#solidstatelife

waynerad@diasp.org

"No more nanometers: It's time for new node naming", says Kevin Morris.

"Intel held the line from '10 micron' in 1972 through '0.35 micron' in 1995, an impressive 23-year run where the node name matched gate length. Then, in 1997 with the '0.25 micron/250 nm' node they started over-achieving with an actual gate length of 200 nm -- 20% better than the name would imply. This 'sandbagging' continued through the next 12 years, with one node (130nm) having gate length of only 70nm -- almost a 2x buffer. Then, in 2011, Intel jumped over to the other side of the ledger, ushering in what we might call the 'overstating decade' with the '22nm' node sporting a gate length of 26 nm. Since then, things have continued to slide further in that direction, with the current '10nm' node measuring in with a gate length of 18 nm -- almost 2x on the other side of the 'named' dimension."

"So essentially, since 1997, the node name has not been a representation of any actual dimension on the chip, and it has erred in both directions by almost a factor of 2."

"For the last few years, this has been a big marketing problem for Intel from a 'perception' point of view. Most industry folks understand that Intel's '10nm' process is roughly equivalent to TSMC and Samsung's '7nm' processes. But non-industry publications regularly write that Intel has 'fallen behind' because they are still back on 10nm when other fabs have 'moved on to 7nm and are working on 5nm.' The truth is, these are marketing names only, and in no way do they represent anything we might expect in comparing competing technologies."

He mentions a proposal from Philip Wong of TSMC. His proposal is you replace "nanometers" with 3 numbers that reflect the actual usefulness of the chip: density of logic transistors in number per square millimeter, density of off-chip DRAM memory in bits per square millimeter, and density of connections between the memory and the logic transistors, in number of interconnects per square millimeter. So a chip could be described as: "DL: 38M, DM: 383M, DC: 12K", or just "38M, 383M, 12K".

That proposal was made in 2020 and so far, hasn't caught on. Maybe it's tough to market as it's not as exciting as a single single-digit number like "5 nm"? But "nanometers" needs to be replaced by something.

I looked online for a website with nice tables of those three numbers for all the recent chips. Couldn't find it.

It looks like these metrics are often kept secret by semiconductor companies to try to get a competitive advantage, but people say there are websites that have information on new chips and the underlying technology, such as WikiChip, SemiWiki, AnandTech, Tom's Hardware, IEDM, VLSI Symposium, IEEE Xplore, EE Times, Nature Electronics, IEEE International Roadmap for Devices and Systems, Semiconductor Industry Association, and so on. (Links to some of these below. I haven't had time to delve too deeply into any of these sites.)

No more nanometers: It's time for new node naming

#solidstatelife #semiconductors

waynerad@diasp.org

What does the election mean for AI? Matt Wolfe says: The upcoming Trump Administration will repeal the Biden Administration's AI regulation executive order, which required developers of foundation models to report their safety tests to the federal government, and in general the upcoming Trump Administration will oppose regulation of AI, seeking for AI to advance as fast as possible so the US will stay ahead of geopolitical competitors. "Make America First in AI". The upcoming Trump Administration might even fund AI "Manhattan Projects" to further speed up AI advancement. J.D. Vance is an advocate for open source AI models. Elon Musk runs X.AI and will be working with the upcoming Trump Administration.

That's the first 6 minutes of the video. The rest is about other stuff: Runway AI's new camera control feature, Kling Face Swap, ByteDance's new X-Portrait 2, Facepoke, BlackForestLabs's Labs FLUX 1.1, Krea.AI Loras (character models), Anthropic PDF reading, Anthropic's Haiku price increase, partnerships with AI companies and the US military/defense industry, Instagram age verification, OpenAI possibly wanting to get into hardware, GPT-4o Predicted Outputs, Prime Video AI recaps, iOS 18.2 new AI features, bolt.new builds a Tetris game from a prompt, Wendy's working with Palentir for supply chain AI, SingularityNET AI that plays Minecraft, Nvidia robotics simulation tools, and a Unitree walking robot and robot dog that a guy tries to beat the heck out of.

