Architecture & Design – Lynne's Take on Tech

Fun Friday: Telescopes and Memories

I’ve been planning to write something for a while, but frankly, there hasn’t been anything really fun to write about.

Everyone is complaining about gas prices and inflation. Global trade is still bottlenecked and tangled in knots. There’s still a pandemic, folks, although you wouldn’t know it from the way people are dancing like it’s the last night before the End of the World.

On the business front, venture is busily grabbing any money they can to hoard while telling their portfolio companies to “tighten the belt”, mainly around the necks of their employees. Companies are eagerly complying by rescinding job offers and instituting layoffs. Folks are nervous as they crowd airports, hoping their flight isn’t one of the hundreds cancelled that day due to lack of flight staff. And the war in Ukraine waged by Russia in a fit of insanity continues to kill innocents and destabilize the entire EU.

Speaking of dead innocents, the US Supreme Court, destined to go down in history as depraved pandering sacks of shit, decided that guns everywhere makes for a stronger America. Their overturning of Roe v Wade, expected after the leaked draft admiring the people who burned innocent people as witches crawled out of the sewers, has been released and to no one’s surprise reduced women to that of beasts. Yes, it is not a Fun Friday for many people. Maybe it’s a Gun Friday. I’m sorry.

Roe v Wade was decided in 1973. I was twelve. It impacted my life and health for the better. Today it is officially overturned in a ruthless precedent-be-damned legal coup. I am sixty, past childbearing age. It cannot impact me directly. Yet I have daughters and young people I care about. I don’t want to see them hurt. Their happiness and livelihood and health matters to me. They should have the same rights to choice and freedom that I had. They may not know how much it matters yet. But they will. I am sure of that.

I spent the morning cleaning one of William’s prototype telescope designs for display in the office. It’s an unusually compact and minimalist design. As I cleaned the mirror and cover plate, I found a cricket living in the focuser. I watched it hop off the picnic table and out of sight, grabbed the telescope, and took it to the office.

It now sits amongst the many creative works William and I did together. Our reliquary. 375 computers. InterProphet low-latency networking boards. 386BSD articles and books and CDROM. An unpopulated six layer 375 motherboard.

In other parts of the office, an EtherSAN prototype unit box, a 386BSD CDROM with the heftiest liner notes ever made, 386 computers of various vintages used for 386BSD, and bins 386BSD and 375 disk drives, boards, and cables. Some complete and some mid-project, designs waiting for a hand to finish the work.

It is a reminder that things are never finished — they are only left in a state of usability for a time. Once that time passes, one either has to toss it away or begin again. I choose both. To toss some things away and to begin again on other things.

Young people also have a choice. They can fight for their freedoms — and they can toss them away. I hope they choose wisely.

Fun Friday: Funding in Transition and Mammalian Distributed Memory Storage

As inflation continues to take its toll on everyone’s investments as well as steak dinner (psst – get the rotisserie chicken at Costco instead), Silicon Valley is clearly in a state of transition. Startups have been told to tighten their belts financially. Layoffs in big tech companies have begun. “Growth” ventures are failing to get follow-on funding, primarily in the consumer space and media (in Substack’s case, they’re also proving the old adage that no one pays attention to writers).

But for every easy money gambit that’s falling out of the sky, there is hope for the dreamer and rogue. Venture firms are still collecting money hand-over-fist from desperate Limiteds eager to get some return with the stock market slowing. Folks with money want to make more money. There are lots of them.

Of course, this doesn’t directly help the small entrepreneur. Big Venture (TM) doesn’t fund the small fry inventing neat technologies anymore — they have too many Series D unicorn mouths to feed. Big tech companies are no longer a safe bet — they may fire you and escort you unceremoniously out the door without any warning. In hard times, loyalty is not its own reward.

But there are a lot of individual investors out there who can drop $1M on a neat tech idea. All the startups William and I founded started with a dream, some code, and a handshake during lousy economic times. They were funded precisely because making easy money on scams and gambits have evaporated.

So if you’ve got a good hard tech project, now may be the best time to go for it. The cash is still plentiful. Just play it cool. It worked for us. It can work for you.

