that's cool um we have a couple of
32:02questions from students uh
32:03maybe there may be a question from ta so
32:06one is how do you make money off of
32:08open source hardware um it's very hard
32:11to so sci-fi sells proprietary
32:13processors as well as maintaining open
32:15source ones
32:18magical specialization yeah
32:21i see a question um i guess
32:25ask towards me when do you know when to
32:27quit an idea um that's that's a good one
32:30i actually had an internal deadline um
32:32not
32:33posed by myself but from my family
32:36saying
32:37you will you will pursue your dream
32:39until
32:40october 2015. and if it doesn't work out
32:43well
32:44you know tough love um you know it's
32:46time to move on so that was kind of my
32:49deadline and it kind of had two effects
32:51on me actually
32:52one was that well if i want to make my
32:54dream happen i kind of think it happened
32:56before october and
32:57kind of um you know i had a good you
33:00know outcome
33:01um as a deadline so i think that that
33:04that was the one that worked for me
33:06i think having a deadline um until you
33:09give it your best and
33:11you know if it doesn't work out maybe
33:12there's another chance for it it's kind
33:14of how i
33:14used to think but you know for me
33:16obviously it's easy to say because
33:18everything worked out um but you know
33:20it's hard to say
33:21it's a really tough call
33:25all right um
33:30oh to you who has been your most
33:33important mentor
33:34during undergrad
33:38um that's a very good question let me
33:40think about it um
33:42i had such great friends and professors
33:47i think it was a professor back at my
33:49undergrad who
33:51was kind of teaching things very
33:54in an interesting way where he would
33:57just
33:57keep things very open-ended and he would
34:00basically
34:02you know tell me a lot of things were
34:05not impossible to do
34:06um but i think it was really
34:10friends as you know that i work with
34:12together
34:13that you know kept doing things that
34:16seemed pretty hard
34:18and you kind of reinforce yourself that
34:21you can if you work hard you can
34:23actually make things happen
34:25um and with you know a dedicated small
34:28team
34:29with small you know smart people you can
34:31actually make things happen
34:32i think it was kind of that
34:33reinforcement mechanism that kept me
34:35going
34:36um and you know eventually say you know
34:39building a processor well you know let's
34:41see
34:41let's see whether it can be done and
34:43come to berkeley and you meet you know
34:45like people like andrew bora
34:47you know krista dave actually borah i
34:49think the one unique thing about the
34:51berkeley phd program is nobody told us
34:53that building a computer was impossible
35:01that's an excellent one any other quick
35:04questions from the others in the
35:14dantes
35:18all right uh let's thank uh yonzap and
35:22andrew um this is a i hope people
35:25understand this is a quite unique
35:27experience
35:28having two founders of this five who
35:30actually
35:31made risk by possible and founded the
35:33company that is making that that are
35:35going to make a risk by pc
35:38um and make it available to the world in
35:41a couple of months
35:42pretty crazy stuff i know if you guys
35:45feel the vibe feel the electricity and
35:48the
35:48in the air how the world is changing
35:51thanks insane andrew
35:52uh we are going to go back to our main
35:54program and deal with the rest
35:57of the stuff that we have in the agenda
35:59thanks so much
36:00thank you so much for joining us folks
36:03go bears thanks again folks that was
36:06great
36:08control yeah i'll take over the next
36:11piece of it um
36:12let's talk a little bit about the quest
36:14uh the quest happened this weekend
36:17hopefully you all knew about it you
36:19could in theory join 621c tomorrow and
36:21find out about the quest
36:23happened a day ago thank you all for the
36:25feedback uh this is our first time
36:28launching this new system prairie learn
36:30i want to certainly put a lot of
36:32love to cc who put a lot of the work
36:34into building in the background and
36:36and making sure she knew all about
36:38technically how to make it run so that
36:39all the questions we were designing
36:40together could get in there correctly
36:42and all that
36:43and i also think a lot of tas are
36:45helping with the auto grading because
36:46one of the things we did is that is
36:48complicated but we couldn't turn the
36:50grading off so you all would have
36:51clicked the button and immediately got
36:52your grade and that we didn't want that
36:54to happen so we
36:55turned that off but it meant that we
36:56couldn't use the default auto graders
36:58that are part of it
36:58we have to now extract all the data out
37:00of the system and then put it back into
37:02grade scope and then figure out how to
37:03make all that work so we're doing all