#solidstatelife #ai #domesticpolitics #election

https://www.youtube.com/watch?v=pt2qXsURflg

waynerad@diasp.org

"I asked the AI to put together a lesson plan on the Great Gatsby for high school students, breaking it into readable chunks and then creating assignments and connections tied to the Common Core learning standard. I also asked it to put this all into a single spreadsheet for me. With a chatbot, I would have needed to direct the AI through each step, using it as a co-intelligence to develop a plan together. This was different. Once given the instructions, the AI went through the steps itself: it downloaded the book, it looked up lesson plans on the web, it opened a spreadsheet application and filled out an initial lesson plan, then it looked up Common Core standards, added revisions to the spreadsheet, and so on for multiple steps. The results are not bad (I checked and did not see obvious errors, but there may be some -- more on reliability later int he post). Most importantly, I was presented finished drafts to comment on, not a process to manage. I simply delegated a complex task and walked away from my computer, checking back later to see what it did (the system is quite slow)."

"But to get a little bit better sense of the limits of the system, I tested it on a game, Paperclip Clicker, which, ironically, is about an AI that destroys humanity in its single-minded pursuit of making paperclips."

Feels like a glimpse of the future.

When you give a Claude a mouse

#solidstatelife #ai #genai #llms #agenticai

waynerad@diasp.org

Generative AI is being added to Notepad.exe. No, I'm not making this up.

"With this update, we are introducing the ability to rewrite content in Notepad with the help of generative AI. You can rephrase sentences, adjust the tone, and modify the length of your content based on your preferences to refine your text."

And MS Paint.

New AI experiences for Paint and Notepad begin rolling out to Windows Insiders

#solidstatelife #ai #genai #llms #computervision

waynerad@diasp.org

"I gave AI control of my computer and asked it to 'solve homework 1 of Stanford discrete math class (Math 61DM)'."

"It found the problem set, downloaded Latex, solved every question, and compiled it to a PDF... in FIVE minutes."

"Will any college student ever do homework again?"

#solidstatelife #ai #genai #llms #multimodal #agenticai

https://twitter.com/deedydas/status/1851802538443706417

waynerad@diasp.org

"Why Recursion Pharmaceuticals abandoned cell painting for brightfield imaging."

"The bet of the whole company is the following: take a drug (either from a chemical screening library or dreamt up by a model), wash it over a plate of cells, visually observe how the cells react to it, and repeat this a few billion times across many different cells + many genetic knockouts of those cells (which acts as a model of a disease) + many different drugs. The whole idea is also known as phenomics -- the study of the visual phenotype of organisms. In this case, the 'organism' is a small set of cells. Once you've done that, train a model on those petabytes of cellular images, along with the genetic or chemical perturbation applied."

"The Recursion bet is that the resulting model would acquire an extremely strong understanding of the interaction between the visual morphology of cells, their genetic makeup, and drugs applied to them -- predicting the rest given one (or two) of the others."

"But this essay isn't about Recursion's drug discovery strategy." "This essay is about how Recursion takes pictures of cells in the first place, why it (officially) changed its approach just a few months ago, and why I think the decision to change it is a lot more interesting than people think."

"The superiority of using raw, unaltered cell painting images as input over hand-crafted features was established by 2019, but, as time went on, the utility of cell painting at all also became suspect. Per a 2022 Scientific Reports paper, one could nearly perfectly predict what the cell painting image would be given a brightfield image."

The article explains that "brightfield" microscopy was what was originally just "looking through a microscope": you put the cell you want to observe on a slide and shine bright light on it from behind and look at it through the microscope. This in contrast with "cell painting" where dyes are used to differentiate between RNA, endoplasmic reticulum, mitochondria, cytoskeleton/cell-membrane, and the cell nucleus.

"Another example of the bitter lesson; pre-imposing structure on your data is useful for a human, but detrimental to a machine."

"But there is another side benefit of going the brightfield route: the ability to do time-lapse microscopy."

Why Recursion Pharmaceuticals abandoned cell painting for brightfield imaging

#solidstatelife #biotech #microscopy

waynerad@diasp.org

"Inside 'Project Rodeo,' the Tesla effort pushing the limits of self-driving technology."

"Operating on open streets with other vehicles, cyclists, and pedestrians, test drivers on Project Rodeo have tested unreleased software that will be crucial to Tesla's push into autonomous driving."

"Test drivers said they sometimes navigated perilous scenarios, particularly those drivers on Project Rodeo's 'critical intervention' team, who say they're trained to wait as long as possible before taking over the car's controls. Tesla engineers say there's a reason for this: The longer the car continues to drive itself, the more data they have to work with."

Inside 'Project Rodeo,' the Tesla effort pushing the limits of self-driving technology

#solidstatelife #ai #computervision #autonomousvehicles #tesla

waynerad@diasp.org

"VibMilk: Nonintrusive milk spoilage detection Via smartphone vibration."