Researchers tracked neural activity across a whole mouse brain to determine what areas were involved in storing a specific memory. Many brain regions found likely to be involved in encoding a memory (top) were also found to be involved in recall upon reactivation (bottom).
Credits: Image courtesy of the Tonegawa Lab/Picower Institute. — Neural activity across a whole mouse brain to determine what areas were involved in storing a specific memory. Many brain regions found likely to be involved in encoding a memory (top) were also found to be involved in recall upon reactivation (bottom).
Credits:Image courtesy of the Tonegawa Lab/Picower Institute. Read the article!

And speaking of distributed memory, a new study from MIT describes the mammalian brain as storing memory, not densely in a few regions, but instead loosely across many regions of the brain. This makes sense in a way. It’s a lot easier to completely lose a memory if it’s in one or a few locations than spread throughout the brain. Also, storing memories in larger “chunks” would result in a lot of wasted storage space since a memory is of varying size. Indirect references to each memory element, even if a few are lost, are more efficient than directly physically mapping a memory.

It does explain the dreamlike aspect of memories, doesn’t it? And also perhaps memories which are completely wrong but feel entirely real and true. Likely we lose a lot of these references that fill in some of the blanks over time. Associated elements, like smell, can track back along a pathway to a memory to give the gist of it, but it may be only a shadow of what was actually recorded.

But is the brain’s memory sparsely allocated as well? It may well be given this highly distributed storage across many parts of the brain. Sparse allocation is common in operating systems because it is usually faster and overall more efficient. But it can use more total memory than that of a densely allocated memory mechanism if most of the elements contain non-repeatable data. Are most of our memories just collages of a few meaningful pieces and a lot of filler? Perhaps dreams look odd precisely because they are just stray strands of sparse referents to redundant memories garbage collected by the brain and reallocated for use.

To dream. Perchance to sleep. Now that is the question.

Fun Friday: AMD-Intel Battle Commences, Big Honking Space Guns, and our Good Friend, the Comb Jelly

The Changing Face of Intel (Economist, 2021)

Another slow Friday. Texas has frozen over. California is becoming ever-drier. The Polar Vortex has crept downwards over the decade to now reach Mexico. Summer will be burning. And the pandemic continues.

Everywhere climate change has come for us all. We are not ready for it. Not. At. All.

Pat Gelsinger has taken the reins at Intel, as the Economist notes (paid subscription). I do enjoy a piece that remembers Andy Grove, the iconoclastic Intel CEO who passed away in 2016 and made Intel a leader in microprocessors.

I remember hearing rollicking stories of the semiconductor industry from Dick Williams, then retired Director of Research from Ford Aerospace / Loral, when we shared an office in the late 1990’s at InterProphet (he was on the Advisory Board).

Back in those days, he would recount, creating semiconductor processes was so difficult that people would talk about it from different firms over drinks after work, so everyone knew everyone. He also hired Bob Noyce into his first job.

Andy Grove and Bob Noyce were among the Traitorous Eight who founded Fairchild Semiconductor. When that got stuffy, Noyce left with Gordon Moore (of the famed “Moore’s Law”) to found Intel, and of course immediately pulled in Grove and dived into DRAM — and did well, until the DRAM Wars drove them out. That’s when the X86 series was born.

By the time the 80386 was established in the late 1980’s, we were outlining a new open source Unix future with 386BSD and A Modest Proposal in 1989. When we were updating 386BSD in 1990, we were given access to their prototype 80486 processor at their headquarters here in Silicon Valley — just so 386BSD would run correctly.

I actually have a fondness for the early Intel. They were supportive of 386BSD and courteous, unlike AMD’s bland officiousness. Later on, when Intel abandoned 386BSD and went exclusively to Linux for what they thought were legal concerns, I was disappointed but not surprised. There was a lot of “fake news” back then about 386BSD and us, and there was no Snopes to counter it.

Only years later was I told by one of my retired Intel executive contacts that they made a mistake in that decision, as none of the catastrophic claims ever came to pass. But it was water under the bridge at that point, and I had gone on to found InterProphet, a TCP low-latency fabless semiconductor company with William. It was just another of the many opportunities in Silicon Valley killed by malign neglect.

