Pages

Sunday, December 3, 2023

Supercomputers And All That Jazz -- December 3, 2023

Disclaimer: in a long note like this there will be content and typographical errors. Facts and opinions will be interspersed and it will be hard to tell the difference. It has not been proofread (yet) and more editing must be done, including some more formatting, but I have family commitments.  

Disclaimer: with regard to all investing discussions -- I never make recommendations and my personal investing horizon has a 30-year horizon.

There have been a number of "items" / subjects that have interested me over the years and I have explored them in depth to get a better understanding. Examples:

  • the Bakken, of course; 
  • Shakespeare;
  • Richard III;
  • renewable energy (wind, solar);
  • Netflix;
  • Apple, Inc.;
  • Devon;
  • BRK;
  • semiconductors, chips, blades;

Most recently, supercomputers. A recent update, November 1, 2023. I don't know for sure, but I think I will be tracking supercomputing here.

Supercomputers are ranked in terms of computing power measured in petaFLOPS or petaflops, defined as a unit of computing speed equal to one thousand million million (10^15) floating-point operations per second (which of course I don't understand at all) -- whatever.

So, anyway, where are we with regard to petaflops?

It's hard to believe, but the subject was actually noted as far back as December 7, 2012, almost eleven years to the day. From that link:

Link to Rigzone.com.

Years ago I followed the dramatic increase in commercial computing power, tracking the "race" between the US and Japan. I pretty much forgot all about that race until this story.
BP has begun building a new supercomputing complex for commercial research that it claims will be the biggest in the world at its Westlake Campus in Houston, the company reported Friday.
The project is designed to keep BP at the forefront of seismic imaging technology and, the firm said, will be a critical tool in its global hunt for oil and natural gas in coming years.
BP's existing HPC center was the world's first commercial research center to achieve a petaFLOP of processing speed (or one thousand trillion calculations per second). But the company said that this has now reached maximum power and cooling capacity in its current space at the Westlake Campus.

So, the update. Whatever happened with regard to BP's "new supercomputing facility in Houston"? From Rigzone, October 23, 2013, about a year later from the original announcement: 


From the linked article:

BP announced late Tuesday that it has opened its new High-Performance Computing (HPC) Center at its US headquarters in Houston.

The new center, which BP began building in December 2012, will serve as a worldwide hub for processing and managing very large amounts of geological and seismic data from across BP's portfolio. It will enable scientists to produce clear images of rock structures that are deep underground.

BP said that better imaging capability will also help the company more safely and efficiently appraise new finds and manage complex reservoirs once production starts. In addition, the company expects the center will open up new possibilities for research into other important aspects of BP’s business activities, from oil refining to enhanced oil recovery.

The new center is housed in a three-story, 110,000-square foot facility at BP's Westlake campus. Equipped with tens of thousands of CPUs, it is designed to have the ability to process data at a rate more than 2.2 petaFLOPS – meaning it can perform 2.2 million, billion simple calculations (or floating-point operations) per second. This almost doubles the firm's previous supercomputing facility.

So, we have one data point: the BP Westlake Houston supercomputer is designed to process data at a rate of more than 2.2 petaFLOPS.

Impressive, huh. Ten years ago, 2.2 petaFLOPS.

Where do we stand now?  

Before we answer that question, we have to introduce a new term: exaflops or exaFLOPS.

Reminder:

  • petaflops: 10^15

Now:

  • exaflops: 10^18; that's three more zeroes;
  • an exaflop is 1,000 times greater than a petaflop

So, if a petaflop is 1,000 trillion, or one thousand thousand billion (1,000,000,000,000,000), an exaflp is 1 billion billion, or one one thousand thousand thousand billion (1,000,000,000,000,000,000) 

How fast are the fast supercomputers today? Measured in:

  • petaflops: 1,000 petaflops
  • exaflops: 1 exaflp

So, rounding, the Houston supercomputer, back in 2013 was able to compute at a speed of 2 petaflop.

Today, supercomputers operate at a speed of 1,000x that much. 

What was Moore's Law? Rhetorical.


 Gazillion: some number greater than 10^24 but less than infinity.

*****************************
Break, Break: Where The Rubber Meets The Road

As an investor, I really don't care about all of that above the fold.

What interests me, as an investor, who's making the blades for those supercomputers?

Who made the blades for that BP Westlake Houston supercomputer? Link to oedigital:

By the end of the year, BP’s computing center will hold 6000 computers. BP worked with HP and Intel to grow its computing power to over 2.2 petaflops, nearly doubling the company’s capability in one year.

Besides that little pearl, much more at the oedigital link.

But back to the question: who made the blades for the BP computer? Intel. Wired by IBM.

*************************
Investing

It was this post on November 22, 2020, that got my attention. That was when I made a huge pivot from energy to Nvidia. Up until then I don't recall ever hearing the word "Nvidia" on CNBC but that at the time, I wasn't watching much CNBC and I knew nothing about blades.