VibMilk "utilizes the ubiquitous vibration motor and inertial measurement unit (IMU) of off-the-shelf smartphones. The method detects spoilage based on the fact that the milk's physical properties change, inducing different vibration responses at various stages of degradation. Using the InceptionTime deep learning model, VibMilk achieves 98.35% accuracy in detecting milk spoilage across 23 different stages, from fresh (pH = 6.6) to fully spoiled (pH = 4.4)."

Alrighty then. Yet another thing I didn't know is possible but is.

VibMilk: Nonintrusive milk spoilage detection Via smartphone vibration

#solidstatelife #ai #imu

waynerad@diasp.org

"How often do you use artificial intelligence in your role?"

Asks Gallup poll.

67% say "never", 5% say "less often than once a year", 2% say "once a year", 7% say "a few times a year", 8% say "a few times a month", 7% say "a few times a week", 4% say "daily".

But the same poll shows 1/3rd of organizations are taking action on AI. 44% for "white-collar" jobs. So it looks like leaders want to boost productivity and innovation, but workers are not using AI much and not delivering on that expectation.

The 3rd question is, of people who do use AI, what do they use it for?

Mostly "to generate ideas" and "to consolidate information or data" -- but leaders and managers more than individual contributors. Next is "to automate basic tasks", which is used more by individual contributors than leaders and managers.

AI in the Workplace: Answering 3 Big Questions

#solidstatelife #ai #genai #technologicalunemployment

waynerad@diasp.org

WalkON Suit F1 exoskeleton demonstration. A person unable to walk and in a wheelchair is able to transfer directly into the exoskeleton and walk using the exoskeleton.

The WalkON Suit F1 exoskeleton was developed by the Korea Advanced Institute of Science and Technology (KAIST) Exoskeleton Laboratory and Angel Robotics.

WalkON Suit F1: The Next-Gen Exoskeleton That Walks Itself - Exoskeleton Lab @ KAIST

#solidstatelife #robotics #exoskeletons

waynerad@diasp.org

"If you had any idea that satellite connectivity isn't a key part of Apple's strategy, well, the company's satellite partner Globalstar has disclosed changes to its deal with Apple, including a new influx of $1.1 billion from Apple tied to capital improvements, and $400 million in equity, which gives Apple a 20-percent stake in the company."

"Apple's been offering satellite connectivity since the introduction of the iPhone 14 line in 2022, which debuted with the Emergency SOS via Satellite feature."

Apple sinks $1.1 billion into Globalstar's satellite network, takes ownership stake

#solidstatelife #satellite #satellitecommunications #apple

waynerad@diasp.org

"Four futures for cognitive labor."

"First: the printing press. If you were an author in 1400, the largest value-add you brought was your handwriting." "With hindsight we see that even as all the terrifying extrapolations of printing automation materialized, the income, influence, and number of authors soared."

"Second: the mechanization of farming." "The per-capita incomes of farmers have doubled several times over but there are many fewer farmers, even in absolute numbers."

"Third: computers. Specifically, the shift from the job title of computer to the name of the machine that replaced it. " "This industry was replaced by a new industry of people who programmed the automation of the previous one."

"Finally: the ice trade. In the 19th and early 20th centuries, before small ice machines were common, harvesting and shipping ice around the world was a large industry employing hundreds of thousands of workers." "By WW2 the industry had collapsed and been replaced by home refrigeration."

Four futures for cognitive labor

#solidstatelife #ai #genai #agi #technologicalunemployment

waynerad@diasp.org

25% of code written at Google is produced by AI. The question, well, 2 questions: 1) is this code as good as before, when code was all human-written? And 2) is Google actually making any good products? If not, they're just making products that aren't any good 25% faster. But Google kills lots of products so maybe they can kill products 25% faster, too?

I guess Gemini is pretty good -- what's not so good is trying to ram it into everything. I don't know about you but I skip those "AI summaries" in search results -- I go directly to Gemini when I want an "AI summary". Google Search itself doesn't seem to be getting any better, although, I know people say it's getting worse, but to me it seems like it's usually good enough -- I still use it. I'm getting text messages from Gemini on my phone -- it wants to help me write messages. I don't want someone else, AI or no, telling me what to say, tho. Who wants this? If you like and use this stuff, comment below.

