Kind: captions
Language: en
the following is a conversation with
David Patterson Turing Award winner and
professor of computer science at
Berkeley he's known for pioneering
contributions to RISC processor
architecture used by 99% of new chips
today and for co-creating RAID storage
the impact that these two lines of
research and development have had in our
world is immeasurable he's also one of
the great educators of computer science
in the world his book with John Hennessy
is how I first learned about and was
humbled by the inner workings of
machines at the lowest level quick
summary of the ads to sponsors the
Jordan Harbinger show and cash app
please consider supporting the podcast
by going to Jordan Harbinger complex and
downloading cash app and using code
Lexx podcast click on the links buy the
stuff it's the best way to support this
podcast and in general the journey I'm
on in my research and startup this is
the artificial intelligence podcast if
you enjoy it
subscribe on YouTube review it five
stars in hype a podcast supported on
patreon or connect with me on Twitter
and Lex Friedman spelled without the e
just Fri DM a.m. as usual I'll do a few
minutes of ads now and never any ads in
the middle that can break the flow of
the conversation
this episode is supported by the Jordan
Harbinger show go to Jordan Harbinger
calm / Lex
it's how he knows I set you on that page
there's links to subscribe to it an
apple podcast Spotify and everywhere
else I've been binging on this podcast
it's amazing Jordan is a great human
being he gets the best out of his guests
- deep calls him out when it's needed it
makes the whole thing fun to listen to
he's interviewed Kobe Bryant Mark Cuban
and Neil deGrasse Tyson and Garry
Kasparov and many more I recently
listened to his conversation with Frank
Abagnale author of catch me if you can
one of the world's most famous Kahneman
perfect podcast length and topic for a
recent long distance run that I did
go to Jordan Harbinger complex to give
him my love and to support this podcast
subscribe also on Apple podcast Spotify
and everywhere else this show is
presented by cash app the greatest
sponsor of this podcast ever and the
number one finance app in the App Store
when you get a used coat Lex podcast
cash app lets you send money to friends
buy bitcoin invest in the stock market
with as little as one dollar since gas
rep allows you to buy bitcoin let me
mention that cryptocurrency in the
context of the history of money is
fascinating I recommend the scent of
money as a great book on this history
also the audio book is amazing debits
and credits on Ledger's started around
30,000 years ago the US dollar created
over two hundred years ago and the first
decentralized cryptocurrency released
just over ten years ago so given that
history cryptocurrencies still very much
in its early days of development but
it's still aiming to and just might
redefine the nature of money so again if
you get cash out from the App Store
Google Play and use the code Lex podcast
you get ten dollars and cash up will
also donate ten dollars to first an
organization that is helping to advance
robotics to stem education for young
people around the world and now here's
my conversation with David Patterson
let's start with the big historical
question how have computers changed in
the past 50 years at both the
fundamental architectural level and in
general in your eyes well the biggest
thing that happened was the invention of
the microprocessor so computers that
used to fill up several rooms could fit
inside your cell phone and not only and
how do they get smaller they got a lot
faster so they're million times faster
than they were 50 years ago and they're
much cheaper and they're RIBA covetous
you know there's seven point eight
billion people on this planet probably
half of them have cell phones but you
know just remarkable
it's probably more micro processors than
there are people sure I don't know what
the ratio is but I'm sure it's above one
maybe it's ten to one or some number
like that what is a microprocessor so a
way to say what a microprocessor is to
tell you what's inside a computer so a
computer forever has classically had
five pieces there's input and output
which kind of naturally as you'd expect
is input is like speech or typing and
output is displays there's a memory and
like the name sounds it it remembers
things so it's integrated circuits whose
job is you put information in and when
you ask for it it comes back out that's
memory and the third part is the
processor where the team microprocessor
comes from and that has two pieces as
well and that is the control which is
kind of the brain of the processor and
the what's called the arithmetic unit
it's kind of the Brawn of the computer
so if you think of the as a human body
the arithmetic unit the thing that does
the number crunching is the is the body
and the control is the brain so those
five pieces input/output memory
arithmetic unit and control are have
been in computers since the very dawn in
the last two are considered the
processor so a microprocessor simply
means a process of the fits on a
microchip and that was invented at about
you know 40 years ago was the first
microprocessor it's interesting that you
refer to the arithmetic unit as the like
he connected to the body and the
controller's of the brain so I guess I
never thought of it that was a nice way
to think of it because most of the
actions the microprocessor does in terms
of literally sort of computation with
the microprocessor does computation it
processes information and most of the
thing it does is basically earth net
arithmetic operations what what are the
operations by the way it's a lot like a
calculator you know so there are add
instructions a subtractive Stressless
multiply and divide and
kind of the brilliance of the invention
of the my computer or the processor is
that it performs very trivial operations
but it just performs billions of them
per second and what we're capable of
doing is writing software that can take
these very trivial instructions and have
them create tasks that can do things
better than human beings can do today
just looking back through your career
did you anticipate the kind of how good
we would be able to get at doing these
small basic operations I think what how
many surprises along the way we just
kind of set back and said wow I didn't
expect it to go this fast this good well
the the fundamental driving force is
what scored Moore's law which was named
after Gordon Moore who's a Berkeley
alumnus and he made this observation
very early in what are called semi
conductors and semiconductors are these
ideas you can build these very simple
switches and you can put them on these
microchips and he made his observation
over 50 years ago he looked at a few
years and said I think what's going to
happen is the number of these little
switches called transistors is going to
double every year for the next decade
and he said this in 1965 and in 1975 he
said well maybe he's gonna double every
two years and that I would other people
since named that Moore's Law guided the
industry and when Gordon Moore makes
that prediction he he wrote a paper back
in I think in the in the 70s and said
not only did this going to happen he
wrote what would be the implications of
that and in this article from 1965 he he
shows ideas like computers being in cars
and computers being in something that
you would buy in the grocery store and
stuff like that so he kind of not only
called his shot he called the
implications of it so if you were in in
the computing field and a few believed
Moore's prediction he kind of said what
the what would be happening in the
future
so so it's not kind of
it's at one sense this is what was
predicted and you could imagine it was
easily believed that Moore's law was
going to continue and so this would be
the implications on the other side there
are these shocking events in your life
like I remember driving in meriem across
the bay in San Francisco and seeing a
bulletin board at a local Civic Center
and had a URL on it uh and it was like
if for all for all that's for the people
at the time these first URLs and that's
the you know ww select stuff with the
HTTP people thought it was looks like
alien alien writing right they'd see
these advertisements and commercials or
bulletin boards that had this alien
writing on it so for the lay people is
like what the hell is going on here and
for those people interesting it's oh my
god this stuff is getting so popular
it's actually leaking out of our nerdy
world and into the real world so that I
mean there is events like that I think
another one was I member with the in the
early days of the personal computer when
we started seeing advertisements in
magazines for personal computers like
it's so popular that it's it made the
newspapers so at one hands you know
Gordon Moore predicted it and you kind
of expected it to happen but when it
really hit and you saw it affecting
society it was it was shocking so maybe
taking a step back and looking both the
engineering and philosophical
perspective what what do you see as the
layers of abstraction in the computer do
you see a computer as a set of layers of
abstractions yeah and I think that's one
of the things that computer science
fundamentals is the these things are
really complicated in the way we cope
with complicated software and
complicated hardware is these layers of
abstraction and that simply means that
we you know suspend disbelief and
pretend that the only thing you know is
that layer and you don't know anything
about the layer below it and that's the
way we can make very complicated things
and probably it started with hardware
that's the way it was done but it's been
proven extremely useful and
you know I would think in a modern
computer today there might be 10 or 20
layers of abstraction and they're all
trying to kind of enforce this contract
is all you know is this interface
there's a set of commands that you can
allow to use and you stick to those
commands that we will faithfully execute
that and it's like peeling the air
layers of a London onion you get down
there's a new set of layers and so forth
so for people who want to study computer
science the exciting part about it is
you can keep peeling those layers you
you take your first course and you might
learn to program in Python and then you
can take a follow-on course and you can
get it down to a lower level language
like C and you know you can go and you
can if you want to you can start getting
into the hardware layers and you keep
getting down all the way to that
transistor that I talked about that
Gordon Moore predicted and you can
understand all those layers all the way
up to the highest level application
software so it's it's a very kind of
magnetic field if you're interested you
can go into any depth and keep going in
particular what's happening right now or
it's happened in software last twenty
years and recently in hardware there's
getting to be open sourced versions of
all of these things so what open source
means is what the engineer the
programmer designs it's not secret the
belonging to a company it's up there on
the World Wide Web so you can see it so
you can look at for lots of pieces of
software that you use you can see
exactly what the programmer does if you
want to get involved that used to stop
at the hardware recently there's been an
efforts to make open-source hardware and
those interfaces open so you can see
that so instead of before you had to
stop at the hardware you can now start
going layer by layer below that and see
what's inside there so it's it's a
remarkable time that for the interested
individual can really see in great depth
what's really going on and the computers
that power everything that we see around
us are you thinking also
when you say open source at the hardware
level is this going to the design
architecture instruction set level or is
it going to literally the the you know
the manufacturer of the of the actual
hardware of the actual chips whether
that's a six specialized a particular
domain or the general yeah so let's talk
about that a little bit so when you get
down to the bottom layer of software the
way software talks to hardware is in a
vocabulary and what we call that
vocabulary we call that the words of
that vocabulary called instructions in
the technical term for the vocabulary is
instruction set
so those instructions are likely we
talked about earlier that can be
instructions like add subtract and
multiply divide there's instructions to
put data into memory which is called a
