• LIVE

  • U.S. Inflation Data

    15m ago

    15m ago

  • Russia-Ukraine War

    1h ago

    1h ago

Seems like the MBAs got their dirty little hands on the company's direction, went for buying products offshore for greater profit and gutted R&D.

Intel for a long time was managed by CEO with marketing background, while AMD and NVDA are managed by engineers.

Well, if nothing else the competition from the likes of AMD, who are clearly eating Intel's lunch right now, is a good thing. Their paradigm shifting chips and underlying architecture sure forced them Intel to up their game. A little, lot, less complacency on their part is never a bad thing in business. You can't sit on your laurels forever. See GM vs Tesla by way of an example in a different industry. Anyway, so now it's game ON!? I say bring it because ultimately, it's the consumer who benefits. At least so long as the thing doesn't crank out the heat of a toaster oven and the like. Or chips get down to 1nm lines at which point it becomes a case of "Okay, NOW what!?" Heh! John~ American Net'Zen

It's ashamed that there's so few Americans who want to become engineers and do the hard work that previously made america lead the world. Now with everyone getting lazier and more focused on $, I tell kids today to forget science; get a degree in finance and work in leveraged buyouts. You'll be sucking the money and lifeblood out of companies, but you'll get hugely wealthy.

So, here is an anathema (in certain lib circles) point, coming out of my 20 years of teaching engineering at a certain top 5 STEM college: top Americans don't do STEM. They do (according to colleagues from other departments) business, finance, law, medicine. For every American-born undergrad I have a foreign (largely Asian) one. For every American grad student I have 7-9 foreign ones. More alarmingly, their attitudes / commitment are vastly different, in general (there are always exceptions). Just like with blue collar work, the "next" gen is not willing to work super-hard in science, math, and engineering, by and large. THAT, and not some "lack of investment" or "lack of long-term vision" is what behind the Intel's latest failure.

Fascinating topic, and it gives me hope that the US ascends back to the top of the chip heap. But I’m particularly in admiration of Ms. Sanchez. To take on the challenge of fixing a chip development challenge— as a woman — in a hyper-male environment, is cooler than heck. She must really be comfortable in her own skin.

". . . in recent years [Intel] lost its longtime lead in manufacturing technology, which helps determine how fast chips can compute." And every year, Intel increases the number of foreign workers (H-1B visa holders) it employs in this country. Coincidence? “We used to have a lot of this kind of muscle that we let atrophy,” Ms. Rivera said. “Now we’re rebuilding.” Hire American tech graduates, support American STEM education, and only employ foreign workers who bring truly unique skills that are not otherwise available in our country (the definition of an H-1B requirement).

10 Recommend

Share

Flag

3 REPLIES

Suppan commented January 11

S

Suppan

San DiegoJan. 11

@NorthernVirginia Your point about H-1B visa holders might have relevance in other contexts, but the quote you are highlighting talks about "lead in manufacturing technology". This was lost due to offshoring the production (manufacturing) of chips, semiconductor equipment and so on. In plain terms, advanced manufacturing cannot be taught in college or university, it is learned by trial-and-error and repeated iteration in the real world. When you export your manufacturing facilities, you are exporting, removing from your country, the labs where the knowledge for advanced manufacturing is developed. These labs cannot be replicated even in the best University or Government Lab. It is fashionable to hate Elon Musk now, so I will not offer his views on engineering and manufacturing, but here's Carver Mead, a Professor at CalTech and one of the pioneers of the Semiconductor revolution. Wisdom in less than 9 minutes:

6 Recommend

Share

Flag

Suppan commented January 11

S

Suppan

San DiegoJan. 11

@Suppan This one is around 21min and in much more detail of some of the history.

1 Recommend

Share

Flag

Karen commented January 11

K

Karen

Bellevue, WAJan. 11

@NorthernVirginia There's no shortage of jobs for engineers. Essentially everyone with capabilities who wants to work is employed. The only people to hire are either the trickle of new graduates (with very raw skills), or taking someone from another job. Almost everyone in my 20+ years of software engineering seems to come by luring them from another company. We need all the h1b people that our idiotic conservative us political parties will allow to come here.

1 Recommend

Share

Flag

i would still prefer an intel chips over an amd chip, or tandy chip. had too many problems with them over my 40 years of engineering software.

Tape-in. Tape-out. I am not chip engineer. I would think the lingo doesn't matter to most readers. Mr. Clark confirmed what I heard from other sources, namely that Intel is struggling to overcome its struggles - and making progress That's news.

In the paragraph starting "But by December 2019, the engineers had hit a milestone called “tape-in.”" It should be "tape-out", not "tape-in": https://en.wikipedia.org/wiki/Tape-out

AMD is executing and Intel is not. Does AMD act like a teenager ... Yup.. but they are crushing it right now. Intel acts like a geriatric old man... and Pat isn't helping. We'll see

Why is a Chinese company like Lenovo allowed to purchase this?

It's "Tape-Out", not "Tape In". How could Clark have made such a blunder. Makes one question everything he writes. (BTW, I was a long time Intel engineer).