Now Pat Gelsinger, architect and technologist, is back at Intel, and he has quite a mess to clean up. Bad business decisions, delays in chip production, and an unimaginative product roadmap might still have not hampered Intel’s profitability. But unlike the Sanders era of AMD, the current CEO, Dr. Lisa Su, is a brilliant technologist and business leader who has led AMD to leadership in the industry.

So now a battle of equals commences. AMD and Intel. It should be interesting, to say the least.

But while we’re waiting, there’s always Big Honking Space Guns from Russia and the ever-amazing comb jelly to bemuse and bewitch you, as they have me.

Have a Fun Friday, everyone!

Fun Friday: AI Technology Investments, Failed Startups, 386BSD and the Open Source Lifestyle and Other Oddities of 2020

First, William Jolitz and I did a comprehensive article entitled Moving Forward in 2020: Technology Investment in ML, AI, and Big Data for Cutter Business Journal (April 2020 – paid subscription). Given the pandemic and upheaval in global economies, this advice is even more pertinent today.

Instead of moving from technology to key customers with an abstracted total addressable market (TAM), we must instead quantify artificial intelligence (AI) and machine learning (ML) benefits where they specifically fit within business strategies across segment industries. By using axiomatic impacts, the fuzziness of how to incorporate AI, ML, and big data into an industry can be used as a check on traditional investment assumptions.

For additional information on this article, please see AI, ML, and Big Data: Functional Groups That Catch the Investor’s Eye (6 May 2020, Cutter Business Technology Advisor).

Techcrunch presented their loser brigade list of 2020 failed startups in December of 2020 – although a few more might have missed the list by days. Some of these investments were victims of “the right startup in the wrong time”. Others were “the wrong startup in the right time”. And some startups were just plain “the wrong startup – period”.

We mourn the $2.45 billion dollars which vanished into the eager pockets of dreamers and fools (we’re looking at you, Quibi – the pig that swallowed $1.75B of investment and couldn’t get any customers) and feel deeply for the Limiteds who lost money in one of the biggest uptick years in the stock market.

Thirty years have passed since we launched open source operating systems with 386BSD. Open source as a concept has been around for over 40 years, as demonstrated by the amazing GNU GCC compiler done by RMS. But until the mid-1990’s, most software was still held under proprietary license – especially the operating system itself. The release of 386BSD spurred the creation of other progeny open source OS systems and a plethora of open source tools, applications and languages that are standard today. However, the “business” of open source is still much misunderstood, as Wired notes in “The Few, the Tired, the Open Source Coders”. Some of the more precious gems excerpted:

But open source success, Thornton quickly found, has a dark side. He felt inundated. Countless people wrote him and Otto every week with bug reports, demands for new features, questions, praise. Thornton would finish his day job and then spend four or five hours every night frantically working on Bootstrap—managing queries, writing new code. “I couldn’t grab dinner with someone after work,” he says, because he felt like he’d be letting users down: I shouldn’t be out enjoying myself. I should be working on Bootstrap!

“The feeling that I had was guilt,” he says. He kept at it, and nine years later he and Otto are still heading up Bootstrap, along with a small group of core contributors. But the stress has been bad enough that he often thought of bailing.”…

…Why didn’t the barn-raising model pan out? As Eghbal notes, it’s partly that the random folks who pitch in make only very small contributions, like fixing a bug. Making and remaking code requires a lot of high-level synthesis—which, as it turns out, is hard to break into little pieces. It lives best in the heads of a small number of people.

Yet those poor top-level coders still need to respond to the smaller contributions (to say nothing of requests for help or reams of abuse). Their burdens, Eghbal realized, felt like those of YouTubers or Instagram influencers who feel overwhelmed by their ardent fan bases—but without the huge, ad-based remuneration.

Been there. Done that.