store instruction and to get data back
which is called the load instructions
and those simple instructions go back to
the very dawn of computing in you know
in 1950 the commercial commercial
computer had these instructions so
that's the instruction set that we're
talking about so up until I'd say ten
years ago these instruction sets are all
proprietary so a very popular one is
Alden by Intel the one that's in the
cloud and then all the pcs in the world
the Intel owns that instruction set it's
referred to as the x86 there have been a
sequence of ones that the first number
was called 8086
and since then there's been a lot of
numbers but they all end in 86 so
there's then that kind of family of
instruction sets and that's proprietary
and that's proprietary the other one
that's very popular is from arm that
kind of powers all of all the cell
phones in the world all the iPads in the
world and a lot of things that are
so-called Internet of Things devices arm
and that one is also proprietary arm
will license it to people for a fee but
they own that so the new idea that got
started at Berkeley kind of
unintentionally ten years ago is
in early in my career we pioneered a way
to do of these vocabularies instruction
sets that was very controversial at the
time at the time in the 1980s
conventional wisdom was these
vocabularies instruction sets should
have you know powerful instructions so
polysyllabic kind of words you can think
of that and and so that instead of just
add subtract and multiply they would
have polynomial vied or sort a list and
the hope was of those powerful
vocabularies that make it easier for
software so we thought that didn't make
sense for microprocessors servers people
at Berkeley and Stanford and IBM who
argued the opposite and we what we
called that was a reduced instruction
set computer in the abbreviation was our
ISC and typical for computer people we
use the abbreviations start pronouncing
it so risk was there so we said for
microprocessors which with Gordon's
Moore is changing really fast we think
it's better to have a pretty simple set
of instructions reduce set of
instructions that that would be a better
way to build microprocessors since
they're going to be changing so fast due
to Moore's law and then we'll just use
standard software to cover the used
generate more of those simple
instructions and one of the pieces of
software that it's in a software stack
going between these layers of
abstractions is called a compiler and it
basically translates it's a translator
between levels we said the translator
will handle it so the technical question
was well since there are these reduced
instructions you have to execute more of
them yeah that's right but maybe they
execute them faster yeah that's right
there's simpler so they could go faster
but you have to do more of them so
what's what's that trade-off look like
and it ended up that we ended up
executing maybe 50 percent more
instructions maybe 1/3 more instructions
but they ran four times faster so so
this risk controversial risk ideas
proved to be maybe factors of three or
four better I love that this idea was
controversial and
most kind of like a rebellious so that's
in the context of what was more
conventional is the complex instruction
set competing so how'd you pronounce
that Sisk Sisk risk vs. Sisk and and
believe it or not this sounds very very
you know who cares about this right it
was it was violently debated at several
conferences it's like what's the
brightman ago is is and people thought
risk was you know was de-evolution we're
gonna make software worse by making
death instructions simpler and they're
fierce debates at several conferences in
the 1980s and then later in the eighties
that kind of settled to these benefits
it's not completely intuitive to me why
risk has for the most part one yes so
why do that happen yeah yeah and maybe I
can sort of say a bunch of dumb things
that could lay the land for further
commentary so to me and this is a this
is kind of interesting thing if you look
at C++ was just see with modern
compilers you really could write faster
code with C++ so relying on the compiler
to reduce your complicated code into
something simple and fast so to me
comparing risk maybe this is a dumb
question but why is it that focusing the
definition the design of the instruction
set on very few simple instructions in
the long run provide faster execution
versus coming up with like I said a ton
of complicated instructions then over
time you know years maybe decades you
come up with compilers that can reduce
those into simple instructions for you
yeah some let's try and split that into
two pieces so if the compiler can do
that for you if the pilot can take you
know a complicated program and produce
simpler instructions then the programmer
doesn't care right programmer yeah yeah
I don't care just how how fast is the
computer I'm using how much does it cost
and so what we what
and kind of in the software industry is
right around before the 1980s critical
pieces of software we're still written
not in languages like C or C++ they were
written in what's called assembly
language where there's this kind of
humans writing exactly at the
instructions at the level then that a
computer can understand so they were
writing add subtract multiply you know
instructions it's very tedious but the
belief was to write this lowest level of
software that the people use which are
called operating systems they had to be
written in assembly language because
these high-level languages were just too
inefficient they were too slow or the
the programs would be too big so that
changed with a famous operating system
called UNIX which is kind of the
grandfather of all the operating systems
today so the UNIX demonstrated that you
could write something as complicated as
an operating system in a language like C
so once that was true then that meant we
could hide the instruction set from the
programmer and so that meant then it
didn't really matter the programmer
didn't have to write lots of these
simple instructions so that was up to
the compiler so that was part of our
arguments for risk is if you were still
writing assembly languages maybe a
better case for sis constructions but if
the compiler can do that it's gonna be
you know that's done once the computer
translates it once and then every time
you run the program it runs that this
this potentially simpler instructions
and so that that was the debate right is
because and people would acknowledge
that these simpler instructions could
lead to a faster computer you can think
of mono syllabic constructions you could
say them you know if you think of
reading you probably read them faster or
say them faster than long instructions
the same thing that analogy works pretty
well for hardware and as long as you
didn't have to read a lot more of those
instructions you could win so that's
that's kind of that's the basic idea for
risk but it's interesting that the in
that discussion of UNIX to see that
there's only one step
of levels of abstraction from the code
that's really the closest to the machine
to the code that's written by human it's
uh at least to me again perhaps a dumb
intuition but it feels like there might
have been more layers sort of different
kinds of humans stacked as well of each
other so what's true and not true about
what you said is several of the layers
of software like so the if you hear two
layers would be suppose we just talked
about two layers that would be the
operating system like you get from from
Microsoft or from Apple like iOS or the
Windows operating system and let's say
applications that run on top of it like
Word or Excel so both the operating
system could be written in C and the
application could be written in C so but
you could construct those two layers and
the applications absolutely do call upon
the operating system and the change was
that both of them could be written in
higher-level languages so it's one step
of a translation but you can still build
many layers of abstraction of software
on top of that and that's how how things
are done today so still today many of
the layers that you'll you'll deep deal
with you may deal with debuggers you may
deal with linkers there's libraries many
of those today will be written in c++
say even though that language is pretty
ancient and even the Python interpreter
is probably written in C or C++ so lots
of layers there are probably written in
these some old fashioned efficient
languages that still take one step to
produce these instructions produce RISC
instructions but they're composed each
layer of software invokes one another
through these interfaces and you can get
ten layers of software that way so in
general the risk was developed here
Berkeley it was kind of the three places
that were
these radicals that advocated for this
against the rest of community where IBM
Berkeley and Stanford you're one of
these radicals and how radical did you
feel how confident did you feel how
doubtful were you that risk might be the
right approach because it may you can
also Intuit that is kind of taking a
step back into simplicity not forward
into simplicity yeah no it was easy to
make yeah it was easy to make the
argument against it well this was my
colleague John Hennessy at Stanford and
I we were both assistant professors and
for me I just believed in the power of
our ideas I thought what we were saying
made sense Moore's Law is going to move
fast the other thing that I didn't
mention is one of the surprises of these
complex instruction sets you could
certainly write these complex
instructions if the programmer is
writing them in themselves it turned out
to be kind of difficult for the compiler
to generate those complex instructions
kind of ironically you'd have to find
the right circumstances that that just
exactly fit this complex instruction it
was actually easier for the compiler to
generate these simple instructions so
not only did these complex instructions
make the hard work more difficult to
build often the compiler wouldn't even
use them and so it's harder to build the
compiler doesn't use them that much the
simple instructions go better with
Moore's Law that's you know the number
of transistors is doubling every every
two years so we're gonna have you know
the you want to reduce the time to
design the microprocessor that may be
more important than these number
instructions so I think we believed in
the that we were right that this was the
best idea then the question became in
these debates well yeah that's a good
technical idea but in the business world
this doesn't matter there's other things
that matter it's like arguing that if
there's a standard with the railroad
tracks and you've come up with a better
with but the whole world has covered
railroad tracks so you'll your ideas
have no chance of success
commercial success it was technically
right but commercially it'll be
insignificant yeah this it's kind of sad
that this world the history of human
civilization is full of good ideas that
lost because somebody else came along
first with a worse idea and it's good
that in the computing world at least
some of these have well well you could
are I mean it's probably still sisk
people that say yeah still are but and
what happened was what was interesting
Intel a bunch of the system companies
with Sisk instruction sets of vocabulary
they gave up but not Intel what Intel
did to its credit because Intel's
vocabulary was in the in the personal
computer and so that was a very valuable
vocabulary because the way we distribute
software is in those actual instructions
it's in the instructions of that
instruction set so they then you don't
get that source code what the
programmers wrote you get after it's
been translated into the last level
that's if you were to get a floppy disk
or download software it's in the
instructions that instruction set so the
x86 instruction set was very valuable so
what Intel did cleverly and amazingly is
they had their chips in hardware do a
translation step they would take these
complex instructions and translate them
into essentially in RISC instructions in
Hardware on the fly you know at at
gigahertz clock speeds and then any good
idea that risk people had they could use
and they could still be compatible with
us with this really valuable PC software
software base and which also had very
high volumes you know a hundred million
personal computers per year so the sisk
architecture in the business world was
actually one in in this PC era so just
going back to the the time of designing
risk when you design an instruction set
architecture do you think like a