37:05that
37:05a ton of great folks are working with
37:06that on the tiers i want to thank those
37:08folks for
37:08working with that as well so i wanted to
37:12say keep it coming keep on
37:13flowing with that kind there's a whole
37:15long thread about quest timing so please
37:17do do
37:17do talk about that we do know that it
37:20was too long
37:21our estimates for how long it should
37:23have taken was was way under under
37:25under budgeted and by the way when that
37:26happens normally here's how
37:28when that happens in a real exam happens
37:29also in real exams too are you under
37:31budget how long the exam should take
37:33what typically happens is
37:34people can't finish and it's really
37:36unfair um so i still think that this
37:38unlimited i mean 24 hour time model is a
37:41better model
37:41even though we'll try to do better not
37:43to say we're going to keep aiming for a
37:44seven hour exam
37:45let's always go back and try to aim for
37:46one hour exam but that gave us a little
37:48bit of breathing room that we normally
37:49wouldn't have had i've had final exams
37:51that were too long
37:52and people were just angry because they
37:53said i know the how to answer question
37:55seven
37:55i just couldn't get to it and so that's
37:57really unfair didn't happen this time so
37:59i do think that this is a better system
38:01taking this thing as homes and we're
38:02gonna try to do better in terms of time
38:03scales
38:04our expectations also need to be better
38:06in terms of making sure you'll get a
38:07sense of
38:08a practice exam being kind of the same
38:10topic and range
38:11uh we did have a you know coding
38:12question the practice exam but people
38:14didn't somehow didn't think that that
38:15was going to be
38:15what you'd be doing for the final one
38:17you know that was kind of a little bit
38:18of a taste of that
38:18we made it a little bit harder than that
38:20but we'll try to have a similar thing so
38:22that
38:22you'll get a sense of roughly the
38:23difficulty and roughly the kind of
38:24problem so maybe we'll write two
38:25problems and give one away and then
38:26we'll
38:27actually ask the second one that are
38:28obviously different topics not exactly
38:30the same problem
38:31um because i mentioned the technical
38:34issues of taking all the
38:35all of your i mean it's a massive file
38:37of of the csv of every submission you've
38:40given not the video part but just the
38:41data from the
38:42from the prairieland and getting it out
38:44and putting into prairieland and then
38:45hacking on that to get the great grades
38:47done
38:47we're going to try to get that done by
38:48tuesday night all of our tas are working
38:51overtime to make this happen and i want
38:52to thank them for again for that work
38:53that they're doing
38:54uh and we're hopefully tuesday night is
38:55the deadline the worst case wednesday
38:57morning but
38:57that's the hard deadline you guys need
38:58to know whether to drop the class or not
39:00i don't want anybody to drop the class
39:03based on bombing the quest
39:04i mean chalk it up to bad expectations
39:07not telling you how much it took if you
39:09started the exam with two hours to go
39:10thinking yeah the game has said an hour
39:12that it's not fair because you probably
39:14needed seven hours or four hours or two
39:15a lot of people did less and and if you
39:18knew you see or
39:19you know more flu and that kind of thing
39:20you probably took less time and
39:23we've learned all about those things and
39:24so uh what we're going to do
39:27get the grades back to you as soon as we
39:28can wednesday morning i want to
39:29encourage you all to stay with the class
39:32okay all to stay with the class i
39:34believe
39:35because we have in yellow a full clobber
39:38available
39:40almost no other class offers it almost
39:42no other class offers it
39:43so what that means is that times two
39:46times yeah
39:47double clobber you can even clobber from
39:49the final
39:50to the quest or from the midterm to the
39:52quest so please stay with it we we want
39:55us to we want you to lose you guys we
39:56want you to stay with the class you put
39:57a lot of work into it
39:58it's a great class we put a lot of you
40:00know effort into hiring great people
40:02stay with this if you can i mean
40:04obviously if you're finally okay you
40:05know what this quest is one data point
40:07to tell me that i can't manage the
40:08semester
40:09we understand if you drop it that case
40:10but don't take a bad score if you you
40:12know had enough time to study for it
40:14uh and you think you can work to fit the
40:16workload in with your schedule please
40:17don't drop it
40:18because of that because we'll give you a
40:19chance to overwrite that grade
40:20completely uh 100
40:22not some people actually only got a
40:24partial credit a clubber where like
40:26you have to own 25 of it you can clap
40:28for 75.