Not many Linux-come-latelies know this, but Linux was actually the second open-source Unix-based operating system for personal computers to be distributed over the Internet. The first was 386BSD, which was put together by an extraordinary couple named Bill and Lynne Jolitz. In a 1993 interview with Meta magazine, Linus Torvalds himself name-checked their O.S. “If 386BSD had been available when I started on Linux,” he said, “Linux would probably never have happened.”

Linus was able to benefit from our two year article series in Dr. Dobbs Journal (the premiere coding magazine of the day, now defunct in an age of github), which along with the how-to details of “Porting Unix to the 386” we also included source code in each article. That, coupled with Lions Commentary on Unix (NB – the old encumbered Edition 6 version, and not Berkeley Unix) allowed Linus to cudgel together Linux. We had no such issues, as we had access to both Berkeley Unix and a source code license from AT&T for our prior company, Symmetric Computer Systems, and hence knew what was encumbered and what was not (Lions was entirely proprietary). Putting together an OS is a group effort to the max. Making an open source OS requires fortitude and knowledge above and beyond that.

Jalopnik, one of my favorite sites, found the ultimate absurd Figure 1 patents with this little gem of an article: Toyota’s Robocars Will Wash Themselves Because We Can’t Be Trusted. Wow, they really knocked themselves out doing their Figure 1, didn’t they? Womp Womp.

And finally, for a serious and detailed discussion of how the pandemic impacted the medical diagnostic side, I recommend this from UCSF: We Thought it was just a Respiratory Virus. We were Wrong (Summer 2020). Looking back, it was just the beginning of wisdom.

Stay safe, everyone!

2020 AMD and Intel: The Grass is Greener on the Other Side of the Chip Business

AMD or Intel? 2020 is the Processor Battleground

AMD, the long neglected stepsister to Intel, has done marvelously well in recent years, primarily due to Intel’s “meltdown” of trust in their flagship processor products and Intel’s delays in shipping new competitive 10nm chips. Coupled with ineffectual senior management and poor board control, Intel, the darling of the Wall Street set, sat wallowing in management paralysis and a moribund stock price until recently.

Meanwhile AMD’s CEO Dr. Lisa Su has been instrumental in moving AMD from its Eeyore approach to life to that of a first-rate competitor in the chip space with its Ryzen, Radeon and Epyc product lines. Dr. Su has not only changed AMD’s attitude – she’s also changed the entire competitive landscape with bold technology moves and strategic partnerships with companies such as Microsoft. Having dealt with the earlier AMD in the 1990’s, where no one would make a decision and the C-suite was filled with ineffectual do-nothings, it has been refreshing to see capable management drive good engineering and product management.

In the last few years, AMD and Intel have swapped places. AMD, the driver in specialist processors, has gone full-bore into the vacuum left by Intel’s strategic blunders and broadened into general processors . Intel, in contrast, has made an old Intel revenue-enhancement approach “new again”, by taking their general processors and specializing them for specific markets.

But believing the grass is greener on the other side of the chip business comes with its consequent perils. Continue reading 2020 AMD and Intel: The Grass is Greener on the Other Side of the Chip Business

Intel’s X86 Decades-Old Referential Integrity Processor Flaw Fix will be “like kicking a dead whale down the beach”

Brian, Brian, Brian. Really, do you have to lie to cover your ass? Variations on this “exploit” have been known since Intel derived the X86 architecture from Honeywell and didn’t bother to do the elaborate MMU fix that Multics used to elide it.

We are talking decades, sir. Decades. And it was covered by Intel patents as a feature. We all knew about it. Intel was proud of it.

Image: Jolitz, Porting Unix to the 386, Dr. Dobbs Journal January 1991

Heck, we even saw this flaw manifest in 386BSD testing, so we wrote our own virtual-to-physical memory mapping mechanism in software and wrote about it in Dr. Dobbs Journal in 1991.

You could have dealt with this a long time ago. But it was a hard problem, and you probably thought “Why bother? Nobody’s gonna care about referential integrity“. And it didn’t matter – until now.

Continue reading Intel’s X86 Decades-Old Referential Integrity Processor Flaw Fix will be “like kicking a dead whale down the beach”

MetaRAM Busts RAMBUS Stranglehold?