programmer do you think like a
microprocessor engineer do you think
like a artist a philosopher do you think
in software and hardware I mean is it
art I see science yeah I'd say I think
designing a goods instruction set as an
art and I think you're trying to balance
the the simplicity and speed of
execution with how well easy it will be
for compilers to use it alright you're
trying to create an instruction set that
everything in there can be used by
compilers there's not things that are
missing
that'll make it difficult for the
program to run they run efficiently but
you want it to be easy to build as well
so it's that kind of so you're thinking
I'd say you're thinking hard we're
trying to find a hardware software
compromise that'll work well and and
it's you know it's you know it's a
matter of taste right it's it's kind of
fun to build instruction sets it's not
that hard to build an instruction set
but to build one that catches on and
people use you know you have to be you
know fortunate to be the right place at
the right time or have a design that
people really like are using metrics
says is it quantifiable because you kind
of have to anticipate the kind of
programs that people will write yet
ahead of time so is that can you use
numbers can use metrics can you quantify
something ahead of time or is this again
that's the art part where you're kind of
knows it's a a big a big change kind of
what happened I think from Hennessey's
and my perspective in the 1980s what
happened was going from kind of really
you know taste and hunches to
quantifiable in in fact he and I wrote a
textbook at the end of the 1980s called
computer architecture a quantitative
approach I heard of that and and it's
it's the thing it it had a pretty big
big impact in the field because we went
from textbooks that kind of listed so
here's what this computer does and
here's the pros and cons and here's what
this computer doesn't pros and cons to
something where there were formulas
in equations where you could measure
things so specifically for instruction
sets what we do in some other fields do
is we agree upon a set of programs which
we call benchmarks and a suite of
programs and then you develop both the
hardware and the compiler and you get
numbers on how well your your computer
does given its instruction set and how
well you implemented it in your
microprocessor and how good your
compilers are and in computer
architecture we you know using
professors terms we grade on a curve
rather than greater than absolute scale
so when you say you know this these
programs run this fast well that's kind
of interesting but how do you know it's
better while you compare it to other
computers at the same time so the best
way we know how to make turned it into a
kind of more science and experimental
and quantitative is to compare yourself
to other computers or the same era that
have the same access the same kind of
technology on commonly agreed benchmark
programs so maybe two toss-up two
possible directions we can go one is
what are the different trade-offs in
designing architectures Ubben are you
talking about Siskin risk but maybe a
little bit more detail in terms of
specific features that you were thinking
about and the other side is what are the
metrics that you're thinking about when
looking at these trade-offs yeah well
let's talk about the metrics so during
these debates we actually had kind of a
hard time explaining convincing people
the ideas and partly we didn't have a
formula to explain it and a few years
into it we hit upon the formula that
helped explain what was going on and I
think if we can do this see how it works
orally just is this so the yes if I can
do a formula or Li L C so the so
fundamentally the way you measure
performance is how long does it take a
program to run a program if you have ten
programs and typically these benchmarks
were sweet because you'd want to have
ten programs so they could represents
lots of different applications so for
these ten programs how long they take to
run
now when you're trying to explain why it
took so long you could factor how long
it takes a program to run into three
factors one of the first one is how many
instructions did it take to execute so
that's the that's the what we've been
talking about you know the instructions
of Academy
how many did it take all right the next
question is how long did each
instruction take to run on average so
you multiply the number instructions
times how long it took to run and that
gets you help okay so that's but now
let's look at this metric of how long
did it take the instruction to run well
it turns out the way we could build
computers today is they all have a clock
and you've seen this when you if you buy
a microprocessor it'll say 3.1 gigahertz
or 2.5 gigahertz and more gigahertz is
good well what that is is the speed of
the clock so 2.5 gigahertz turns out to
be 4 billions of instruction or 4
nanoseconds so that's the clock cycle
time but there's another factor which is
what's the average number of clock
cycles that takes per instructions so
it's number of instructions average
number of clock cycles in the clock
cycle time
so in these risks ist's debates we would
we they would concentrate on but wrist
makes needs to take more instructions
and we'd argue what maybe the clock
cycle is faster but what the real big
difference was was the number of clock
cycles per instruction or instruction as
fascinating what about the mess up the
beautiful mess of parallelism in the
whole picture parallelism which has to
do was say how many instructions could
execute in parallel and things like that
you could think of that as affecting the
clock cycles per instruction because
it's the average clock cycles per
instruction so when you're running a
program if it took a hundred billion
instructions and on average it took two
clock cycles per instruction and they
were four nanoseconds you could multiply
that out and see how long it took to run
and there's all kinds of tricks to try
and reduce the number of clock cycles
per instruction but it turned out that
the way they would do these complex
instructions is they would actually
build what we would call an interpreter
in a simpler a very simple hardware
interpreter but it turned out that for
sis constructions if you had to use one
of those interpreters it would be like
10 clock cycles per instruction where
the risk instructions could be too so
there'd be this factor of five advantage
in clock cycles per instruction we have
to execute say 25 or 50 percent more
instructions so that's where the wind
would come and then you could make an
argument whether the clock cycle times
are the same or not but pointing out
that we could divide the benchmark
results time per program into three
factors and the biggest difference
between risk consists was the clock
cycles per you execute a few more
instructions but the clock cycles per
instruction is much less and that was
what this debate once we made that
argument then people say okay I get it
and so we went from it was outrageously
controversial in you know 1982 that
maybe probably by 1984 so people said oh
yeah
technically they've got a good argument
what are the instructions in the RISC
instruction set just to get an intuition
okay
1995 I was asked scientific the future
of what microprocessor so I and that
well as I'd seen these predictions and
usually people predict something
outrageous just to be entertaining right
and so my prediction for 2020 was you
know things are gonna be pretty much
they're gonna look very familiar to what
they are and they are in if you were to
read the article you know the things I
said are pretty much true the
instructions that have been around
forever are kind of the same and that's
the outrageous prediction actually yeah
given how fast computers and well you
know Moore's law was gonna go on we
thought for 25 more years you know who
knows but kind of the surprising thing
in fact you know Hennessy and I you know
won the the ACM a.m. Turing award for
both the RISC instruction set
contributions and for that textbook I
mentioned but you know we are surprised
that here we are 35 40 years later after
we did our work and the the conventional
wisdom of the best way to do instruction
sets is still those RISC instruction
sets that look very similar to what we
look like you know we did in the 1980s
so those surprisingly there hasn't
some radical new idea even though we
have you know a million times as many
transistors as we had back then but what
are the basic constructions and how did
they change over the years so we're
talking about addition subtract these
are the specific so the the to get so
the things that are in a calculator you
are in a computer so any of the buttons
that are in the calculator in the crater
so the little button so if there's a
memory function key and like I said
those are turns into putting something
in memories called a store bring
something back Scott load just as a
quick tangent when you say memory what
does memory mean well I told you there
were five pieces of a computer and if
you remember in a calculator there's a
memory key so you you want to have
intermediate calculation and bring it
back later
so you'd hit the memory plus key M plus
maybe and it would put that into memory
and then you'd hit an REM like return
instruction and it bring it back in the
display so you don't have to type it you
don't have to write it down bring it
back again so that's exactly what memory
is if you can put things into it as
temporary storage and bring it back when
you need it later
so that's memory and loads and stores
but the big thing the difference between
a computer and a calculator is that the
computer can make decisions and in
amazingly the decisions are as simple is
is this value less than zero or is this
value bigger than that value so there's
and those instructions which are called
conditional branch instructions is what
give computers all its power if you were
in the early days of computing before
the what's called the general-purpose
microprocessor people would write these
instructions kind of in hardware and but
it couldn't make decisions it would just
it would do the same thing over and over
again with the power of having branch
instructions that can look at things and
make decisions automatically and it can
make these decisions you know billions
of times per second and amazingly enough
we can get you know thanks to advances
machine learning we can we can create
programs that can do something smarter
than human beings can do but if you go
down that very basic level it's the
instructions are the keys on the
calculator plus the ability to make
decisions of these conditional branch
instructions you know and all decisions
fundamental can be reduced down to these
- assumptions yeah so in in fact and so
you know going way back in the sack back
to you know we did for risk projects at
Berkeley in the 1980s they did a couple
at Stanford in the 1980s in 2010 we
decided we wanted to do a new
instruction set learning from the
mistakes of those RISC architectures of
1980s and that was done here at Berkeley
almost exactly 10 years ago in the the
people who did it I participated but
other Christos Sanne and others
drove it
they called it risk 5 to honor those
risk the four risk projects of 1980s so
what is risk 5 involved so leaders 5 is
another instruction set of vocabulary
it's learned from the mistakes of the
past but it still has if you look at the
there's a core set of instructions it's
very similar to the simplest
architectures from the 1980s and the big
difference about risk 5 is it's open so
I talked early about proprietary versus
open and kind of sauce software so this
is an instruction set so it's a
vocabulary it's not it's not hardware
but by having an open instruction set we
can have open source implementations
open source processors that people can
use where do you see that going says
it's the really exciting possibilities
but she's just like in the Scientific
American if you were to predict 10 20 30
years from now that kind of ability to
utilize open source instruction set
architectures like risk 5 what kind of
possibilities might that unlock yeah and
so just to make it clear because this is
confusing the specification of risk 5 is
something that's like in a text book
there's books about it so that's what
that's kind of defining an interface
there's also the way you build hardware
is you write it in languages they're
kind of like sea but they're specialized
for hardware that gets translated into
hardware and so these implementations of
this specification are what are the open
