AMD's Dr. Lisa Su Thinks That Moore's Law is Still Relevant - Innovation Will Keep Legacy Going

[T0@st]

Barron's Magazine has been on a technology industry kick this week and published their interview with AMD CEO Dr. Lisa Su on May 3. The interviewer asks Su about her views on Moore's Law and it becomes apparent that she remains a believer of Gordon Moore's (more than half-century old) prediction - Moore, an Intel co-founder passed away in late March. Su explains that her company's engineers will need to innovate in order to carry on with that legacy: "I would certainly say I don't think Moore's Law is dead. I think Moore's Law has slowed down. We have to do different things to continue to get that performance and that energy efficiency. We've done chiplets - that's been one big step. We've now done 3-D packaging. We think there are a number of other innovations, as well." Expertise in other areas is also key in hitting technological goals: "Software and algorithms are also quite important. I think you need all of these pieces for us to continue this performance trajectory that we've all been on."

When asked about the challenges involved in advancing CPU designs within limitations, Su responds with: "Yes. The transistor costs and the amount of improvement you're getting from density and overall energy reduction is less from each generation. But we're still moving (forward) generation to generation. We're doing plenty of work in 3 nanometer today, and we're looking beyond that to 2 nm as well. But we'll continue to use chiplets and these type of constructions to try to get around some of the Moore's Law challenges." AMD and Intel continue to hold firm with Moore's Law, even though slightly younger upstarts disagree (see NVIDIA). Dr. Lisa Su's latest thoughts stay consistent with her colleague's past statements - AMD CTO Mark Papermaster reckoned that the theory is pertinent for another six to eight years, although it could be a costly endeavor for AMD - the company believes that it cannot double transistor density every 18 to 24 months without incurring extra expenses.



View at TechPowerUp Main Site | Source

 

[bonehead123]

the company believes that it cannot double transistor density every 18 to 24 months without incurring extra expenses.

Soooooo.... apparently she is saying that they will be able to continue making some advances in chip tech, but ONLY if she stops spending so much money on her anti-jacket man jackets

 

[RH92]

That's cool for Lisa but what the market needs is less talk and more acts and preferably not only on the CPU side but also on the GPU side !

 

[Paranoir andando]

Lisa "undershipping" Su.
She don't want to sell. How does this affect to the progress?

 

[Daven]

"I would certainly say I don't think Moore's Law is dead. I think Moore's Law has slowed down.”

The whole law is based on a time premise. That would be equivalent to saying all planets have the same gravity except objects fall faster or slower depending on the planet.

Sorry Lisa but you have no idea what you are talking about and Moore’s law was just a product roadmap timetable for the marketing department. Nothing more.

 

[DemonicRyzen666]

"Moore's Law" was never a law.
It's a theory.
Too many outliers to be a classified as a "true law".
It's a "pseudo law" excepted by the tech as a real law.

 

[DrCR]

excepted by the tech as a real law.

Except, while not dead, that acceptance has slowed down

 

[Daven]

"Moore's Law" was never a law.
It's a theory.
Too many outliers to be a classified as a "true law".
It's a "pseudo law" excepted by the tech as a real law.

It is not a theory either. Its just an Intel product roadmap. Its not part of the scientific method. Strategies for building a human invention over time do not classify as scientific inquiry. It is classified under capitalism and profits.

A scientific theory is knowledge about how nature works that has been exhaustively proven through experimentation. A theory becomes law when it is 100% certain that the natural phenomenon occurred or occurs in the way predicted. The timetable for burning grooves into a substrate is NOT part of this. Moore’s law is like Murphy’s law. The use of law here is just jargon.

Here are some examples of scientific laws and theories:

10 Scientific Laws and Theories You Really Should Know

Even if you don't operate a scanning electron microscope all that frequently, you might want to brush up on these 10 scientific laws and theories. After all, they're some of our best tools for describing how the universe works.

science.howstuffworks.com

 

[Chry]

If we liberally interpret Moore's Law, taking is doubling of performance, then it kind of still applies if we consider best case performance scenarios using latest instruction sets and whole system's performance. Which in the end is what matters most.

 

[TumbleGeorge]

It is not a theory either. Its just an Intel product roadmap. Its not part of the scientific method. Strategies for building a human invention over time do not classify as scientific inquiry. It is classified under capitalism and profits.