Sand Hill Road envies RAMBUS. Oh, they don’t envy them their lawsuits, precarious business model or turbulent management structure. But they do envy them their ruthless monopoly of the high-speed DRAM market. RAMBUS successfully competed against the behemoths with a clever architectural enhancement, kept belief in their approach against huge odds, fought back as hard and dirty as the big boys and made licensing deals stick. They are survivors.

When RAMBUS went IPO back in 1997, I was completing work on the first preliminary patent application for InterProphet’s SiliconTCP technology, while William began his hunt for investment. RAMBUS’s IPO was on the minds of many VCs, but it wasn’t in a good way, surprisingly. RAMBUS’s 7 prior years had been fraught with changes in business model and personnel. Instead of setting up a fab, RAMBUS chose to license their technology. Finally, RAMBUS chose to make their stand on the basis of their patents. Don’t let me fool you — investors may crab about the need for “intellectual property protection” but when it comes to playing with the big boys, they believe about as much in IPR as the tooth fairy.

RAMBUS has been remarkably successful in defending and enhancing their patents (and yes, I know about their “steering committee” games — coming from the OS side I’ve seen Microsoft and others play the same games, even to the point of doing software patents on work pre-existing by decades). Essentially, they’ve played dirty like Intel, Hynix and all the other guys the VCs said you could never win against. But it has been a very long wild and crazy ride for the payoff — too much for the “10x in 3 years” crowd.

But despite all of RAMBUS’s remarkable turbulence, it has been amazingly successful. During one incredible record-setting day in 1998, I listened to a top-tier VC say that he’d never want a single share of RAMBUS’s stock no matter *how* much money they made. He just hated them. Another top-tier VC rambled on about how “you could make a lot of money with a RAMBUS business model, but they weren’t interested in that”. What they really hated is how there wasn’t a single massive success where they could bow and take their winnings (like VMWARE in 2006 for example, but they didn’t invest in that one either because it was run by a husband-wife team that believed open source was valuable — hmm, beginning to see a pattern). As Magdelena Yesil (at the time partner at USVP) liked to intone to me “Venture Capitalists are more capitalist than venture these days”. Chip risk wasn’t as exciting when you could respin any company as an Internet venture and go public with no revenue. And semiconductor companies *are* risky.

Semiconductor companies are also the historical lifeblood of Silicon Valley — hey, that’s why it’s called “Silicon Valley” and not “Internet Valley”! So now we come to MetaRAM, an attempt to steal RAMBUS’s monopoly on architecture. According to Ryan Block of Engadget “MetaRam uses a specialized “MetaSDRAM” chipset that effectively bonds and addresses four cheap 1Gb DRAM chips as one, tricking any machine’s memory controller into using it as a 4x capacity DIMM.”

Is the technology innovative? Not likely — it sounds like a combination cache and bank decoder, which is not innovative in the least. In fact, you need 4x the number of components on the DIMM, which means 4x the number of current spikes and decoupling capacitors, even if you put the chips together in the same package. Because you have a fifth chip, you complicate things even more. There is no way you can approach the triple-zero (volume, power, cost) sacred to chip designers with such a design, because one single high-speed high-capacity chip will eventually win out given the proliferation of small expensive gadgets demanding the lowest of volume and power. In a world of gadgets like IPODs, cellphones, laptops, PDAs and the like, cost is very important but *not* the most important quantity. So RAMBUS doesn’t have a lot to worry about here.

Hynix has been fighting a losing battle against RAMBUS ever since getting hit with a whopping $306M patent infringement judgment in 2006 (since reduced to $133M), and RAMBUS is still going for more. These are the same guys who pleaded guilty in 2005 to a DOJ memory price-fixing scheme from 1999-2002 and paid a $185M fine. There is no love lost in the memory biz.