source so they're written in something
that's called Verilog or VHDL but it's
put up on the web
like that you can see the C++ code for
Linux on the web so that's the open
instruction set enables open source
implementations at risk five so you can
literally build a processor using this
instruction set people are and people
are so what happened to us that the
story was this was developed here for
our use to do our research and we made
it we licensed under the berkeley
software distribution license like a lot
of things get licensed here so other
academics use it they wouldn't be afraid
to use it and then about 2014 we started
getting complaints that we were using it
in our research in our courses and we
got complaints from people in industries
why did you change your instruction set
between the fall and the spring semester
and well we get complaints of additional
time why the hell do you care what we do
with our instruction set and then when
we talked to him we found out there was
this thirst for this idea of an open
instruction set architecture and they
had been looking for one they stumbled
upon ours at Berkeley thought it was boy
this looks great we should use this one
and so once we realize there is this
need for an open instruction set
architecture we thought that's a great
idea and then we started supporting it
and tried to make it happen so this was
you know kind we accidentally stumbled
into this and to this need in our timing
was good and so it's really taking off
there's a you know universities are good
at starting things but the not good it's
sustaining things so like Linux has the
Linux Foundation there's a risk 5
foundation that we started there's
there's an annual conferences and the
first one was done I think January 2015
and the one that was just last December
in it you know it had 50 people at it
and the last one last December had kind
of 1,700 people were at it and the
companies excited all over the world
so if predicting into the future you
know if we were doing 25 years I would
predict that risk 5 will be you know
possibly the most popular instruction
set architecture out there because it's
a pretty good instruction set
architecture and it's open and free and
there's no reason
lots of people shouldn't use it and
there's benefits just like Linux is so
popular today compared to 20 years ago I
and you know the fact that you can get
access to it for free you can modify it
you can improve it for all those same
arguments and so people collaborate to
make it a better system for all
everybody to use and that works in
software and I expect the same thing
will happen in hardware so if you look
at arm Intel mips if you look at just
the lay of the land and what do you
think just for me because I'm not
familiar how difficult this kind of
transition would how much challenges
this kind of transition would entail do
you see let me ask my dumb question
another one no that's I know where
you're headed well there's a budget I
think the thing you point out there's
there's these proprietary popular
proprietary instruction sets the x86 and
so how do we move to risk five
potentially in sort of in the span of
five 10 20 years a kind of a unification
in given that the device is the kind of
way we use devices IOT mobile devices
and and the cloud keeps changing well
part of it a big piece of it is the
software stack and what right now
looking forward there seem to be three
important markets there's the cloud and
then the cloud is simply companies like
Alibaba and Amazon and Google Microsoft
having these giant data centers with
tens of thousands of servers in maybe a
hunt maybe a hundred of these data
centers all over the world and that's
what the cloud is so the computer that
dominates the cloud is the x86
instruction set so the instructions are
the vocal instructor sets using the
cloud of the x86 almost almost 100% of
that today is x86 the other big thing
are cell phones and laptops those are
the big things today I mean the PC is
also dominated by the x86 instruction
set but those
sales are dwindling you know there's
maybe 200 million pcs a year and there's
I serve one and a half billion phones a
year there's numbers like that so for
the phones that's dominated by arm and
now and a reason that I talked about the
software stacks and then the third
category is Internet of Things which is
basically embedded devices things in
your cars and your microwaves everywhere
so what's different about those three
categories is for the cloud the software
that runs in the cloud is determined by
these companies Alibaba Amazon Google
Microsoft so that they control that
software stack for the cell phones
there's both for Android and Apple the
software they supply but both of them
have marketplaces where anybody in the
world can build software and that
software is translated or you know
compiled down and shipped in the
vocabulary of arm so that's the the
what's referred to as binary compatible
because the actual it's the instructions
are turned into numbers binary numbers
and shipped around the world so and the
size just a quick interruption so arm
what is arm as arm is an instructions
like a risk-based yeah it's a risk-based
instruction as a proprietary one arm
stands for advanced RISC machine erm is
the name where the company is so it's a
proprietary RISC architecture so and
it's been around for a while and you
know the surely the most popular
instruction set in the world right now
they every year billions of chips are
using the arm design in this post PC era
is what it was the one of the early risk
adopters of the risk yeah yeah the first
arm goes back I don't know 86 or so so
Berkeley instead did their work in the
early 80s their arm guys needed an
instruction set and they read our papers
and it heavily influenced them so
getting back my story what about
Internet of Things well software's not
shipped in Internet of Things it's the
the embedded device people control that
software stack so you would the
opportune
these four risk five everybody thinks is
in the internet of things embedded
things because there's no dominant
player like there is in the cloud or the
smartphones and you know it's it's
doesn't have a lot of licenses
associated with and you can enhance the
instruction set if you want and it's a
in and people have looked at instruction
sets and think it's a very good
instruction set so it appears to be very
popular there it's possible that in the
cloud people those companies control
their software stacks so that it's
possible that they would decide to use
verse five if we're talking about ten
and twenty years in the future the one
of the be harder it would be the cell
phones since people ship software in the
arm instruction set that you'd think be
the more difficult one but if if risk
five really catches on and you know you
could in a period of a decade you can
imagine that's changing over to give a
sense why risk five our arm is dominated
you mentioned these three categories why
has why did arm dominate why does it
dominate the mobile device base and
maybe the my naive intuition is that
there are some aspects of power
efficiency that are important yeah that
somehow come along with risk well part
of it is for these old Siskin structions
that's like in the x86 it it was more
expensive to these for the you know
they're older so they have disadvantages
in them because they were designed forty
years ago but also they have to
translate in hardware from sis
constructions to risks instructions on
the fly and that costs both silicon area
that the chips are bigger to be able to
do that and it uses more power so arm
his which has you know followed this
risk philosophy is seen to be much more
energy-efficient and in today's computer
world both in the cloud in cell phone
and you know things it isn't the
limiting resource isn't the number of
transistors you can fit in the chip it's
what how much power can you dissipate
for your application so by having a
reduced instruction set you that's
possible to have
a simpler hardware which is more energy
efficient in energy efficiency is
incredibly important in the cloud when
you have tens of thousands of computers
in a datacenter you want to have the
most energy-efficient ones there as well
and of course for embedded things
running off of batteries you want those
to be energy efficient in the cell
phones too so it I think it's believed
that there's a energy disadvantage of
using these more complex instruction set
architectures so the other aspect of
this is if we look at Apple Qualcomm
Samsung Huawei all use the ARM
architecture and yet the performance of
the systems varies I mean I don't know
whose opinion you take on but you know
Apple for some reason seems to perform
better and try these implementations
architecture so where's the magic and
sure that happened yeah so what arm
pioneered was a new business model as
they said well here's our proprietary
instruction set and we'll give you two
ways to do it eat there we'll give you
one of these implementations written in
things like C called Verilog and you can
just use ours well you have to pay money
for that not only pay will give you the
you know will license use to do that or
you could design your own and so we're
talking about numbers like tens of
millions of dollars to have the right to
design your own since they it's the
instruction set belongs to them so Apple
got one of those the right to build
their own most of the other people who
build like Android phones just get one
of the designs from arm and to do it
themselves
so Apple developed a really good
microprocessor design team they you know
acquired a very good team that had was a
building other microprocessors and
brought them into the company to build
their designs so the instruction sets
are the same the specifications are the
same but their hardware design is much
more efficient than I think everybody
else's and that's given Apple an
advantage in the marketplace and that
the iPhones tend to be the faster than
most everybody else's phones that are
they
it'd be nice to be able to jump around
and kind of explore different little
sides of this but let me ask one sort of
romanticized question what to you is the
most beautiful aspect or idea of risk
instruction set or instruction sets for
this you know what I think that you know
I I'm you know I I was always attracted
to the idea of you know smallest
beautiful why is that the temptation in
engineering it's kind of easy to make
things more complicated it's harder to
come up with a it's more difficult
surprising they come up with a simple
elegant solution and I think that
there's a bunch of small features of of
risk in general that you know where you
can see this examples of keeping it
simpler makes it more elegant
specifically in risk five which you know
I'm I was kind of the mentor in the
program but it was really driven by
christos sama and two grad
students Andrew Waterman Yin Sibley is
they hit upon this idea of having a
subset of instructions a nice simple
instruction subset instructions like
40-ish instructions that all software
the software status v can run just on
those forty instructions and then they
provide optional features that could
accelerate the performance instructions
that if you needed them could be very
helpful but you don't need to have them
and that that's a new really a new idea
so risk five has right now maybe five
optional subsets that you can pull in
but the software runs without them if
you just want to build the just the core
forty instructions that's fine you can
do that so this is fantastic
educationally is so you can explain
computers you only have to explain forty
instructions and not thousands of them
also if you invent some wild and crazy
new technology like you know biological
computing you'd like a nice simple
instruction set and you can risk 5e if
you implement those core instructions
you can run you know really interesting
programs on top of that so this idea of
a core set of instructions that the
software stack runs on
and then optional features that if you
turn them on the compilers where used
but you don't have to I think is a
powerful idea what's happened in the
past if for the proprietary instruction
sets is when they add new instructions
it becomes required piece and so that
all all microprocessors in the future
have to use those instructions so it's
kind of like is for a lot of people as
they get older they gain weight
all right is it that weight and age are
correlated and so you can see these
instruction sets get getting bigger and
bigger as they get older so risk five
you know let's you be as slim as your as
a teenager and you only have to add
