David Patterson - A New Golden Age for Computer Architecture: History, Challenges and Opportunities
Abstract:
In the 1980s, Mead and Conway democratized chip design and high-level language programming surpassed assembly language programming, which made instruction set advances viable. Innovations like Reduced Instruction Set Computers (RISC), superscalar, and speculation ushered in a Golden Age of computer architecture, when performance doubled every 18 months. The ending of Dennard Scaling and Moore’s Law crippled this path; microprocessor performance improved only 3% last year! In addition to poor performance gains, Spectre recently demonstrated timing attacks on modern microprocessors that leak information at high rates.
The ending of Dennard scaling and Moore’s law and the deceleration of performance gains for standard microprocessors are not problems that must be solved but facts that if accepted offer breathtaking opportunities. We believe high-level, domain-specific languages and architectures, freeing architects from the chains of proprietary instruction sets, and the demand from the public for improved security will usher in a new Golden Age for computer architecture. Aided by open source ecosystems, agilely developed chips will convincingly demonstrate advances and thereby accelerate commercial adoption. The instruction set philosophy of the general-purpose processors in these chips will likely be RISC, which has stood the test of time. We envision the same rapid improvement as in the last Golden Age, but this time in cost, energy, and security as well as in performance.
Like the 1980s, the next decade will be exciting for computer architects in academia and in industry!
Bio:
David Patterson is a professor emeritus of Computer Science at UC Berkeley, a distinguished engineer at Google, and Vice-Chair of the Board of Directors of the RISC-V Foundation. He received his BA, MS, and PhD degrees from UCLA.
His most successful research projects were likely Reduced Instruction Set Computer (RISC), Redundant Array of Inexpensive Disks (RAID), and Network of Workstation (NOW). All three projects helped lead to multibillion-dollar industries. This research led to many papers and seven books, with the best known being Computer Architecture: A Quantitative Approach co-authored by John Hennessy, now in its sixth edition. His most recent book is The RISC-V Reader: An Open Architecture Atlas , co-authored by Andrew Waterman.
Patterson is a member of the National Academy of Engineering, the National Academy of Sciences, and the Silicon Valley Engineering Hall of Fame. His teaching was honored with the ACM Karlstrom Award and the IEEE Mulligan Medal. As a past president of ACM and a past Chair of CRA, he received Distinguished Service Awards from ACM, CRA, and SIGARCH and the Tapia Achievement Award for Scientific Scholarship, Civic Science, and Diversifying Computing.
His most recent award is the ACM A.M Turing Award, shared with John Hennessy, which is the highest award in computer science.
Пікірлер: 81
“Don't sacrifice your salary for your ‘incredibly valuable’ stock ’cause, you know, that may not work” = most practically useful quote of the lecture lol
He looks like walter white of computers
@cognominal
3 жыл бұрын
I am not in the meth empire, I am in the silicon empire.
@insanitysreign6195
3 жыл бұрын
@@cognominal *Stares intently into camera* My Instruction sets are pure.
@fpgamadeeasy
3 жыл бұрын
Dear Electronic engineer, please subscribe to my channel.
very good !!
amazing!! thank you
@roaaalsadeq3530
3 жыл бұрын
hello..can you help me to translate code from C language to RISC-V?
@32:16. About ARM. Chromebooks and MacBooks M1 are ARM64. Amazon AWS has ARM64 based EC2 servers. ARM64 is gradually replacing x86_64.
@bnsonline5099
3 жыл бұрын
I think in the same way, ARM is growing very fast and RISC-V too.
@qqryqq123
3 жыл бұрын
@@bnsonline5099 ARM is IP. As soon as RISC-V "matures" companies will jump on it and ARM might be in trouble. Not going to happen overnight but it's certainly closer than it was a few years ago.
@madmotorcyclist
3 жыл бұрын
@@bnsonline5099 RISC-V has one disadvantage. Because their licensing allows developers/manufacturers to add hardware extensions there will be incompatibilities that the 86x systems did not have. ARM doesn't have that issue. Also at the moment RISC-V is more power hungry for less performance than ARM chips now (~2-4 years behind).
@kayakMike1000
2 жыл бұрын
Specifically it's I think it's called aarchv8-a
@kayakMike1000
2 жыл бұрын
@@madmotorcyclist well .... If you're a developer, you could interrogate the risc-v hart for any built-in extensions when your program starts... You could then fail fast with a reasonable error message OR fail over to emulation of the non standard instructions at a performance cost.
I'm pretty surprised that this has only 800 likes. This is good information.
@julianskidmore293
Жыл бұрын
It's good, but it's very specialised. It's like asking how many copies of "Computer Architecture: A Quantitive Approach" have been bought (I have Edition 2, which I bought in 1996). It's never been a huge best-seller, because it's so specialised and most computing types (especially, ironically, hardware people), don't tend to care about computer architecture. Nevertheless, it's one of the most influential computing books of all time, because the right people have copies and they've put the ideas into practice for the benefit of all of us. Anyway, 1.3K likes so far!
@radivojevasiljevic3145
Жыл бұрын
@@julianskidmore293 I sometime think that 2nd edition nailed it and that it is best of all editions. Newer editions have newer examples, but sadly some things were moved to appendix.