There's also the issue of the 25% itself. At my work there is tremendous pressure for programmers to vastly increase productivity by using AI tools. Subjectively it seems like the expectation industrywide is about 5x, although at my work nobody has said 5x specifically. 25% is 1.25x, nowhere near 5x.

Sundar Pichai said on the earnings call today that more than 25% of all new code at Google is now generated by AI.

#solidstatelife #ai #genai #llms #codingai #google

waynerad@diasp.org

Export controls not working on semiconductor equipment to China.

"Huawei has been able to scoot around TSMC's weak end customer checks and secure thousands of leading edge wafers through Bitmain / Sophgo, and many other established Chinese design firms. While this is a huge failure in enforcement of the export controls, it's also relatively low volume compared to domestic fabrication of the Ascend 910B. Huawei has been relying on primarily Semiconductor Manufacturing International Corporation (SMIC) to produce their domestic AI chip and they have run tens of thousands of wafers of the main compute chiplet on their domestic SMIC N+2 (~7nm) and N+3 (~6nm) process nodes."

"SMIC produces 7nm-class chips including the Kirin 9000S mobile system-on-chip (SoC) and Ascend 910B AI accelerator. Two of their fabs are connected via wafer bridge, such that an automated overhead track can move wafers between them. For production purposes, this forms a continuous cleanroom and effectively one fab. But for regulatory purposes, they are separate! One building is entity-listed by the US and working on advanced logic for AI chips, a clear national security concern. The other is free to import 'dual use' tools as it runs only 'legacy processes.' Do you believe they aren't sharing anything over the wafer bridge?"

"An isotropic etch chamber, essential to producing the latest 2nm Gate-All-Around transistors, cannot be exported to China from Lam's US factories. This same etch chamber, manufactured in Lam's Malaysia facility, can legally be sold to an advanced logic fab in China if no US persons are involved (in manufacturing, sales, installation, and servicing). This includes even customers on the US entity list. Other companies follow the same playbook, including Applied Materials & KLA with their Singapore facilities."

"ChangXin Memory Technologies (CXMT)'s advanced DRAM with 18nm half pitch was subject to the original 2022 export controls. A new method of calculating half-pitch in the 2023 rules put them back above the minimum line for controls. Without changing the underlying process, they went from restricted to not. They also changed the node name from 17nm or 18.5nm to 19nm, to avoid any appearance of impropriety. The subtleties of the rule change that so specifically moved them from under to a literal nanometer or two over the restriction are, at least, very fortunate for CXMT. Lobbying efforts by giants such as Applied Materials who has made over $3 billion from CXMT may be noteworthy."

"Pengjin High-Tech, a gallium nitride (GaN) startup that is not entity-listed, is building its cleanroom across the street from entity-listed advanced logic producer Peng Xin Wei (PXW). Pengjin is free to import nearly all advanced Western production equipment, including equipment critical to producing advanced logic at 7nm or below (again the exceptions are a small number of tools on the control list, including extreme ultraviolet (EUV))."

Fab whack-a-mole: Chinese companies are evading US sanctions

#solidstatelife #semiconductors #geopolitics

waynerad@diasp.org

Gerrymandering with simulated annealing with Monte Carlo Markov chains. Found this video from 3 years ago. Basically you randomly flip pixels on a map, then, as your simulated "temperature" decreases, you decrease the randomness of how much you run with the random pixel flips regardless of whether they increase whatever gerrymandering metric you put in, and increase the degree to which the map zeroes in on a solution that's close to optimal. Your optimization function takes into account such things as to what degree funky shapes are allowed, populations of all districts are close, and of course, the degree to which the final outcome of the state is proportional to the population in terms of political parties, or whether it gives one party or the other disproportionate representation.

Algorithmic Redistricting: Elections made-to-order - AlphaPhoenix

#solidstatelife #domesticpolitics #gerrymandering #simulatedannealing

waynerad@diasp.org

"Russia says it might build its own Linux community after removal of several kernel maintainers."

The Russian Ministry of Digital Development said it is planning to create its own Linux community, which "will unite developers from those countries that are ready to work with Russia." (Me quoting the Google Translate version of the Russian text, see the rbc.ru link below).

"Russia's response came after the Linux community blocked 11 Russians from maintaining the Linux kernel -- the operating system's core code -- citing 'various compliance requirements.' Linux creator Linus Torvalds stated that this decision 'is not getting reverted,' adding that as a Finn, he will not 'support Russian aggression.'" (This time quoting the English news article, which provides some context.)

