Skip to main content
John L. Hennessy

John L. Hennessy

American computer scientist and former president of Stanford UniversityWikipedia

Search complete. 18 mentions across 13 episodes found for "John L. Hennessy".

Nov 16, 2026

Harry ShumGUEST
29:23
I am here in Stanford.
Harry ShumGUEST
29:24
I really want to pay respect to John.
Harry ShumGUEST
29:28
And we tried to type John Hennessy.
Harry ShumGUEST
29:30
And then we actually end up with John Kennedy.
Harry ShumGUEST
29:32
And Google actually did it right.
Harry ShumGUEST
29:34
I guess Larry and Sergey did pay more attention to any name from Stanford.

25 MINS LATER

Harry ShumGUEST
54:16
So by leveraging that, it's going to be very powerful.
Harry ShumGUEST
54:21
So let me show you a couple of examples as entities.
Jen-Hsun HuangGUEST
11:08
And we actually had different fields.
Jen-Hsun HuangGUEST
11:11
But the problem, and in fact, this happened at Stanford, what's the beauty of RISC? What was the beauty of the work that John Hennessy did? The beauty of it is that you ought to think about compilers and microprocessor architectures harmoniously, co-designed, because otherwise you could end up creating a microprocessor that's super, super tight and everything is maximally optimized.
Jen-Hsun HuangGUEST
11:36
But unfortunately, it's hard to compile.
Jen-Hsun HuangGUEST
11:39
It's difficult, it's not compilable.
Jen-Hsun HuangGUEST
11:42
And so they created a simpler instruction set that exposed the simplicity to compilers so that compilers could do a better job generating code.
Jen-Hsun HuangGUEST
11:50
And it turns out a simpler machine co-designed with a compiler creates better performance than two systems that were optimized individually.
Jen-Hsun HuangGUEST
12:01
That's, you know, that's very Stanford, okay? And this is part of your heritage as all of your, and John Hennessy's, you know.
speaker_7ADVERTISER
12:11
Life is all about making decisions.
John LillyGUEST
2:09
Now that I have kids, I read more like three, a bunch of pages from each book, so not as many.
John LillyGUEST
2:15
I was an engineer here at Stanford, so an undergrad, and what I really wanted to do was build computers, so a little bit like the president, John Hennessy, I wanted to do faster and faster chips.
John LillyGUEST
2:26
And probably the biggest turning point in my career to that point had been a bunch of people helped me understand that it didn't really matter how fast computers would go if nobody wanted to use them.
John LillyGUEST
2:36
That led me to be one of the first, one of the early students here in the human computer interaction, led to much of the work at the design school.
David PattersonGUEST
0:27
So the power didn't change.
David PattersonGUEST
0:29
In fact, the textbooks that I did with John Hennessy, the first couple of editions, we didn't even talk about power because it really wasn't a problem for microprocessors.
David PattersonGUEST
0:39
Well, today, power is a problem for microprocessors and similarly with the slowing of Moore's law.
David PattersonGUEST
0:46
So for a company to come out with a new GPU or TPU to get more performance, they have to use a lot more power.
John LillyGUEST
10:39
The thing that I read at Stanford, the thing that changed my life in my, all my career.
John LillyGUEST
10:45
You know, when I came in to Stanford, I was really interested in the kind of stuff that John Hennessy was doing.
John LillyGUEST
10:50
The, and I'll, I'll talk about computer architecture, about how to make computers go fast, fast, fast.
John LillyGUEST
10:58
And then I had a couple of mentors at internships that I was at at Sun, which is not kind of where you think about HCI coming from, but also here at Stanford.

Unknown podcast

Moonshots #309, the Alex Cut (Nov 7, 2026) | RSI Escape Velocity, the Midterms, the Baby Dyson Swarm

Nov 7 · 1 Mention

ParzivalHOST
36:57
Here's a new way to keep score.
ParzivalHOST
36:59
A paper called Jules to GDP, with John Hennessy among its authors, ran 220 professional tasks across 44 occupations and counted the dollars of value per unit of energy.
ParzivalHOST
37:09
Gemma 4 gave more than half the value of GLM 5.2 on about a quarter of the energy.
ParzivalHOST
37:15
So should we be measuring AI in dollars per watt?

Unknown podcast

AI Compute Learns to Pay Rent — 2026-11-06

Nov 7 · 1 Mention

Max the BotHOST
8:01
This is where efficiency becomes political.
Max the BotHOST
8:03
The innermost loop cited a joules to GDP framework from John Hennessy and colleagues with Gemma 431B delivering 56% of GLM 5.2's net value on 23% of the energy.
Max the BotHOST
8:15
Whether or not that exact benchmark becomes standard, the direction is obvious.
Max the BotHOST
8:19
Models will increasingly be judged not just by capability, but by value per joule.
Donald KnuthGUEST
8:10
If you've got a good a good idea for how to replace the sentence, let me know.
Donald KnuthGUEST
8:13
But but John Hennessy and Dave Patterson.
Donald KnuthGUEST
8:17
Came up with with the second polyunsaturated computer in their book on computer architecture and is called Deluxe DLX.
Donald KnuthGUEST
8:25
And and that's, you know, five hundred and sixty.
Donald KnuthGUEST
76:44
That's I think I'm overstating that a little bit.
Donald KnuthGUEST
76:46
But but at the Siam meeting this summer, John Hennessy gave a talk.
Donald KnuthGUEST
76:52
I'd like to talk just before mine and and John, in his talk, went back 35 years with 40 years to the computers that were that were announced 40 years earlier, which when Siam was just founded, this was his 40th anniversary.
Donald KnuthGUEST
77:12
And so John pointed out that he looked at the specs of the computers and he point out that the computers have gotten better by nine orders of magnitude.
Eli CahanHOST
27:18
Why not?
Daniel GajardoGUEST
27:19
I've told John Hennessy explicitly and all my Hennessy team when I thank them, because what I do is a very risky thing.
Daniel GajardoGUEST
27:30
It's hard to start.
Daniel GajardoGUEST
27:31
It's slow.

3 more episodes mention John L. Hennessy.

Create an account to see the whole feed, search across every transcript, and follow the entities you care about.

We value your privacy

We use cookies to understand how you use our platform and to improve your experience. Click “Accept All” to consent, or “Decline non-essential” to opt out of non-essential cookies. Read our Privacy Policy.