Anybody developing a system should know at least something about the fields above and below (or before and after, depending on how you visualise processing stacks) their area. A software person needs to know something about the problem domain and about hardware. The hardware types need to understand software and materials science, and so on.
@lordmushroom723
Жыл бұрын
Agree 100%
Fred Brooks Jr gave a talk at SIGGRAPH once. He apologized for creating a notation that had syntactic significance for spaces...
Yeppp. Very Amazing.
Thanks a lot! Amazing lecture. Explains a lot about m1 from apple and trends in future cpu for servers today
In talk He said Berkeley students made the chip?Can Anyone tell me How did they did it ? In Rudimentary steps.For Research I need this Because It would be cool I can Do this.
59:10 is a typical "I bet you say that to all the girls" moment :)
Coolest ACM A. M. Turing Award laureate ever
Why is Walter white teaching Microprocessor. I thought he was a chemistry teacher
It’s been a little more than a year since this lecture and so much has changed already... CISC is practically doomed and RISC V keeps getting bigger and bigger...
@pentiumou8993
3 жыл бұрын
i just read his book on hardware software interface and get to be familiar with RISC-V, and came here to view some videos on it and finally come across the inventor’s lecture on computer architecture, and find that RISC is in such a development process right now...
@cat-.-
2 жыл бұрын
RISC-V still a couple years behind AArch (when it comes to microarch) and it's like triple the price. That said, this RISCV is valuable beyond imagination and it's on the right track!
@suntzu1409
2 жыл бұрын
"The longer the icon of RISC-V is on earth, the stronger it will become" "RISC-V's power double every year"
@mikafoxx2717
5 ай бұрын
@@cat-.-Thing is, it's much like arm in the first place. It was small and efficient but powerful enough, but not enough, power came with time and need. Now just look at what apple is doing with M3, it's seriously competing favorably with x86 and much more efficient. Risc-V will take time to get the high performance cores developed and desire to do so, which isn't hard considering it's free to use.
@cat-.-
5 ай бұрын
@@mikafoxx2717 I hope! But I don't see how, without major commercial adoption, RiscV could compete with arm and x86 in terms of uarch development. Vendors are very secretive and protective of their uarch. Oh, and, besides, it's becoming increasingly clear that we need a open ISA for GPU compute as well!
Please do at least 1080p. We're about to end the second decade of the 21st century.
@MaxPrehl
3 жыл бұрын
At least they got a decent mic on him. (those cuts to the audience mic aren't great tho)
@suntzu1409
2 жыл бұрын
Yeah plus 720p would have been fine if slides were bigger But they arent
good
timing attacks 20:10
I'm not convinced that Neural Networks are THE answer to AI. Adding massive computer power does not mean anything if its not the answer.
@arunavaghatak6281
3 жыл бұрын
Me too. Neural networks are not as data efficient as the human brain is. We need some radically different architecture for AI.
@bitflogger
3 жыл бұрын
@@arunavaghatak6281 Radical? How about zeros and ones, and Xs?
@schumachersbatman5094
3 жыл бұрын
@@arunavaghatak6281 But the brain is made of neural networks right? I don't really understand this criticism. People skeptical of neural networks, do they have a narrow definition of neural networks they don't like, one that excludes the biological brain? Or are they critiquing neural nets writ large, all variants, including 'graph neural networks' and other artificial models, as well as the human brain itself?
@youtubiuttoni
3 жыл бұрын
@bitflogger why not? They may not be YET. What’s the difference between artificial intelligence and intelligence? Neural networks are exactly what gave birth to intelligence in the first place...
@bitflogger
3 жыл бұрын
@@youtubiuttoni Nural Networks are an abstraction of the real thing. A good enough abstraction? I assume it would get better with time, but would it ever become good enough? I've read the NN go back 30 years, not yet? 30 years from now, not yet? Seems like fusion tech.
Great video! one thing though, I believe Mr Patterson actually drank out of someone else's water bottle!
Walther White shaved and got into computer science
@babeslash
Жыл бұрын
WW is already bald
Hello Mr. Patterson it's Aleisha Holt or Angel Olsen I'm still here ,no one is going brak my stride ... What does Google mean the circle ...
@fredsands7437
2 жыл бұрын
Damian Marley life is a circle ⭕
CISC was always intended for humans to use, RISC makes more sense if machine code is being generated by a computer
@brokula1312
3 жыл бұрын
You mean like compilers :)
@godofcows4649
3 жыл бұрын
@@brokula1312 yes!
@roaaalsadeq3530
3 жыл бұрын
hello..can you help me to translate code from C language to RISC-V?
@godofcows4649
3 жыл бұрын
@@roaaalsadeq3530 no, i'm sure there's a compiler for it for free somewhere
@crhu319
3 жыл бұрын
I must be a computer I hand coded 6502 assembler with just an 8x8 table of opcodes in 1980. Fight me.
Walter White
🤎💜
hello..can you help me to translate code from C language to RISC-V? أعجبني تعليق مشاركة
@fpgamadeeasy
3 жыл бұрын
Dear Electronic engineer, please subscribe to my channel.
@roberthindle5146
3 жыл бұрын
GCC
You're copying those big guys out there. The problem you got is those big guys aren't gonna tell you their Trade secrets. Did you tell of your Jujutsu.
So this guy voted from Biden. Ha! I'm smarter than him.