A scientific theory is knowledge about how nature works that has been exhaustively proven through experimentation. A theory becomes law when it is 100% certain that the natural phenomenon occurred or occurs in the way predicted. The timetable for burning grooves into a substrate is NOT part of this. Moore’s law is like Murphy’s law. The use of law here is just jargon.

Here are some examples of scientific laws and theories:

10 Scientific Laws and Theories You Really Should Know

Even if you don't operate a scanning electron microscope all that frequently, you might want to brush up on these 10 scientific laws and theories. After all, they're some of our best tools for describing how the universe works.

science.howstuffworks.com

I disagree. Humans are a natural product. From a philosophical point of view, nature uses us to create development that cannot happen directly. We are only working tools.

 

[Wirko]

Rock's law or Moore's second law, named for Arthur Rock or Gordon Moore, says that the cost of a semiconductor chip fabrication plant doubles every four years.

Maybe we should also be interested in this one. And I don't see it slowing down.

 

[Paganstomp]

Vision has the answers

 

[kapone32]

I remember creating a post that established that in some cases the 5600 was in some cases 100% faster than the 1600 but some people refused to believe it.

 

[R-T-B]

Anyone who thought a commercial thing like moores law was an actual law of nature needs to retake high school science.

Thankfully, I don't think many here were confused on that point.

 

[eidairaman1]

Anyone who thought a commercial thing like moores law was an actual law of nature needs to retake high school science.

Thankfully, I don't think many here were confused on that point.

Who coined it a law to begin with? It was a Theory, I mean we haven't received 10GHz CPUs either.

 

[Daven]

Who coined it a law to begin with? It was a Theory, I mean we haven't received 10GHz CPUs either.

Again its not a theory. Theories lead to laws. Saying its one also says it could become the other.

Gordon Moore was a founder of an influential company. Because of that, people listened to his thoughts and opinions. One of his opinions was that Intel could make products at a certain pace. For awhile, that was the case. Now its not. Everyone needs to stop elevating the musings of one person to some sort of fundamental, universal process.

 

[Dr. Dro]

Again its not a theory. Theories lead to laws. Saying its one also says it could become the other.

Gordon Moore was a founder of an influential company. Because of that, people listened to his thoughts and opinions. One of his opinions was that Intel could make products at a certain pace. For awhile, that was the case. Now its not. Everyone needs to stop elevating the musings of one person to some sort of fundamental, universal process.

I mean, are you absolutely, positively sure that they can not? Keeping mindful that processors are more advanced than ever and their cost hasn't gone dramatically up like GPUs have? Or observing what a processor such as Raptor Lake brings to the table compared to Rocket Lake which released only 3 years ago? Even intra-node, Raptor offers a lot on top of Alder, not enough to make an Alder owner switch or upgrade, but it brought the i9-12900KS (which is still a mighty processor I may add) to the Core i7 tier. If anything, the i7-13700K is a little faster than the 12900KS.

It's the same problem that NVIDIA is facing. The Ada architecture is leaps and bounds ahead of Ampere. This is true, undisputable fact. So why are the 40 series product stack so bad? The answer isn't simple greed, I'm afraid. I'm not gonna dispute that there is some of that (there is - Jensen Huang isn't the kind of guy who will openly forfeit his cut), but their production and R&D costs went up harshly, so they had little choice in there but to dramatically increase the product's cost while simultaneously releasing high-yield, cutdown products that barely - when at all - offer an increase in performance per dollar. In fact, across many segments, it's an actual regression in that regard. For this reason, I wouldn't be surprised if the Blackwell architecture was actually a refinement of Ada focused on upkeeping performance but lowering cost throughout, then making use of configurations that would be possible with Ada but were not used due to yield and profitability concerns.

 

[DemonicRyzen666]

Again its not a theory. Theories lead to laws. Saying its one also says it could become the other.

Gordon Moore was a founder of an influential company. Because of that, people listened to his thoughts and opinions. One of his opinions was that Intel could make products at a certain pace. For awhile, that was the case. Now its not. Everyone needs to stop elevating the musings of one person to some sort of fundamental, universal process.

