19:02 - "Hitachi -- maker of the Hitachi Magic Wand"

They will never be able to escape this image

Maker of the wand... they never living that down.

I was a chip engineer from 79 for a few decades, some of this story is just missing a lot of detail and gets other stuff wrong.

The HP story of 6x was to do with the alpha problem, Japanese DRAMs were mostly the same as US DRAMs except for one detail that was not appreciated in the US. The DRAMs were packaged in plastic to shield the die from the environment. What was not widely understood at that time was cosmic radiation from space and the trace amounts of uranium found in the plastic producing alpha decays. For some reason the Japanese sourced their plastic from a supplier that had vastly purer materials so they were getting 6x less alpha hits on the DRAM cells. The US suppliers were all using the same contaminated  supplies. At first the US suppliers denied the problem, but eventually understood it. The solution was to switch to the purer plastics that Japan used plus they also added another barrier, a polyimide layer right over the die. The problem had been less severe in the metal lidded part IIRC.

I worked at INMOS which for a while became a premier DRAM company that added new features such as the nibble mode and CAS before RAS as well as turbo charged clock driver circuitry that sped things up. That improved the system levels performance for bursting data into caches. By this time Intel was irrelevant to DRAMs by the low 80s, we never took them seriously as a player. Also INMOS had hired the best engineers from MOSTEK which was far more important to the story than this article suggests. The early 3t DRAMs were very quickly replaced by the 1t cell for many reasons, the 3t cell DRAM was really a complete piece of garbage and it was a different piece of garbage from every vendor. Take an 1103 from 5 companies that were supposed to be pin compatible, but were functionally different in timing parameters. A mainframe company could not mix and match chips from different vendors, the memory boards would not work. Also the DRAM was missing very basic features that needed to be inside the chip but were outside, the DRAM data line bit sensing, the word level clock drivers and so on, as well as many power supplies. These were 12V, 5V, -5V parts. The first of the 1t cell chips that went onto huge productions came from MOSTEK the 4027, they were quickly able to go to a 5V only part with 100s of innovations to get the DRAM into a simple to use design with tight reproducible specs. When these parts were mask copied by competitors US and Japanese, they all worked perfectly except for the plastic protection part exposed later on. Indeed there was a company called Mosaid that lived in Canada whos job was to reverse engineer the leading designs and sell reports to all comers. With a compatible MOS foundry and a Mosaid report you could get into DRAM design with a year of effort saved. Of course we had reports on all the leading designs, I used to enjoy reading them. MOSTEK eventually faded away but the legendary engineers from that company went on to form INMOS and MICRON. The later are still around, they pivoted right from the beginning though on good enough performance and the absolute smallest die for highest yield while INMOS went all in on complexity for the premium market. The shift from NMOS to CMOS was largely driven by Japan, and once DRAMs went to CMOS the US completely gave up on DRAM except Micron.

I could say more, but this article concentrates way to much on Intel as if they were the leader in everything DRAM, they were largely also rans even though they started the commercial DRAM. But before Intel, IBM had created the first CPU mainframe DRAMs but those were captive. Over the 80s when I went to the ISSCC conference, we often enjoyed papers from all the DRAM players teasing out a few details about the next gen DRAM, SRAM  and non volatile storage etc. INMOS had a hand in all 3 memory markets but the nitride based non volatile storage never really took off. It also ejected quite a few daughter companies like Fairchild had, but none of them really took off.

Biggest news of the day - we finally badgered Jon into saying D-RAM instead of Dram! 🎉🎉🎉

The debates about whether to exit the memory business began as early as 1982.  I attended the first exec staff and BOD meetings where the issue was debated.  The pain on the faces of Grove and Noyce, as they pondered killing off their startup vision, was palpable.  The debates went on seemingly endlessly until Grove and Moore finally saw the light.

lol the small laugh at 12:47 when he said "grope around a bit"

Being among the first comments is wild...
However: "the japanese businessman trested him like a god. As they should, he co-invented the IC"
With the changes that guy helped bring to the world... Yeah. Loved the quip. Thanks for the video!

“..they opened the kimono and shared hard data showing them as such.”   🤣  
Killed Me!!

He ain't lying about the magic wand, now everyone mixes their paint this way

A mistake in this video at 6:00.  The Altair 8800 used the Intel 8008 processor, it was the IMSAI 8080 that was the first to use the 8080.

As a Hillsboro resident, living about 1 mile from 3 different Intel sites including the main fab, is cool. It's a shame we're losing about 1500 staff from the current round of layoffs.

19:20 - I wasn't expecting at all to hear such adult joke.

I’m in the midst of re-reading Woz’s book, iWoz, it’s easy to not realize just how costly RAM was back then, we take it so much for granted now. He talks about how new it was for his Cream Soda computer to have 256 bytes of RAM and how he didn’t have to use core memory and the complex methods in which it has to be implemented.

This episode feels like a documentary and an amazing story. Love listening to your episodes, your eloquence and the flow is gripping!

Made in Germany was introduced to make sure that these cheap low quality products would not be mistaken for the British industrial products.
Made in Japan used to be a joke.
Korean? Used to be cheap knockoffs
Cars, electronics, steel etc etc
Now China is shaking this off.
In 10-20 years it will be Vietnam and India

Copy exactly was a bit of a nightmare for tool suppliers. Newer, better performing tools had to be adapted/tweaked/cherry-picked to match older tools in the fab...

Arrrggghhh!  I hoped I'd never hear that Kimono metaphor (9:23) ever again.  About 20 years ago, it was a trend where, if you discussed a controversial solution with a manager, they'd reply "I'm approaching this with an open kimono".

I worked with reliability testing in 1986 and Fujitsu memory circuits was best. Motorola 68000 processors we ran for 2 years in 160 C with zero failures so the test was stopped.

I worked at National Semiconductor in the 80s and lived through much of this. The Japanese simply kicked our backsides in both cost and quality. By the early 90s we were laying everyone off and they eventually closed the fab.

Intel struggling with yield and quality and losing market share? Well, well, well, how the turntables...