So where does little MetaRAM come in. When technology fails, maybe a clever business model will do. MetaRAM’s big claim to fame is cost reduction — not for gadgets or laptops, but according to Fred Weber, CEO of MetaRAM, for “personal supercomputers” and “large databases”. And who is the big licensee for this so-called technology. Why, it’s Hynix of course, who announced they will make this lumbering memory module. They claim it will be lower power. I think I’d like an independent evaluation on this point, but it will probably be lower cost. Is it worth it? Given reliability considerations, that also remains to be seen. But the moral of this saga is simple — human memories are longer than memory architectures in this business, and the real puppet-master behind the throne (Kleiner-Perkins) is sure to walk away with the money. I wish I could say the same for the customers.

When Video Kills Your Drive – Quicktime Waxes Track 0

Alright! Yes, sometimes I do read slashdot when it’s amusing, and the discussion of how you can create your own custom panic screen (or BSOD window) for OS/X via an API is amusing (my son Ben points this stuff out for me – he feels it’s one of his sacred tasks). Joke panic screens have been around a long time, but the battle over “how much information to give people” has led to many not-so-amusing battles, especially when we were creating 386BSD releases and did the Apple approach as an option long before Apple switched to BSD and did the same thing we did. I like information and transparency, but not at the cost of frustrating and annoying lots of people…

But aside from this old debate, probably the funniest *real* programming error discussed (yes Apple, you did it again) was the Quicktime capture bug a video engineer discussed, where it merrily fills up the drive with video and then, when you’ve run out of space on the disk, overwrites allocated blocks. Yes, allocated blocks! And where did that leave our poor engineer? With no track 0. Even Apple couldn’t put that one back together again – they had to give him a replacement drive, and that’s a complete admission of defeat, because as everyone knows Apple is very cheap. (Disclaimer – I have worked in manufacturing, and we all are very cheap in this regard because hardware returns make a nasty entry on your income statement – show me a systems vendor that doesn’t care about hardware returns, and I’ll show you a vendor that won’t please shareholders).

But the question begs – is there another way to recover from a track 0 loss on a disk drive? Well, we faced the same problem at Symmetric Computer Systems years ago, and we recovered that disk, albeit not the way you might think…

Itanic Readies for Final Sinking – Multiflow and HP Tech Aren’t Enough

Alas, “itanic”, aka “Itanium” is resurrected again, this time debuted by Intel as a “chip with two brains”. But the critics aren’t impressed, and to save money for senior management Itanium soirees, Intel middle managers are to be tossed into the cold big blue.

Well, this is no surprise to those chip-watchers who talked down Itanium from the beginning. While Dell and IBM have fled to calmer X86 waters with the likes of AMD, HP has been steadfast, selling 80% of the chips Intel has shipped.

Of course, many wonder why HP is such a firm holdout when everyone else is baling? To understand how we got where we are, we actually have to ask the question “Where did the Itanium come from?” If you think it all sprang fully formed from Andy Grove’s head, you’re quite wrong. And you’d miss the story of long-ago acquisitions, agreements, and ambitions…

Microsoft’s Ultimate Throughput – Change the Compiler, Not the Processor

I like people who go out on a limb to push for some much needed change in the computer biz. Not that I always like the idea itself – but moxie is so rare nowadays that I have to love the messenger despite the message. So here comes Herb Sutter, Microsoft Architect, pushing the need for real concurrency in software. Sequential is dead, and it’s time for parallelism. Actually, it’s long overdue in the software world.

In the hardware world, we’ve been rethinking Von Neumann architecture for many years – SiliconTCP from InterProphet, a company I co-founded, uses a non-Von Neumann dataflow architecture (state machines and functional units – not instruction code translated to Verilog because that never works) to bypass the old-styled protocol stack in software, because an instruction based general processor can never be as efficient for streaming protocols like TCP/IP as our method. Don’t believe me? Check out Figures 2a-b for a graphic on how much you wait for store and forward instead of doing continuous flow processing – the loss for one packet isn’t bad, but do a million and it adds up fast.

It’s all about throughput now – and throughput means dataflow in hardware. But what about user-level software applications? How can we get them the performance they need when the processor is reaching speed-of-light limits? If on a typical processor from one end to the other end you get one clock cycle at the speed of light at 7-8 GHz, anyone stuck in sequential processing will be outraced by Moore’s Law, multiple cores and specialized architectures like SiliconTCP.