"One of the Linux maintainers later explained that the restrictions would apply to developers whose companies are owned or controlled by entities on the US Office of Foreign Assets Control list, designated as involved in activities that 'threaten the national security, foreign policy, or economy' of the country."

Russia says it might build its own Linux community after removal of several kernel maintainers

#solidstatelife #linux #geopolitics #ukraineconflict

waynerad@diasp.org

Grunty is a "self-hosted desktop app to have AI control your computer, powered by the new Claude computer use capability. Allow Claude to take over your laptop and do your tasks for you (or at least attempt to, lol). Written in Python, using PyQt."

"If it wipes your computer, sends weird emails, or orders 100 pizzas... that's on you."

Grunty

#solidstatelife #ai #llms #genai #agenticai #anthropic

waynerad@diasp.org

"GenAI's dirty secret: It's set to create a mountainous increase in e-waste."

The study, which looks at the rate AI servers are being introduced to datacenters, claims that a realistic scenario indicates potential for rapid growth of e-waste from 2.6 kilotons each year in 2023 to between 400 kilotons and 2.5 million tons each year in 2030, when no waste reduction measures are considered.

The team "considered four scenarios with varying degrees of generative AI production and application, ranging from an aggressive scenario with widespread applications to a conservative scenario with more specific applications."

GenAI's dirty secret: It's set to create a mountainous increase in e-waste

#solidstatelife #environment #pollution

waynerad@diasp.org

"TSMC isn't a pure monopsony in the wafer fab equipment market. Intel and Samsung are still buying tools, just not as often as wafer fab equipment manufacturers would like. Memory manufacturers also buy wafer fab equipment tools, and trailing-edge foundries do too."

"Monopsony refers to a market in which there is only a single buyer, i.e., producers cannot find alternative buyers of their product."

"Monopsony permits the buyer to establish prices, terms and conditions that are quite different from those that would result from a market structure in which there were many competing buyers and sellers."

The only seller the article mentions is ASML. Single-seller market, too? At least when it comes to high-numeric-aperture (high-NA) extreme ultraviolet (EUV) lithography tools.

"When there are only a few buyers, it's an oligopsony. Given the consolidated nature of the semiconductor industry, most markets within the semiconductor supply chain are already oligopsonistic."

Monopsony and TSMC

#solidstatelife #semiconductors

waynerad@diasp.org

OpenAI o1 isn't as good as an experienced professional programmer, but... "the set of tasks that O1 can do is impressive, and it's becoming more and more difficult to find easily demonstrated examples of things it can't do."

"There's a ton of things it can't do. But a lot of them are so complicated they don't really fit in a video."

"There are a small number of specific kinds of entry level developer jobs it could actually do as well, or maybe even better, than new hires."

Carl of "Internet of Bugs" recounts how he spent the last 3 weeks experimenting with the o1 model to try to find its shortcomings. /

"I've been saying for months now that AI couldn't do the work of a programmer, and that's been true, and to a large extent it still is. But in one common case, that's less true than it used to be, if it's still true at all."

"I've worked with a bunch of new hires that were fresh out with CS degrees from major colleges. Generally these new hires come out of school unfamiliar with the specific frameworks used on active projects. They have to be closely supervised for a while before they can work on their own. They have to be given self-contained pieces of code so they don't screw up something else and create regressions. A lot of them have never actually built anything that wasn't in response to a homework assignment.

"This o1 thing is more productive than most, if not all, of those fresh CS graduates I've worked with.

"Now, after a few months, the new grads get the hang of things, and from then on, for the most part, they become productive enough that I'd rather have them on a project than o1."

When I have a choice, I never hire anyone who only has an academic and theoretical understanding of programming and has never actually built anything that faces a customer, even if they only built it for themselves. But in the tech industry, many companies specifically create entry-level positions for new grads."

"In my opinion, those positions where people can get hired with no practical experience, those positions were stupid to have before and they're completely irrelevant now. But as long as those kinds of positions still exist, and now that o1 exists, I can no longer honestly say that there aren't any jobs that an AI could do better than a human, at least as far as programming goes."

"o1 Still has a lot of limitations."

Some of the limitations he cited were writing tests and writing a SQL RDBMS in Zig.

ChatGPT-O1 Changes Programming as a Profession. I really hated saying that. - Internet of Bugs

#solidstatelife #ai #genai #llms #codingai #openai #technologicalunemployment