these extra features if you're really
gonna use them rather than every you
have no choice you have to keep growing
with the instruction set I don't know if
the analogy holds out but that's a
beautiful notion that there's it's
almost like a nudge towards here's the
simple core that's the essential yeah I
think the surprising thing is still if
we if we brought back you know the
pioneers from the 1950s and showed them
the instruction set architectures they'd
understand it they that doesn't look
that different well you know I'm
surprised and it's if there's it may be
something you know to talk about
philosophical things I mean there may be
something powerful about those you know
forty or fifty instructions that all you
need is these commands like these
instructions that we talked about and
that is sufficient to build to bring
upon you know artificial intelligence
and so it's a remarkable surprising to
me that is complicated
Minoo microprocessors where the line
widths are narrower than the wavelength
of light you know is this amazing
technologies at some fundamental level
the commands that software execute are
really pretty straightforward and
haven't changed that much in in decades
it's what a surprising outcome so
underlying all computation all Turing
machines all artificial intelligent
systems perhaps might be a very simple
instruction
set like like a risk 5 or it's yeah I
mean I that's kind of what I said I was
interested to see I had another more
senior faculty colleague and he he had
written something in Scientific American
in you know his 25 years in the future
and his turned out about when I was a
young professor and he said yep
I checked it I was interest to see how
that was going to turn out for me and
it's pretty held up pretty well but yeah
so there's there's probably there's
something I you know there's there must
be something fundamental about those
instructions that were capable of
creating you know intelligence and from
pretty primitive operations and just
doing them really fast you kind of
mentioned the different maybe radical
computational medium like biological and
there's other ideas so there's a lot of
spaces in a6 or domain-specific and then
there could be quantum computers and
wood so we couldn't think of all those
different mediums and types of
computation what's the connection
between swapping out different Hardware
systems and the instruction set do you
see those as disjoint or they
fundamentally coupled yeah so what's so
kind of if we go back to the history you
know when Moore's Law is in full effect
and you're getting twice as many
transistors every couple of years you
know kind of the challenge for computer
designers is how can we take advantage
of that how can we turn those
transistors into better computers faster
typically and so there was an era I
guess in the 80s and 90s where computers
were doubling performance every 18
months and if you weren't around then
what would happen is you had your
computer and your friend's computer
which was like a year year and a half
newer and it was much faster than your
computer and you he he or she could get
their work done much faster than your
typical user so people took their
computers perfectly good computers and
threw them away to buy a newer computer
because the computer one or two years
later was so much faster so that's what
the
world was like in 80s and 90s well with
the slowing down of Moore's law that's
no longer true right he not now with you
know not decide computers with the
laptops I only get a new laptop when it
breaks right well damn the disk broke or
this display broke I got to buy a new
computer but before you would throw them
away because it just they were just so
sluggish compared to the latest
computers so that's you know that's a
huge change of what's gone on so but yes
since this lasted for decades kind of
programmers and maybe all society is
used to computers getting faster
regularly it we now now believe those of
us who are in computer design it's
called computer architecture that the
path forward is instead is to add
accelerators that only work well for
certain applications
so since Moore's law is slowing down
we don't think general-purpose computers
are gonna get a lot faster so the Intel
processors of the world are not going to
haven't been getting a lot faster
they've been barely improving like a few
percent a year
it used to be doubling your 18 months
and now it's doubling every 20 years so
it was just shocking so to be able to
deliver on what Moore's law used to do
we think what's going to happen what is
happening right now is people adding
accelerators to their microprocessors
that only work well for some domains and
by sheer coincidence at the same time
that this is happening has been this
revolution in artificial intelligence
called machine learning so with as I'm
sure your other guess I've said you know
a I had these two competing schools of
thought is that we could figure out
artificial intelligence by just writing
the rules top-down or that was wrong you
had to look at data and infer what the
rules are the machine learning and
what's happened in the last decade or
eight years this machine learning has
won and it turns out that machine
learning the hardware you built from
learning is pretty much multiply the
matrix multiply is a key feature for the
way people machine learning is done so
that's a godsend for computer designers
we know how to make metrics multiply run
really fast so general-purpose
microprocessors are slowing down we're
adding accelerators from machine
learning that fundamentally are doing
matrix multiplies much more efficiently
than general-purpose computers have done
so we have to come up with a new way to
accelerate things the danger of only
accelerating one application is how
important is that application turns it
turns out machine learning gets used for
all kinds of things so serendipitously
we found something to accelerate that's
widely applicable and we don't even
we're in the middle of this revolution
of machine learning we're not sure what
the limits of machine learning are so
this has been kind of a godsend if
you're going to be able to Excel deliver
on improved performance as long as
people are moving their programs to be
embracing more machine learning we know
how to give them more performance even
as Moore's Law is slowing down and
counter-intuitively
the machine learning mechanism you can
say is domain-specific but because it's
leveraging data it's actually could be
very broad in terms of in terms of the
domains it could be applied in yeah
that's exactly right sort of it's almost
sort of people sometimes talk about the
idea of software 2.0 we're almost taking
another step up in the abstraction layer
in designing machine learning systems
because now you're programming in the
space of data in the space of hyper
parameters it's changing fundamentally
the nature of programming and so the
specialized devices that that accelerate
the performance especially neural
network based machine learning systems
might become the new general yes so the
this thing that's interesting point out
these are not coral these are not tied
together the it's enthusiasm about
machine learning about creating programs
driven from data that we should figure
out the answers from
rather than kind of top down which
classically the way most programming is
done in the way artificial intelligent
used to be done
that's a movement that's going on at the
same time coincidentally and the the
first word machine learnings machines
right so that's going to increase the
demand for computing because instead of
programmers being smart writing those
those things down we're going to instead
use computers to exam a lot of data to
kind of create the programs that's the
idea and remarkably this gets used for
all kinds of things very successfully
the image recognition the language
translation the game playing and you
know it gets into pieces of the software
stack like databases and stuff like that
we're not quite sure how journal
purposes but that's going on independent
as Hardware stuff what's happening on
the hardware side is Moore's Law is
slowing down right when we need a lot
more cycles it's failing us it's failing
us right when we need it because there's
going to be a greater in peace a greater
increase in computing and then this idea
that we're going to do so-called
domain-specific here's a domain that
your greatest fear is you'll make this
one thing work and that'll help you know
5% of the people in the world well this
this looks like it's a very
general-purpose thing so the timing is
fortuitous that if we can perhaps if we
can keep building hardware that will
accelerate machine learning the neural
networks that'll beat the timing D right
that that neural network revolution will
transform your software the so called
software 2.0 and the software the future
will be very different from the software
the past and just as our microprocessors
even though we're still going to have
that same basic risk instructions to run
a big pieces of the software stack like
user interfaces and stuff like that we
can accelerate the the kind of the small
piece that's computationally intensive
it's not lots of lines of code but there
it takes a lot of cycles to run that
code that that's going to be the
accelerator piece and so this that's
what makes this from a computer
designer's perspective a really
interesting decade but Hennessy and I
talked about
that the title of our Turing warrant
speech is a new golden age we we see
this as a very exciting decade much like
when we were assistant professors and
the wrists stuff was going on that was a
very exciting time was where we were
changing what was going on we see this
happening again tremendous opportunities
of people because we're fundamentally
changing how software is built and how
we're running it so which layer of the
abstraction do you think most of the
acceleration might be happening the if
you look in the next ten years that
Google is working on a lot of exciting
stuff with the TPU sort of there's a
closer to the hardware that could be
optimizations around the IROC closer to
the instruction set that could be
optimization at the compiler level it
could be even at the higher level
software stack yeah it's going to be I
mean if you think about the old risks
this debate it was both
it was software hardware it was the
compilers improving as well as the
architecture improving and that that's
likely to be the way things are now
with machine learning they they're using
domain-specific languages the languages
like tensorflow and pi torch are very
popular with the machine learning people
that those are the raising the level of
abstraction it's easier for people to
write machine learning in these
domain-specific languages like like a PI
torch in tensorflow
so where the most of the optimization
but yeah and so that and so there'll be
both the compiler piece and the hardware
piece underneath it so as you kind of
the fatal flaw for hardware people is to
create really great hardware but not
have brought along the compilers and
what we're seeing right now in the
marketplace because of this enthusiasm
around hardware for machine learning is
getting you know probably a billions of
dollars invested in start-up companies
we're seeing startup companies go
belly-up because they focus on the
hardware but didn't bring the software
stack along we talked about benchmarks
earlier so I participated in machine
learning didn't really have a set of
benchmarks I think just two years ago
they didn't have a set of benchmarks and
we've created something called ml perf
which
machine learning benchmark suite and
pretty much the companies who didn't
invest in the software stack couldn't
run a ml per fairy wall and the ones who
did invest in software stack did and
we're seeing you know like kind of in
computer architecture this is what
happens you have these arguments about
risk versus ist's people spend billions
of dollars in the marketplace to see who
wins and it's not it's not a perfect
comparison but it kind of sorts things
out and we're seeing companies go out of
business and then companies like like
there's a company in Israel called
Habana they came up with machine
learning accelerators that they had good
ml perf scores Intel had acquired a
company earlier called nirvana
a couple years ago they didn't reveal
the amount of Perth's cores which was
suspicious but month ago
Intel announced that they're cancelling
the Nirvana product line and they've
bought Habana for two billion dollars
and Intel's going to be shipping Habano
chips which have hardware and software