40:29now we are certainly uh doing this uh
40:32doing this
40:33oh look at this i didn't know this i'm
40:34seeing some everybody gets an extra
40:36emergency drop so even there you can use
40:38the emergency drop you get one free one
40:40one free get out of jail card that's get
40:41a jail free card um
40:44so that's great uh so that's the full
40:46clubber um we will
40:48there are a lot of questions about you
40:49know how do i know when i need to retake
40:50it
40:51we're all just trying to survive this
40:53week so we're going to try to if we can
40:54get the grades back to you by wednesday
40:56i'll consider the week a success
40:58the week starting monday not starting on
41:00sunday because we we
41:01did learn about the quest but we're
41:03going to start monday get the grades
41:04back to you then phew you all make a
41:05decision get
41:06there there's obviously people can be
41:07unhappy with this um
41:10we are not the current planners we are
41:12not going to look at code if you
41:13couldn't get it to compile
41:14it might be 99 there but if you couldn't
41:17get to compile we're not going to give
41:18you points for those things
41:19it has to compile but we're going to
41:20give a lot of partial credit for all the
41:22pieces if you only turned a bit on
41:24but couldn't rotate right or couldn't
41:25turn a bit off we'll give you points for
41:27turning the bit on
41:28if you only turn it off we'll give you
41:29points for that we only rotate it
41:30without turning on and off
41:31like so we'll try to do that part of it
41:33give you that some part a partial if you
41:35couldn't get the whole problem done
41:36similar to the other part if you've got
41:37some of it working if you didn't do any
41:39malice if you didn't do the freeze
41:40you have supposed to free it and then
41:42move it over and do the split
41:43let's see did this put without the
41:44without the freeing well we give you
41:46some partial for that so there's partial
41:47for all the pieces of it as much as we
41:48can
41:49but if it doesn't work we're not going
41:50to give partial credit that's kind of
41:51how this works we give you enough time
41:52to get it working and if you can get it
41:54working we're sorry that's the current
41:55plan
41:57now here we go what's going to happen is
42:01next week we're going to take a look at
42:03how the grades looked
42:04look at what all the technical issues
42:06were there's a ton boy i i've never been
42:08on piazza for six hours straight like i
42:10was
42:10all of us were just on it trying to
42:12answer questions as fast as we could
42:14um try to give you that feedback um so
42:17what we're going to do is take a look at
42:18who needs it based on how
42:20you know if your camera dropped off for
42:2220 seconds we're not going to make you
42:23retake it okay if you
42:24drop off for 30 seconds you know so
42:26we'll look at how the camera drop-off
42:28was all the issues every issue needs to
42:30be looked at one by one and that takes a
42:31lot of time that's a thousand twelve
42:33hundred people
42:33when you look at all the and uh boys
42:35some people have technical issues well
42:36my
42:37thing was i didn't run out of space so i
42:38lost the last minute you know what
42:40you did you did fine last minute that's
42:42okay whatever we'll look at the numbers
42:43we'll look how you did
42:44we'll decide whether you know to
42:46basically lose the berkeley you all
42:47signed the book the honor code you all
42:49said hours and cheating
42:50well to make a look at how you do how if
42:52you had no video we're probably going to
42:53ask you to
42:54retake it but if you only drop for a
42:55little bit we can maybe work out and
42:56talk about it
42:57and maybe maybe just say it will be fine
42:59we'll take your grade okay so we want to
43:00ask
43:01everybody but we may ask for some of you
43:02to do a retake
43:04all the details we told next week we'll
43:05announce this next monday we'll announce
43:07how that's going to work
43:08we're going to have different problems
43:09we're going to give you not exactly the
43:11same problem different
43:12problems similar and then go it's just
43:14some amazing amount
43:15i had by the way how much footage i have
43:17i had the last second cc knows this i
43:18clicked the button on the forum
43:20to allow for one terabyte on that form
43:22because it was like a hundred gigs
43:24as another gigs no that's two uh how do
43:26you don't know let's make our tera
43:27literally is the biggest number i could
43:29choose on the silly form there's a
43:30terabyte of data now and by the way
43:32my computer started to hum i was like
43:34why is my computer because my computer
43:35synced that stupid folder so my google
43:37drive was downloading every
43:40video as i'm ah that says free
43:43why is my free space going from 200 gigs
43:45to zero
43:46so i unsynced it but it's there and it's
43:48crazy so anyway summary is left
43:50never exactly my ssd is in paid uh so
43:53we're gonna um
43:54free snt no don't free ssd
43:57rm minus rf star okay no so long story
44:01short
44:02um we are gonna let you know next week
44:03okay and that's the idea
44:05now you ready for the next one the next
44:07one are you ready for it for the next
