Lisa Su believes AI will soon be an integral part of chip design

[nanoguy]

In brief: Nvidia isn't the only company that wants to capitalize on the broad interest in AI tools for accelerating traditional workloads in a variety of industries. Like Team Green, the Lisa Su-powered AMD also wants a piece of the market for AI-fueled business tools and is already using AI internally to assist in designing future chips.

According to a recent survey, the so-called AI revolution is turning out to be quite depressing for many executives in the tech space. There are growing fears that machine-learning technologies are set to replace humans faster than we can retrain them, and it also doesn't help that companies big and small have been laying off thousands of workers, sometimes with the explicit intent to offload their work to generative artificial intelligence models.

Of course, not every tech leader is worried about AI. People like Nvidia CEO Jensen Huang are more than enthusiastic about its potential applications across a variety of fields of work. It's also no secret that Nvidia is well-positioned to sell most of the GPU shovels for the AI gold rush. AMD CEO Lisa Su is also a believer, albeit at the helm of a company that's only recently shifted more of its focus to developing hardware accelerators for artificial intelligence.

During the World Artificial Intelligence Conference in Shanghai, Su made a few remarks about the AI revolution and how AMD is leveraging advancements in this area for future product design. Su believes hardware engineering will soon require a larger set of skills to produce competitive chip designs. AMD CTO Mark Papermaster also confirmed that AMD is already using AI models in semiconductor design, testing, and verification, with plans to gradually incorporate it into its chip design workflow alongside traditional Electronic Design Automation (EDA) tools.

If the marketing language sounds familiar, it's because we've heard this before from Nvidia as well as every single EDA vendor looking to make their software suite more attractive for manufacturers. Like AMD, Team Green has a growing team of researchers and engineers dedicated to automating and improving chip design using AI tools, with a focus on solving hard problems like routing congestion and signal integrity as well as optimizing the most basic features of the transistor logic used in GPUs and other advanced silicon.

As for the hardware used to accelerate AI tasks as well as the software needed to take advantage of it, AMD has traditionally lagged behind Nvidia. Recently, Team Red has been making some progress in the hardware department with its Radeon Instinct products, but its ROCm software libraries have yet to see the kind of adoption that Nvidia's CUDA has.

Masthead credit: David Baillot | University of California San Diego

Permalink to story.

https://www.techspot.com/news/99351-lisa-su-believes-ai-soon-integral-part-chip.html

 

[wiyosaya]

IMO, like in the medical field, electronic design is a field where AI is likely to provide excellent results. As I see it, it is a good thing humans will be watching over any designs AI develops.

 

[kiwigraeme]

IMO, like in the medical field, electronic design is a field where AI is likely to provide excellent results. As I see it, it is a good thing humans will be watching over any designs AI develops.

Don't we already have a perfectly fine cheap system to farm out analysis to millions in developing countries.
grading those pennies per image workers , statistical analysis is all we need /s

For medical analysis , grading fruit , metal , wood commercial products - neural networks pretty good already - and lots of investment already

But yes AI preparing data , requesting data in a feedback loop with a large AI medical database is essential - all of us - not just bad police , or doctors can be struck on pursuing one path.

Doctors/Consultants do have access to an international group think - Hey this has me stumped - any ideas

 

[wiyosaya]

Don't we already have a perfectly fine cheap system to farm out analysis to millions in developing countries.
grading those pennies per image workers , statistical analysis is all we need /s

For medical analysis , grading fruit , metal , wood commercial products - neural networks pretty good already - and lots of investment already

But yes AI preparing data , requesting data in a feedback loop with a large AI medical database is essential - all of us - not just bad police , or doctors can be struck on pursuing one path.

Doctors/Consultants do have access to an international group think - Hey this has me stumped - any ideas

All I can say is that I hope that humanity makes it a priority to mandate that AI work for the best of humanity.

I cannot say whether the group think you speak of would have come to the same helpful conclusions that AI has come to in the medical cases where AI has helped. It takes time and effort for humans to put into something like that, and it may be an advantage that AI is, essentially, available on demand.

Where AI goes from here is anyone's guess at this point. Like all other technological advances, there are people who will use it for ill, and people who will use it for good.

However, if humanity is ever going to rid itself of those who would use something like this for ill, I think a fundamental change in society is needed. What that change is and how it should look is difficult to determine, and even if it is a change for the better, I think that there will be people who will be against it.

 

[wiyosaya]

@kiwigraeme Here's an interesting article that covers the subject you may want to read https://techxplore.com/news/2023-07-productivity-effects-chatgpt-college-educated-professionals.html
However, from the article

The researchers found that volunteers using ChatGPT took 40% less time to complete their assignments than those who did not use the app. They also found that those who used the app produced results that were judged to be 18% higher in quality. The researchers acknowledge that they did not conduct fact-checking on the writing produced by the volunteers; thus, it is not known if the increase in efficiency and quality came at the cost of accuracy.

 

[kiwigraeme]

@kiwigraeme Here's an interesting article that covers the subject you may want to read https://techxplore.com/news/2023-07-productivity-effects-chatgpt-college-educated-professionals.html
However, from the article

Cheers - what is annoying is news sites - that publish click bait titles about research - with no qualifiers - eg paid for by Big something , or data sample of 4 , or we took respondents claims at face value( they are better than average driver , if was not their fault ... ). Or even worse is published solely in Big Somethings monthly propaganda journal only.

Most people don't click link
Look at that latest Supreme Court case in USA - the Standing for the case was not even checked - for a wedding site web designer - The guy named was not gay , happily married and a web designer in a far off State

I currently destroying a report by a govt dept - for a while I skimmed over expert witness clause - The report failed Expert Witness testimony on the 3 most fundamental issues
eg note any additional tests etc
show a clear line of evidence ( all waffle) to conclusion
Give extensive reasoning it opinion changes - absolutely none give from first part of report 4 months earlier
I read the report and just attacking it as the POS it was - selective, prejudicial , vague , no real brief, remit - inconsistency with earlier report - sso foolishly missed expert witness clause