and run the ml perf programs pretty well
and that's going to be their product
line in the future brilliant so maybe
just a linker briefly I'm a love metrics
I love standards that everyone can
gather around what are some interesting
aspects of that portfolio of metrics
well one of the interesting metrics is
you know what we thought it was you know
we I was involved in the start you know
we that Peter Matson is leading the
effort from Google Google got it off the
ground but we had to reach out to
competitors and say there's no
benchmarks here this we didn't we think
this is bad for the field it'll be much
better if we look at examples like in
the wrist days there was an effort to
create a for the the people in the risk
community got together competitors got
together a building risk microprocessors
to agree on a set of benchmarks that we
called spec and that was good for the
industry is rather before the different
risk architectures were arguing well you
can believe my performance others but
those other guys are liars and that
didn't do any good so we agreed on a set
of benchmarks and then we could figure
out who is faster between the various
risk architectures but it was a little
bit faster but that drew the market
rather than you know people were afraid
to buy anything so we argued the same
thing would happen
with him helper you know companies like
Nvidia were you know maybe worried that
it was some kind of trap but eventually
we all got together to create a set of
benchmarks and do the right thing right
and we agree on the results and so we
can see whether TP use or GPUs or CPUs
are really faster than how much the
faster and I think from an engineer's
perspective as long as the results are
fair Europe you can live with it okay
you know you have a tip your hat to to
your colleagues at another institution
boy they did a better job than this what
you what you hate is if it's it's false
right they're making claims and it's
just marketing and you know in
that's affecting sales so you from an
engineer's perspective as long as it's a
fair comparison and we don't come in
first place that's too bad but it's fair
so we wanted to create that environment
frame all perf and so now there's ten
companies I mean ten universities and
fifty companies involved so pretty much
AML perf has is the is the way you
measure machine learning performance and
and it didn't exist even two years ago
one of the cool things that I enjoy
about the Internet has a few downsides
but one of the nice things is people can
see through BS a little better with the
presence yes has a metrics it's so it's
really nice a companies like Google and
Facebook and Twitter now it's the cool
thing to do is to put your engineers
forward and to actually show off how
well you do on these metrics there's not
sort of it well there's a less of a
desire to do marketing a less so in my
in my sort of naive no I don't think
well I was trying to understand that you
know what's changed from the 80s in this
era I think because of things like
social networking Twitter and stuff like
that if you if you put up you know
stuff right that's just you
know miss purposely misleading you know
that you you can get a violent reaction
in social media pointing out the flaws
in your arguments right and so from a
marketing perspective you have to be
careful today that you didn't have to be
careful that there'll be people who put
off the flaw you can get the word out
the flaws and what you're saying much
more easily today than in the past you
used to be it was used to be easier to
get away with it and the other thing
that's been happening in terms of
starting off engineers it's just in the
software side people have largely
embraced open-source software it it was
20 years ago it was a dirty word at
Microsoft and today Microsoft is one of
the big proponents of open source
software the kind of that's the standard
way most software gets built which
really shows off your engineers because
you can see if you look at the source
code you can see who are making the
commits who's making the improvements
who are the engineers at all these
companies who are are you know really
great programmers and engineers and
making really solid contributions which
enhances their reputations and the
reputation of the companies so but
that's of course not everywhere like in
this space that I work more in is
autonomous vehicles and they're still
the machinery of hype and marketing is
still very strong there and there's less
willingness to be open in this kind of
open source way and sort of benchmark so
ml Perez represents the machine learning
world is much better being open-source
about holding itself to standards of
different the amount of incredible
benchmarks in terms of the different
computer vision naturally new processing
- inaudible
it you know historically it wasn't
always that way I had a graduate student
working with me David Martin so for in
computer in some fields benchmarking is
been around forever so computer
architecture databases maybe operating
systems benchmarks are the way you
measure progress but he was working with
me and then started working with gender
Malik and he's a gender Malik in
computer vision space who I guess you've
you interviewed yes and David Martin
told me they don't have benchmarks
everybody has their own vision algorithm
in the way that my here's my image look
at how well I do and everybody had their
own image so David Martin
back when he did his dissertation
figured out a way to do benchmarks he
had a bunch of graduate students
identify images and then ran benchmarks
to see which algorithms run well and
that was as far as I know kind of the
first time people did benchmarks in
computer vision in which was predated
all you know the things that eventually
led to imagenet himself like that but
then you know the vision community got
religion and then once we got as far as
image net then that let the guys in
Toronto be able to win the image net
competition and then you know that
changed the whole world it's a scary
step actually because when you enter the
world of benchmarks you actually have to
be good to participate as opposed to
yeah you can just you just believe
you're the best in the world and I think
the people I think they weren't
purposely misleading I think if you
don't have benchmarks I mean how do you
know you know you could have your
intuition it's kind of like the way we
did used to do computer architecture
your intuition is that this is the right
instruction set to do this job I believe
in my experience my hunch is that's true
we had to get to make things more
quantitative to make progress and so I
just don't know how you know in fields
that don't have benchmarks I don't
understand how they figure out how
they're making progress we're kind of in
the vacuum tube days of quantum
computing what are your thoughts in this
wholly different kind of space of
architectures uh you know I actually you
know quantum computing his ideas been
around for a while and I actually
thought well sure hope I retire before I
have to start teaching this I'd say
because I talked about give these talks
about the slowing of Moore's law and you
know when we need to change by doing
domain-specific accelerators common
questions say what about quantum
computing the reason that comes up it's
in the news all the time so I think the
keep and the third thing to keep in mind
is quantum computing is not right around
the corner there have been two national
reports one by the national campus of
engineering another by the computing
consortium where they did a frank
assessment of quantum computing
in both of those reports said you know
as far as we can tell before you get
error corrected quantum computing it's a
decade away so I think of it like
nuclear fusion right there been people
who've been excited about nuclear fusion
a long time if we ever get nuclear
fusion it's going to be fantastic for
the world I'm glad people are working on
it but you know it's not right around
the corner that those two reports to me
say probably it'll be 2030 before
quantum computing is a something that
could happen and when it does happen you
know this is going to be big science
stuff this is you know microkelvin
almost absolute zero things that if they
vibrate if truck goes by it won't work
right so this will be in data center
stuff we're not gonna have a quantum
cell phone and and it's probably a 2030
kind of thing so I'm happy that other
people are working on it but just you
know it's hard with all the news about
it not to think that it's right around
the corner and that's why we need to do
something as Moore's Law is slowing down
to provide the computing keep improving
getting better for this next decade and
and you know we shouldn't be betting on
quantum computing are expecting quantum
computing to deliver in the next few
years it's it's probably further off you
know I I'd be happy to be wrong it be
great if quantum computing is gonna
commercially viable but it will be a set
of applications it's not a
general-purpose computation so it's
gonna do some amazing things but
there'll be a lot of things that
probably you know the the old-fashioned
computers are gonna keep doing better
for quite a while and there'll be a
teenager 50 years from now watching this
video saying look how silly David
Patterson was saying I said what did
I didn't say sorry I never we're not
gonna have quantum cellphones so he's
gonna be watching and well I mean III
think this is such a you know given
we've had Moore's law I just I feel
comfortable trying to do projects that
are thinking about the next decade I I
admire people who are trying to do
things that are 30 years out but it's
such a fast-moving field I just don't
know how to I'm
not good enough to figure out what
what's the problems gonna be in 30 years
you know 10 years is hard enough for me
so maybe if it's possible to untangle
your intuition a little bit
I spoke with Jim Keller I don't know if
you're familiar with Jim and he he is
trying to sort of be a little bit
rebellious and to try to think that he
quotes me as being wrong yeah so what
are your the relationship for the record
Jim talks about that he has an intuition
that Moore's law is not in fact in fact
dead yet and then it may continue for
some time to come
what are your thoughts about Jim's ideas
in this space yeah this is just this is
just marketing so but Gordon Moore said
is a quantitative prediction if we can
check the facts right which is doubling
the number of transistors every two
years so we can look back at Intel for
the last five years and ask him let's
look at DRAM chips six years ago so that
would be three two-year periods so then
our DRAM chips have eight times as many
transistors as they did six years ago we
can look up Intel microprocessors six
years ago
if Moore's law is continuing it should
have eight times as many transistors as
six years ago the answers in both those
cases is no the problem has been because
Moore's law was kind of genuinely
embraced by the semiconductor industries
they would make investments in severe
equipment to make Moore's Law come true
semiconductor improving in Moore's law
in many people's mind are the same thing
so when I say and I'm factually correct
that Moore's law is no longer holds we
are not doubling transistors every years
years the downside for a company like
Intel is people think that means it
stopped that technology has no longer
improved and so Jim is trying to react
at AraC the impression that
semiconductors are frozen in 2000
nineteen are never gonna get better so I
never said that I said was Moore's law
is no more and I'm strictly looking at a
number of transistors because that's
what more that's what Moore's law is
there's the I don't know there's been
this aura associated with Moore's law
that they've enjoyed for fifty years
about look at the field we're in we're
doubling transistors every two years
what an amazing field which is an
amazing thing that they were able to
pull off but even as Gordon Moore said
you know no exponential can last forever
it's lasted for 50 years which is
amazing and this is a huge impact on the
industry because of these changes that
we've been talking about so he claims
because he's trying to act and he claims
you know Patterson says Moore's laws
know more and look at all look at it
it's still controlling and tsmc to say
it's as no longer but there but there's
quantitative evidence that Moore's law
is not continuing so what I say now to
try and okay I understand the perception
problem when I say Moore's law is
stopped okay so now I say Moore's law