If I read a description of a microprocessor as an “electronic brain” once more, I think I’ll scream. This naive and simplistic article does nothing to illuminate the extraordinary human efforts required to bring a product to market embodying 10s of billions of transistors, tens of thousands of patented methods, cutting edge materials science, and brand new packaging technology. Disappointing lack of any insight whatsoever…NYTimes needs to dig deeper to adequately cover important technology business issues such as Intel’s ongoing return to leadership in the development of arguably the most important product in the world.

Intel's future rests on its management and engineers. Regardless of how much money Intel gets from the government, the future success is not certain. AMD already has developed the best logical chips using "chiplet" architecture and design. Every other company including many in China and S. Korea is working on it. Even if "Sapphire Rapids" succeeds, other will soon develop faster and better chip products. This chip dominance is a marathon rather than a 100-meter sprint.

And does this Intel chip use so much energy and produce so much heat that it's cooling fan needs to run at quicker speeds and for longer times than AMDCPU's?

I would not bet against Intel in the medium to long-term just yet. Pat is exactly the right CEO at at time when Intel needs him most. It will be a gruelling climb upward, but still very much possible. Building state of the art microprocessors is no easy task and AMD could yet stumble (as it has before in the past). And Intel has the additional challenge of upgrading its fabs - a double-whammy. I would not buy INTC anytime soon, but INTC could well be back in the game within 5 years.

The U.S. used to have engineers for everything it needed to conceptualize, then mock up, and then produce. My laymen's words here. A company I work with wanted to create a product, which involved moving metal and plastic parts, complicated parts, a compute to run it, and software to mesh everything together. Finding talent for all of this was hard, really hard. But not impossible. Then production engineers, to manufacture it in the U.S. Almost impossible to find. It seems China has thousands of production engineers. Thousands. Maybe tens of thousands. But the U.S.? A handful it seems. What gives?

The term is "tape out", not "tape in". Tape out is the final process of chip design where all of the various transistors are wired together. The term dates back to the time when circuit traces (wires) were defined by placing thin tape on clear plastic sheets, often using Xacto knives to cut it. These plastic sheets were then used to photographically reduce the size to make a mask for making a chip. For each layer, a metal layer is created on the chip then the mask is used to photographically etch the metal away like a photograph, leaving the wires behind. While tape hasn't been used in decades, having been replaced by CAD software, the name hasn't changed. The process usually starts with creating a floorplan, where various groups of transistors are placed in the chip area, then the wires are laid down. With high-speed chips like CPUs, this can grow extremely complex where the length of the wires must be within a certain length range so that the electrical signal gets from point A to point B in a specific amount of time. Not only that, but the width of the wires and the spacing between them can become critical as well at high speeds. After everything is laid out, it is simulated, then from the results of the simulation the layout is changed and it's simulated again. It's an iterative process. All it takes is one wire being too long to cause a major headache.

I recall one of the major themes in Tolstoy's War and Peace being the extent to which history assigns leaders the agency for any outcome, despite the course of history being largely driven by the combined actions of legions of individuals. This hero CEO will surely wave a magic wand and fix any difficulty.

I've never heard of "tape-in". Every company I've ever worked with calls releasing a chip for fabrication "tape out". Maybe Intel has their own term, they are Intel after all. The term comes from the olden days when the data describing a chip was stored on a magnetic tape. Most companies now don't use the term "validation" to describe (some of) the steps that are used to prepare a chip for fabrication. We call it "verification". "Validation" occurs after the fabricated chips are sautered to boards and tested.

Intel should adopt IBM's concept of millicode to update errant hardware with a never-ending series of "millicoded" errata. See "Millicode in an IBM zSeries processor" "Because of the complex architecture of the zSeries processors, an internal code, called millicode, is used to mplement many of the functions provided by these systems. While the hardware can execute many of the logically less complex and high-performance instructions, millicode is required to implement the more complex instructions, as well as to provide additional support functions related primarily to the central processor. ...."

The latest biggest microprocessor chips are a technological miracle. I am amazed they can design and build them as well as they do. Components on these chips are often less than 100 atoms wide. The billions of transistors make the hardware as complex as a large software system. Verifying the design is correct is an enormous task. You cannot just change a few lines of code to fix a problem. Any bugs that sneak through costs millions of dollars and months to fix. Verifying that each production device is working is also extremely difficult, when a flaw in a single transistor out of billions of incredibly tiny components could make the device unusable. The designs have to add some redundancy to the hardware to get a better yield of usable chips. Don't expect computer hardware to continue to increase exponentially in power, it is getting very hard to improve. Probably computer power will level off unless quantum computing ever becomes practical.

I used to say "never bet against intel in the long term" but I've updated it- "never bet against AMD in the long term". That's just how well AMD has done compared to Intel.

One can only wonder if Intel had taken the billions of dollars it used to manipulate financial markets with that ubiquitous tool of incompetent management known as the stock buy-back and instead invested in its design processes and fabrication technologies whether this chip would have been so delayed and whether the industry as a whole would have needed the federal life preserver/incompetent management bailout known as the CHIPS act. Perhaps the CHIPS act should have had claw-back provisions to take back the bonuses of the managers and executives who engineered this mess...