That's even worse than what I said then. That means it's all marketing lingo to drive sales & everyone believes it's a Frikken "law", when it's nothing but a scheme to sell cpus.
I actually feel the same way about the guy who made claims about 1% low & 0.1% lows. The guy who claims to shed light on all that was in marketing well before, he did anything like reviews.

 

[R-T-B]

Who coined it a law to begin with? It was a Theory, I mean we haven't received 10GHz CPUs either.

Law and theory are just words. They have no meaning outside of their applicable fields, and Moore never proposed his law to be a natural one, only a commercial prediction. People here are abusing that definition with gusto.

Any stock market "law" is not the same as a natural law and everyone should know that.

Also, scientific theories are signifigantly more grounded then that of any commercial prediction. Don't give Moore's law that kind of credibility, it really shows you have no understanding of what you are saying.

 

[JustBenching]

Law and theory are just words. They have no meaning outside of their applicable fields, and Moore never proposed his law to be a natural one, only a commercial prediction. People here are abusing that definition with gusto.

Any stock market "law" is not the same as a natural law and everyone should know that.

Also, scientific theories are signifigantly more grounded then that of any commercial prediction. Don't give Moore's law that kind of credibility, it really shows you have no understanding of what you are saying.

Even natural laws are malleable, they are not set in stone. They are descriptive after all, not prescriptive.

 

[Vayra86]

I mean, are you absolutely, positively sure that they can not? Keeping mindful that processors are more advanced than ever and their cost hasn't gone dramatically up like GPUs have? Or observing what a processor such as Raptor Lake brings to the table compared to Rocket Lake which released only 3 years ago? Even intra-node, Raptor offers a lot on top of Alder, not enough to make an Alder owner switch or upgrade, but it brought the i9-12900KS (which is still a mighty processor I may add) to the Core i7 tier. If anything, the i7-13700K is a little faster than the 12900KS.

It's the same problem that NVIDIA is facing. The Ada architecture is leaps and bounds ahead of Ampere. This is true, undisputable fact. So why are the 40 series product stack so bad? The answer isn't simple greed, I'm afraid. I'm not gonna dispute that there is some of that (there is - Jensen Huang isn't the kind of guy who will openly forfeit his cut), but their production and R&D costs went up harshly, so they had little choice in there but to dramatically increase the product's cost while simultaneously releasing high-yield, cutdown products that barely - when at all - offer an increase in performance per dollar. In fact, across many segments, it's an actual regression in that regard. For this reason, I wouldn't be surprised if the Blackwell architecture was actually a refinement of Ada focused on upkeeping performance but lowering cost throughout, then making use of configurations that would be possible with Ada but were not used due to yield and profitability concerns.

Nice story & theory.

Practice - 65~55% profit margins

 

[david salsero]

What you have to do is produce more AMD ZEN 4 7040 Phoenix = DDR5 + RDNA 3 + USB 4.0 + HDMI 2.1 + AI with XDNA architecture developed by Xilinx and all at 4nm vs 10nm Intel.
You have the best processor in history and we still can't enjoy any ultrabook.
You have the opportunity to offer ultrabook under 2.2lb without heating up and performance for AAA gaming and we don't see it in the market and people are getting impatient.

 

[Daven]

The main way to know if words like law and theory should be used is experimentation. Gordan Moore didn’t do any experimentation. He just stated an opinion/prediction based on his knowledge of Intel manufacturing practices at the time.

If a scientist observes something in the environment, a hypothesis is made and a set of experiments are designed to prove the hypothesis. If the experiments generate a reproducible result then a theory might be postulated. After an exhausting number of reproducible experiments from a multitude of scientists demonstrate that the theory holds up, a law will be created.

Sorry for belaboring this point but I’m a scientist and the scientific method keeps getting abused by tech companies, tech news sites and CEOs that have lost perspective. I just wish everyone would stop calling it Moore’s law and call it something like Moore’s prediction. I’m guessing the word law was used to make it seem Intel discovered something significant rather than just come up with a general product release schedule.

 

[TumbleGeorge]

This, of course, is not scientific, but a marketing speech by Mr. Gordon Moore. Which is additionally abused by other persons and companies.

 

[Wirko]

Ah, Moore's law has finally met the law of diminishing returns. The latter also doesn't care if it's called theory or prediction or premise or hyporhesis, it just... holds true.