slowing down and I think Jim which is
another way if he's if it's predicting
every two years and I say it's slowing
down then that's another way of saying
it doesn't hold anymore and and I think
Jim wouldn't disagree that it's slowing
down because that sounds like it's
things are still getting better just not
as fast which is another way of saying
Moore's law isn't working anymore
it's still good for marketing but uh but
what's your you're not you don't like
expanding the definition of Moore's law
sort of uh well yeah that's really yeah
it's an educator you know are you know
is this like bonding politics is
everybody get their own facts
or do we have Moore's law was a crisp
you know amorous Carver Mead looked at
his observations drawing on a log-log
scale a straight line and that's what
the definition of Moore's law is there's
this other what Intel did for a while
interestingly before Jim joined them
they said oh no Morris lies in the
number of doubling isn't really doubling
transistors every two years
Moore's law is the cost of the
individual dressed
sister going down cutting in half every
two years now that's not what he said
but they reinterpreted it because they
believed that the that the cost of
transistors was continuing to drop even
if they couldn't get twice as many
people industry have told me that's not
true anymore that basically then the in
more recent technologies that got more
complicated the actual cost of
transistor went up so even even the a
corollary might not be true but
certainly you know Moore's law that was
the beauty of Moore's law it was a very
simple it's like equals mc-squared right
it was like wow what an amazing
prediction it's so easy to understand
the implications are amazing and that's
why it was so famous as a as a
prediction and this this
reinterpretation of what it meant and
changing is you know his revisionist
history and I I'd be happy and and
they're not claiming there's a new
Moore's law they're not saying by the
way it's instead of every two years it's
every three years I don't think the I
don't think they want to say that I
think what's going to happen is the new
technology Commission's H ones get a
little bit slower so it it is slowing
down the improvements will won't be as
great and that's why we need to do new
things yeah I don't like that the the
idea of Moore's law is tied up with
marketing I it would be nice if it's
whether it's marketing or it's it's well
it could be affecting business but they
could also be infecting the imagination
of engineers is if if Intel employees
actually believe that we're frozen in
2019 well that's that would be bad for
Intel they not just Intel but everybody
it's inspired Moore's law is inspiring
yeah everybody but what's happening
right now talking to people in who have
working in national offices and stuff
like that a lot of the computer science
community is unaware that this is going
on right that we are in an era that's
going to need radical change at lower
levels that could affect the whole
software stack this you know if
if the Intel if you're using cloud stuff
and servers that you get next year are
basically only a little bit faster than
the servers you got this year you need
to know that and we need to start
innovating to start delivery blow on it
if you're counting on your software your
software going to add a lot more
features assuming the computers can get
faster that's not true so are you gonna
have to start making your software stack
more efficient or are you gonna have to
start learning about machine learning so
it's you know it's kind of a it's a
morning or call for arms that the world
is changing right now and a lot of
people a lot of computer science PhDs
are unaware of that so a way to try and
get their attention is to say that
Moore's law is slowing down and that's
gonna affect your assumptions and you
know we're trying to get the word out
and when companies like TSMC and Intel
say oh no no no Moore's law is fine then
people think okay that I don't have to
change my behavior I'll just get the
next servers and you know if they start
doing measurements though realize what's
going on it'd be nice to have some
transparency and metrics for for the
layperson
to be able to know if computers are
getting faster and there are yeah there
are there are a bunch of most people
kind of use clock rate as a measure
performance you know it's not a perfect
one but if you've noticed clock rates
are more or less the same as they were
five years ago computers are a little
better than they aren't they haven't
made zero progress but they've made
small progress so you there's some
indications out there and in our
behavior right nobody buys the next
laptop because it's so much faster than
the laptop from the past four cell
phones I think I don't know why people
buy new cell phones you know because of
the new ones announced the cameras are
better but that's kind of
domain-specific right they're putting
special purpose hardware to make the
processing of images go much better so
that's that that's the way they're doing
it they're not particularly it's not
that the ARM processor there's twice as
fast as much as they'd added
accelerators to help eat the experience
of the phone
can we talk a little bit about one other
exciting space arguably the same level
of impact as your work with risk is raid
and in your in 1988 you co-authored a
paper a case for redundant array of
inexpensive disks hence our AI D rate so
you that's where you introduce the idea
rate incredible that that little I mean
little that paper kind of had this
ripple effect and had a really
revolutionary effect so first what is
rate what is rate so this is work I did
with my colleague Randy Katz and a star
graduate student Garth Gibson so we had
just done the fourth generation risk
project and Randy Kass which had early
Apple Macintosh computer at this time
everything was done with floppy disks
which are old technologies that to could
store things that didn't have much
capacity and you had to to get any work
done you're always sticking in your
little floppy disk in and out because
they didn't have much capacity but they
started building what are called hard
disk drives which is magnetic material
that can remember information storage
for the Mac and Randy asked the question
when he saw this disk next to his Mac
jeez he's a brand-new small things
before that for the big computers that
the disk would be the size of washing
machines and here's something the size
of a kind of the size of a book or so
this is I wonder what we could do with
that well we the Randy was involved in
the in the fourth generation risk
project here at Berkeley 80s so we
figured out a way how to make the
computation part the processor part go a
lot faster but what about the storage
part can we do something to make it
faster so we hit upon the idea of taking
a lot of these disks developed for
personal computers and mackintoshes and
putting many of them together instead of
one of these washing machine sized
things and so we were to rub the first
draft of the paper and we'd have 40 of
these little PC DOS
instead of one of these washing machine
size things and they would be much
cheaper because they're made for PCs and
they could actually kind of be faster
because there was 40 of them rather than
one of them and so he wrote a paper like
that and send it to one of a former
Berkeley students at IBM and he said
well this is all great and good but what
about the reliability of these things
now you have 40 of these devices each of
which are kind of PC quality so they're
not as good as these IBM washing
machines IBM dominated the the the
storage Genesis so you
reliably gonna be awful and so when we
calculated it out instead of you know it
breaking on average once a year it would
break every two weeks so we thought
about the idea and said well we got to
address the reliability so we did it
originally performance but we had do
reliability so the name redundant array
of inexpensive disks is array of these
disks inexpensive life for pcs but we
have extra copies so if one breaks we
won't lose all the information will have
enough redundancy that we could let some
break and we can still preserve the
information so the name is an array of
inexpensive discs this is a collection
of these pcs and the are part of the
name was the redundancy so they'd be
reliable and it turns out if you put a
modest number of extra disks in one of
these arrays it could actually not only
be as faster and cheaper that one of
these washing machine discs it could be
actually more reliable because you could
have a couple of breaks even with these
cheap discs whereas one failure with the
washing machine thing would knock it out
did you did you have a sense just like
with risk that in the 30 years that
followed raid would take over as a as a
man I think George I I'd say I think I'm
naturally an optimist but I thought our
ideas were right
I thought kind of like Moore's law it
seemed to me if you looked at the
history of the disk drives
they went from washing machine size
things than they were getting smaller
and smaller and the volumes were with
the smaller disk drives because that's
where the PCs were so we thought that
was a technological trend that disk
drives
the volume
disk drives was going to be small
getting smaller and smaller devices
which were true they were the size of
the I don't know eight inches diameter
than five inches than three inches of
diameters and so that it made sense to
figure out how to deal things with an
array of disks so I think it was one of
those things where logically we think
the technological forces were on our
side that it made sense so we expected
it to catch on but there was that same
kind of business question you know IBM
was the big pusher of these disk drives
in the real world where the technical
advantage get turned into a business
advantage or not it proved to be true it
did in so you know we thought we were
sound technically and it was unclear
worth of the business side but we kind
of as academics we believe the
technology should win and and it did and
and if you look at those thirty years
just from your perspective are there
interesting developments in the space of
storage that have happened in that time
yeah the big thing that happened both a
couple of things that happened what we
did had a modest amount of storage so as
redundancy as people built bigger and
bigger storage systems they've added
more we doesn't see so they could have
more failures and they have biggest
thing that happened in storage is for
decades it was based on things
physically spinning called hard disk
drives where you used to turn on your
computer and it would make a noise what
that noise was was the disk drive
spinning and they were rotating it in
like 60 revolutions per second and it's
like if you remember the vinyl vinyl
records if you've ever seen those that's
what it looked like and there was like a
needle like on a vinyl record that was
reading it so the big drive a change is
switching that over to a similar
technology called flash so within the
last I'd say about decade is increasing
fraction of all the computers in the
world are using semiconductor for
storage the flash drive instead of being
magnetic their optical their there well
their semiconductor writing of
information into very densely
and that's been a huge difference so all
the cell phones in the world use flash
most of the laptops use flash all the
embedded devices use flash instead of
storage still in the cloud magnetic
disks are more economical than flash but
they used both in the cloud so it's been
a huge change in the storage industry
this the switching from primarily disk
to being primarily semiconductor for the
individual discs but still the raid
mechanism applies to those different
kinds of yes the the people will still
use raid ideas because it's kind of
what's different you know kind of
interesting kind of psychologically if
you think about it people have always
worried about the reliability of
computing since the earliest days so
kind of but if we're talking about
computation if your computer makes a
mistake and the computer says the
computer has worries to check and say we
screwed up we made a mistake what
happens is that program that was running
you have to redo it which is a hassle
for storage if you've sent important
information away and it loses that
information you go nuts yeah yeah this
is the worst I oh my god so if you have
a laptop and you're not backing it up on
the cloud or something like this and
your disk drive breaks which it can do
you'll lose all that information and you
just go crazy right so the importance of