44:08one i click the next slide here we go
44:10the next slide is that the drum roll
44:16hey we're giving you an extension of the
44:20project
44:23so basically a lot of people said and we
44:25heard you guys we are hearing you i
44:27can't think of a course or we're not
44:28listening more than we are we're
44:29listening everything you think you say
44:31you're saying dan i expected to spend
44:32one hour on the exam
44:34i spent seven hours that was the time i
44:36would have done my project
44:38we were gonna and all of us don't just
44:39thank me thank the whole team remember
44:41the whole team
44:42thought about this and made this call uh
44:44all of us heads top tiers
44:46everybody was bartered at this so long
44:49story short you get
44:50one extra day on 1a and 1b but no else
44:53okay only one that that just give you
44:56one day
44:57to kind of make up for that but i mean
44:59it's
45:00look folks what's my most most famous
45:02phrase from my number representation
45:04come on everybody type it famous phrase
45:05four words
45:07ain't no free lunch what it means is if
45:10i
45:10give you a day on this it means you have
45:13less days on the stuff after it okay i'm
45:15sorry about that but that's how it works
45:16we got we can't be slipping too much
45:18ain't no free lunch over here so just
45:20let you know that's what we got
45:22so don't don't don't don't like oh i'll
45:25figure that but let me
45:26keep going you got to keep pushing this
45:27is hard by the way 61c this
45:29is the hard weeks am i bored give me
45:31some love this is exactly the hardest
45:33weeks of this
45:34if you can survive these weeks where we
45:35throw sea at you and throw risk at you
45:37and then
45:38the project one in the project it's like
45:39literally back to back it br you get to
45:41breathe more
45:42it's really hard to have any time to
45:43breathe here this is just a push push
45:45through this
45:46my wife i have two kids it's like labor
45:49you gotta just
45:50push through push push and then you can
45:52get to the breathing then breathe
45:53breathe
45:55push and then breathe so this is a push
45:57time four
45:59yeah go i mean if you listen to just dan
46:01speaking now
46:02that's how he speaks in lectures he's
46:04about 20 faster than the average i'm
46:06about 20
46:07slower than the average so you actually
46:09get a breather this week
46:14so if you watch lectures at half speed
46:17or double speed and you can adjust that
46:19depending on what you're exactly yeah
46:22oh goodness we are learning i mean we're
46:24learning how to
46:25we haven't done a class like this yeah
46:28this is all of us i've never done this
46:29for a while
46:30exactly and we're learning this and you
46:32know you're going to adjust for the for
46:33limiter we're going to see a little bit
46:35different
46:36style of homeworks that are going to
46:38help you prepare you know so you don't
46:39get the surprise
46:41of how does the you know the style of a
46:44question
46:45um and then we'll practice a little bit
46:47more and
46:49hopefully the we're going to take out
46:52the surprise
46:53effect element from the exactly from the
46:55from the midterm and the final
46:57so your knowledge yeah exactly exactly
47:00and that's basically it but yes this
47:02i mean week two uh is the toughest i
47:05mean this is the fastest
47:06quickest store to see and if you
47:10if you got it you got it if not i you
47:13get more time to to nail it down to
47:16to basically code and see that's a
47:19really
47:19that thing is going to stay forever i
47:21wrote my masters and see
47:23i lost my speed i mean i lost i wrote
47:26twenty thousand lines i'm sticking i
47:28wrote my master's in c2 i brought my
47:30masters into
47:30two i love it it's the old school old
47:32school i wrote a power line power line
47:34modem
47:35uh in c
47:38anyway back to that okay yeah yeah yeah
47:40well we have about five minutes or four
47:42minutes three minutes left now look at
47:43the time
47:43okay so let's quickly go to what those
47:45calendar before you want to take
47:46ourselves for the calendar what we're
47:47looking at here
47:48we got a couple of projects and stuff
47:49yeah we have a
47:51yeah we have project due dates i can't
47:53see this uh
47:55well over there so we have a couple of
47:57project dues uh
47:58do we have a homework and i think uh
48:00there is a
48:01floating point homework that's out there
48:03if you're going to release
48:05a risk five homework um relatively early
48:08this week
48:09um so you can get going if as soon as
48:12you finish
48:12um floating point this is hopefully a
48:14little bit later week
48:16conceptually is i would think it's
48:18easier risk 5 is a little easier but
48:20it's a
48:21completely different concept so you're
48:23just gone
48:24you know from c dove into assembly
48:28some people might have seen c before but
48:31nobody
48:32probably has seen assembly so it's a
48:33different concept we are going a little
48:35bit slower
48:37and you're going to pick up a bit of a
48:38speed uh however things when you see how
48:41does the memory system really work
48:43in assembly that will help really
48:46understand how does you know why does
48:47the sea do things
48:49see do things in a particular way all
48:51these