But two things in 2020:

  • that wiki entry at that link
  • tech was selling off due to the Covid lockdown
    • Nvidia, Zoom, Cisco, etc, were doing well but felt to be temporary.

For Nvidia, things looked worse, immediately coming out of the lockdown. Do folks remember why? That's "Nvidia 101."  Look at 2022. I remember that as if it happened yesterday.

Look at a five-year and a "max" chart for Nvidia. 

**********************
Where Do We Stand Today? December 3, 2023

First of all, and I pat myself on the back for this one -- all the blogs I have posted regarding semiconductors since November 22, 2020. That, for me was as eye-opening as my original posts on Netflix and Devon. 

But then, this post on December 2, 2023. Look at the screenshot from Forbes. The "X86" is Intel's.

That Forbes article was dated August 8, 2023, and I think Forbes was already late in publishing this story. The movers and shakers would have known this in 2020 (maybe, perhaps not -- there's more to the story) but certainly by late 2022 / early 2023.

************************
Is There More? The Cherry On Top!
They're Reading The Blog

In the on-line edition The WSJ places their stories by "position." Position 1 is the top of the page, the most important story on that page, above the fold. Position 2 is the second most important story.

Note this URL: https://www.wsj.com/tech/personal-tech/intel-inside-apple-arm-microsoft-nvidia-a435183b?mod=wsjhp_columnists_pos2. 

A WSJ article in position 2. An important story. Note that it was published after the market was closed for the week.

Here's the headline:


What day was December 1st? Oh, that's right -- it was Friday. Note the time the story was posted, 9:00 p.m. ET, too late for traders on Friday. They will read it on. Saturday in the weekend edition (perhaps). Then Monday. Tomorrow.

It's behind a paywall, but this is how the article begins:

It might not look like it yet, but Intel is in a fight for its life.

The stakes for its employees and investors are high, and are likely to turn on some fierce battles for market share that will play out in 2024 and beyond.

For the everyday consumer, what’s at stake is mostly nostalgia. One day, the little “Intel Inside” sticker that’s been on PCs since 1991 could cease to exist.

Instead of an Intel chip, these computers could have processors from an array of manufacturers, principally Qualcomm , but also possibly Nvidia , AMD, and lesser-known companies like Santa Clara, Calif.-based Amlogic and Taiwan-based MediaTek .

What’s happening now is a tipping point decades in the making. Ever since a little chip-design company called ARM built the mobile processor for Apple ’s first Newton personal digital assistant, which came out in 1993, it’s been gaining steam, primarily in the mobile-phone business. By the time Intel sought to enter the mobile-processor business in 2011, it was too late.

Apple was the first company to bet that ARM-based processors—thought by many to be useful only in phones—could be the brains of even the most powerful desktop computers. This gave Apple a huge head start over Intel, and the rest of the industry, in designing chips that prioritized power-sipping performance in a world where that’s become the primary limiting factor in the performance of all devices, not just phones.

Now, Google, Qualcomm, Amazon , Apple and others can use ARM’s blueprints to custom-design the chips that power everything from phones and notebooks to cloud servers. These chips are then typically produced by Samsung or Taiwan-based TSMC , which focus on making chips for other companies.

The threats to Intel are so numerous that it’s worth summing them up: The Mac and Google’s Chromebooks are already eating the market share of Windows-based, Intel-powered devices. As for Windows-based devices, all signs point to their increasingly being based on non-Intel processors. Finally, Windows is likely to run on the cloud in the future, where it will also run on non-Intel chips.

The article goes on and on and on. It's a long, long, long article.

There is a fly in the ointment and it has nothing to do with Intel or Nvidia but that's another story for another day.

Traders are going to love this story. Traders: get in, make your money, then get out. For investors, it's much, much, much more difficult. So much can change. Maybe it's time to sell / short NVDA. Maybe it's time to start a position.

Example: Apple wasted a lot of time and resources on an autonomous vehicle when in hindsight it appears the company could / should have diverted those efforts to supercomputers. Governments around the world have deep pockets and are racing to build faster supercomputers. China alone will keep the industry busy for years. If not decades. 

We're starting to see how EVs and autonomous vehicles are working out for folks. 

Disclaimer: this is not an investment site. Do not make any investment, financial, job, career, travel, or relationship decisions based on what you read here or think you may have read here. 

All my posts are done quickly: there will be content and typographical errors. If anything on any of my posts is important to you, go to the source. If/when I find typographical / content errors, I will correct them. 

Again, all my posts are done quickly. There will be typographical and content errors in all my posts. If any of my posts are important to you, go to the source.

*****************************
Of Course There's More But I Have To Take A Break

I've got a great Apple post -- but that will be later this afternoon / evening.

*************************
Parting Shot

I've posted these graphs before but these graphs are important for many, many reasons.

Nvidia has been around for a long, long time. Everyone had the same information from which to work.


No comments:

Post a Comment

Note: Only a member of this blog may post a comment.