reliability for storage is tremendously
higher than the importance of
reliability for computation because of
the consequences of it so yes so raid
ideas are still very popular even with
the switch of the technology although
you know flash drives are more reliable
you know if you're not doing anything
like backing it up to get some
redundancy so they handle it you're
you're you're taking great risks you
said that for you and possibly from any
others teaching and research don't
conflict with each other as right one
might suspect and in fact they kind of
complement each other so maybe a
question I have is how is teaching
helped you in your research or just in
your
entirety as a person who both teaches
and does research and just thinks and
creates new ideas in this world yes I
think I think what happens is is when
you're a college student you know
there's this kind of tenure system and
doing research so kind of this model
that you know is popular in America I
think America really made it happen is
we can attract these really great
faculty to research universities because
they get to do research as well as teach
and that especially in fast-moving
fields this means people are up-to-date
and they're teaching those kind of
things so but when you run into a really
bad professor a really bad teacher I
think the students think well this guy
must be a great researcher because why
else could he be here so is I you know I
I after 40 years at Berkeley we had a
retirement party and I got a chance to
reflect and I looked back to some things
that is not my experience there's a I
saw a photograph of five of us in the
department who won the distinguished
Teaching Award from campus a very high
honor you know what I've got one of
those when the highest honors so they're
five of us on that picture
there's Manuel Blum Richard Karp me
Randy Katz and John osterhaus
contemporaries of mine I mentioned Randy
already all of us are in the National
Academy of Engineering we've all run the
distinguished Teaching Award Blum Karp
and I are all have turing award just
going away that's right you know the
highest award in computing so the
opposite right it's what happens if you
it's it's they're highly correlated so
probably the other way to think of it if
you're very successful people may be
successful at everything they do it's
not an either/or and but it's an
interesting question whether
specifically that's probably true but
specifically for teaching if there's
something in teaching that it's the
Richard Fineman right right yeah is
there something about teaching that
actually makes your research makes you
think deeper and more outside the box
and yeah absolutely so yeah I was going
to bring up Fineman I mean he criticized
the Institute of Advanced Studies
he says there's Advanced Studies was
this thing that was created in your
Princeton where Einstein and all these
smart people
and when he was invited he said he
thought it was a terrible idea his this
is a university was it was supposed to
be heaven right a university without any
teaching but he thought it was a mistake
is getting up in the classroom and
having to explain things to students and
having them ask questions like well why
is that true makes you stop and think so
he to think he thought and I agree I
think that interaction between a retina
research university and having students
with bright young man's asking hard
questions the whole time is synergistic
and you know a university without
teaching wouldn't be as vital and
exciting a place and I think it helps
stimulate the the research another
romanticized question but what's your
favorite concept or idea to teach what
inspires you or you see inspire the
students is there something to pasta my
or or puts the fear of God in them I
don't know II whichever is most
effective I mean in general I think
people are surprised I've seen a lot of
people who don't think they like
teaching
come come give guest lectures or teach a
course and get hooked on seeing the
lights turn on right his people you can
explain something to people that they
don't understand and suddenly they get
something you know that's that's not
that's important and difficult and just
seeing the lights turn on is a you know
it's a real satisfaction there I don't
think there's any in a specific example
of that it's just the general joy of
seeing them seeing them understand I
have to talk about this because I've
wrestled I do usual arts yes yes I love
Russ I'm a huge I'm Russian so I'll sure
I'd have talked to Dan Gable oh yeah I
guess so fine yang Gables my era kind of
guy so you wrestled UCLA among many
other things you've done in your life
competitively in sports and science on
you've wrestled maybe again continue in
their immense sessions but what have you
learned about life yeah and maybe even
size from wrestling or from that's in
fact I wrestled at UCLA but also at El
Camino can be
College and just right now we were in
the state of California we were state
champions at El Camino and the fact I
was talking to my mom and I got into
UCLA but I decided to go to the
Community College which is it's much
he's harder to go to UCLA than Community
College and I asked why did I make the
decision because I thought that was
because of my girlfriend she said well
it was the girlfriend and and you
thought the wrestling team was really
good and we were right we had a great
wrestling team it we actually wrestled
against UCLA at a tournament and we beat
UCLA it's a community college which just
freshmen and sophomores and the reason I
brought this up is I'm gonna go they've
invited me back at El Camino if give a
lecture next month and so I'm Liev my
friend who was on the wrestling team
that we're still together we're right
now reaching out to other members of the
wrestling team you can get together
every Union but in terms of me it was a
huge difference
I was I was both I was kind of the age
cutoff I was who's December first and so
I was almost always the youngest person
in my class and I matured later on you
know our family badgered later so I was
almost always the smallest guy so you
know I took in kind of nerdy courses but
I was wrestling so wrestling was huge
for my you know self-confidence in high
school and then you know I kind of got
bigger at El Camino and in college and
so I had this kind of physical
self-confidence and it's translated into
research self-confidence and and also
kind of I've had this feeling even today
in my 70s you know if something if
something going on and streets there's
bad physically I'm not gonna ignore it
right I'm gonna stand up and try and
straighten that out and that kind of
confidence just carries through the
entirety of your life yeah and the same
things happens intellectually if there's
something going on where people are
saying something that's not true I feel
it's my job to stand up and just like I
would in the street if there's something
going on somebody attacking some woman
or something I'm not I'm not standing by
and letting that
so I feel it's my job to stand up so
it's kind of ironically translates the
other things that turned out for both I
had really great college in high school
coaches and they believed even though
wrestling's an individual sport that
would be be more successful as a team if
we bonded together you do things that we
would support each other rather than
everybody you know in wrestling it's
one-on-one and you could be everybody's
on their own but he felt if we bonded as
a team we'd succeed so I kind of picked
up those skills of how to form
successful teams and how do you from
wrestling and so I think one of most
people would say one of my strengths is
I can create teams of faculty watch
teams of faculty grad students pull all
together for a common goal and you know
and you often be successful at it but I
got I got both of those things from
wrestling also I think I heard this line
about if people are in kind of you know
collision you know sports with physical
contact like wrestling or football and
stuff like that people are a little bit
more you know assertive or something so
I think I think that also comes through
is you know in I was I didn't shy away
from the risk debates you know I was
yeah I enjoyed taking on the arguments
and stuff like that so it was it was a
I'm really glad I did wrestling I think
it was really good for my self-image and
I learned a lot from it so I think
that's you know sports done well you
know there's really lots of positives
you can take about it leadership you
know how to how to form teams and how to
be successful so we've talked about
metrics a lot there's a really cool in
terms of bench press and weightlifting
pioneers metric do you develop that we
don't have time to talk about but it's
it's a really cool that people should
look into
it's rethinking the way we think about
metrics and weightlifting but let me
talk about metrics more broadly since
that appeals Cu in all forms let's look
at the most ridiculous the biggest
question of the meaning of life if you
were to try to put metrics on a life
well-lived what would those metrics be
yeah a friend Randy Katz said this he
said you know when when it's time to
sign off it's
it's the measure isn't the number of
zeros in your bank account it's the
number of inches in the obituary in The
New York Times he said it I I think you
know having and you know this is a
cliche is that people don't die wishing
they'd spent more time in the office
right is I reflect upon my career there
have been you know a half a dozen or a
dozen things say I've been proud of a
lot of them aren't papers or scientific
well certainly my family my wife we've
been married more than 50 years kids and
grandkids that's really precious
education thinks I've done I'm very
proud of you know books and courses I
did some help with underrepresented
groups that was effective so it was
interesting just seeing what were the
things I reflected you know I had
hundreds of papers but some of them
weren't the papers like the risk and
rate stuff wasn't proud of but a lot of
them were or not those things so people
who are just spend their lives you know
going after the dollars are going after
all the papers in the world you know
that's probably not the things that are
afterwards you're gonna care about when
I was a yeah just when I got the offer
from Berkeley but before I showed up I
read a book where they interviewed a lot
of people in all walks of life and what
I got out of that book was the people
who felt good about what they did was
the people who affected people as
opposed to things that were more
transitory so I came into this job
assuming that it wasn't going to be the
papers it was gonna be relationships
with the people over time that I would I
would value and that was a correct
assessment right it's it's the people
you work with the people you can
influence the people you can help is the
things that you feel good about towards
into your career it's not not the the
stuff that's more transitory I don't
think there's a better way to end it
then talking about your family the the
over 50 years of being married to your
childhood sweetheart is how do when you
tell people you've been married 50 years
they want to know why how why I can tell
you the nine magic words that you need
to say to your partner to keep a good
relationship in the nine magic words are
was wrong you were right I love you okay
and you got to say all nine you can't
say I was wrong you were right you're a
jerk you know you guess so yeah a freely
acknowledging that you made a mistake
the other person was right and that you
love them really gets over a lot of
bumps in the road so that's what I pass
along beautifully put David is a huge
honor thank you so much for the book
you've written for the research you've
done for changing the world thank you
for talking to that oh thanks for the
interview
thanks for listening to this
conversation with David Patterson and
thank you to our sponsors the Jordan
Harbinger show and cash app please
consider supporting this podcast by
going to Jordan Harbinger complex and
downloading cash app and using colex
podcast click the links buy the stuff
it's the best way to support this
podcast and the journey I'm on if you
enjoy this thing subscribe on youtube
review it with five stars in a podcast
supported on patreon or connect with me
on Twitter and lex Freedman spelled
without the e try to figure out how to
do that it's just fr ID ma n and now let
me leave you with some words from Henry
David Thoreau our life is frittered away
by detail simplify simplify
thank you for listening and hope to see
you next time
you