48:53mallocs and things like those are going
48:55to suddenly make sense
48:57sure and we can look forward to next
48:59week's calendar as well
49:00so just to give you this is we're in
49:01week four and week five we can see
49:03so this is all the week of risk five
49:05we're deep into risk five folks it's
49:07five
49:07straight lectures of and now we already
49:10had one on friday so we've got
49:11you know load word story decisions
49:13procedures
49:14instruction formats so now so this week
49:16is kind of how to write risk five
49:18so you're there and then we lift the
49:20hood next week which is exciting and
49:21boris covered these lectures i can't
49:22wait to see them
49:23which is how what does it actually mean
49:25to look at the ones and zeros that what
49:27these each of these mean so look at the
49:29format of each of these guys how do you
49:30translate it from
49:31assembly code to machine code i'm going
49:33to come back on friday of next week and
49:35do a compiling assembly linking and
49:37loading and then we're going to
49:38do the low level we'll jump all the way
49:39down to the low-level stuff after that
49:41so we're looking forward to that being
49:42kind of a break
49:43kind of a breather the first five weeks
49:44a third of the class is done and week
49:46six we get to the low level stuff which
49:48is really really fun too
49:49so i'll come back in like at week six
49:52and again i mean
49:53if you watch me i can't match that i
49:56mean there is
50:01[Laughter]
50:06but i i think i tried to really
50:10go relatively slowly and thoroughly
50:11through the concepts and
50:13they're there they're in a fairly
50:15compact form uh we're gonna release the
50:17rest of the videos tomorrow for this
50:19week
50:20and then the next week is also going to
50:22be is
50:23kind of a bit boring about uh how the do
50:26we translate assembly to machine code
50:28but then you will see you know some of
50:30these things that andrew mentioned about
50:32the rats nest or
50:33young suspension of transmission
50:35immediates
50:38we get to the details details exactly by
50:40the way i want to say
50:42go ahead bart go ahead well i just want
50:45to say
50:45you don't realize everybody doesn't
50:47realize how lucky you are
50:49borah helped make risk five happen do
50:52you realize this you're learning about
50:53risk five from the guy that made it
50:54happen that's very exciting i
50:56i had nothing to do with it i'm just
50:57writing on the coattails of him
50:59so you guys please give that love that
51:01way because that's
51:02great i'm so pleased i've learned to be
51:04honest what did i learn from risk five i
51:05learned from borah okay if you ask me a
51:07question i'll say what did borah teach
51:08me last year
51:09that's what it is so borah is the
51:11awesome guy in the risk five
51:12conversation so i love it it's great
51:14stuff
51:14but a bigger biggest thing would be to
51:17get
51:17you know uh dan record all of these
51:20videos eventually no no
51:21no no no no keep up keep it up
51:25keep it up you guys are awesome a lot of
51:26love both ways both ways good stuff
51:29we'll try to bring also very interesting
51:31people i mean one thing that we have we
51:33have both we have connections to this
51:35industry and that there is an
51:36opportunity here to bring joseph and
51:37andrew and
51:38some other people that there is there is
51:40no way
51:41to bring life into a lecture so i think
51:44stay tuned for the next week's lecture
51:46we're going to be lucky this is amazing
51:48and borah knows a lot of people his
51:50list of people he's going to contact is
51:51much bigger than a list of people i
51:52would contact so we're going to have a
51:53lot of really great people
51:54all relevant to 621c in some way that's
51:56the idea so it's just great so so i'm
51:58really thankful for all the work you put
52:00into to get the guest
52:01and today was great i really appreciate
52:02meeting those two guys i had met them
52:03before that was nice
52:05good stuff all right so
52:08that's it i think we try to answer most
52:10of these guys in the future
52:12uh partial credit we're again we're not
52:14gonna do partial credit we're gonna pass
52:15your credit on the
52:16auto grading kind of stuff but not
52:17looking at code and next time if you
52:19need to go to the bathroom that's why i
52:20broke it up into pieces
52:21hit the restroom and then jump in and
52:23lock in there and then hit the restroom
52:24again for pieces so we'll break up an
52:26example a lot more pieces like that and
52:27we'll do that next time all right
52:28thanks everybody we'll see you next week
52:34thanks everybody we got 200 people all
52:36right good stuff
52:38hey we got 120 at some point you did you
52:40got me on this i can't believe i was
52:42i was thinking we'd have four cc
52:47you win the price is right conversation
52:49good stuff see you guys we'll see you
52:50next week all right
52:52thanks folks okay bye

About this transcript

This page contains the full transcript of [CS61C FA20] Weekly Lecture 04.LIVE - RISC-V by CS 61C Departmental, generated from the public captions YouTube serves with the video. The transcript has 9,882 words across 1,623 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.