It tells the amazing story of how Robert Noyce, Arthur Rock, Gordon Moore, and Andrew Grove were absolute pioneers and insanely gifted. A huge loss.
Here is the first 15 minutes of the documentary on YouTube: https://www.youtube.com/watch?v=LcOoQP7nhl4
They didn't have it easy like Microsoft did where they were virtually the only game in town. They almost lost their position to AMD on more than one occasion, and were under constant pressure from Motorola, IBM, Sun, MIPS, DEC, and others all intent on capturing a dominant position in the highly lucrative desktop and server markets.
Where hard drive makers would have a few years of glory before burning out or getting folded into another company, and where few CPU manufacturers had the talent to survive more than a decade, Intel not only popularized the CPU, but kept a near strangle-hold on the market for over two decades. They're still number one by a number of key metrics even with the increasing pressure from ARM.
Few companies from the Intel era in the 1970s have survived intact. The only one I can think of is Seagate out of hundreds from that era.
I doubt few will ever measure up to Grove in terms of impact.
Not to take anything away from Grove, but a lot of the focus on deep instruction pipelines can probably be traced to certain software partners being concerned about scaling across multi-core.
As for VLIW/Itanium, I expect some of the same motivations were in play--plus, a desire to reset and lock-out the various x86 clone makers.
Remember that in Grove's time Intel was getting clobbered in benchmarks by the DEC Alpha chip and was struggling to keep up with the PowerPC consortium. It was through their relentless pursuit of clock-speed at any cost that they prevailed.
Unfortunately they slammed into the 4GHz barrier earlier and harder than the competition and went into a spiral of confusion for nearly half a decade. Adding more and more stages to their pipelines was the primary strategy for pushing higher and higher clockspeeds.
What saved them in the end was the very pragmatic and practical Pentium III platform, the one introduced a few years before Grove retired, not the outlandish and massively power hungry Pentium 4 architecture.
Grove did seem to keep the company focused on relentless incremental improvement. This is in stark contrast to the later Intel which showed unbelievable hubris with their VLIW effort thinking that people would spontaneously adopt a whole new architecture just because. Maybe they listened to their partner HP way too much as that company is quite clearly utterly clueless.
You mean the Pentium Pro platform? Pentium III is just Pentium Pro with off-package L2 cache and SSE instructions. The P6 microarchitecture kept Intel in business from 1998 (when the P5 line concluded) until the Nehalem architecture launched in 2008.
http://www.intel.com/content/www/us/en/processors/xeon/xeon-...
I wouldn't call it dead yet. It's still paying dividends.
You're way overstating that. You're whole post, but especially that one, totally ignores the concept of lock-in due to ISA, OS, and library backward compatibility. And first mover advantage, patents, and the Microsoft/Intel partnership. It's amazing how successful a product can be when attempting to clone or ditch it can sink an entire business or its profit margins. And especially if it's cheaper than UNIX/RISC or mainframes on top of that. ;)
Note: I'm an Intel and Windows opponent who can't do shit for specific customers of theirs with legacy apps due to the above. They want benefits of other things but can't switch. I can say same about certain mainframe shops. I'll be able to say same about Oracle/Java and certain cloud shops 10 years from now. I bet you $100.
AMD managed to navigate a patent minefield and ended up with a 32-bit compatible chip that was basically a drop-in replacement, the user would barely notice the difference unless they went looking for one. The Athlon XP became a serious competitor to Intel's then flagship Pentium series. AMD was also the first to come up with a workable 64-bit IA instruction set that Intel was later forced to adopt because of widespread popularity.
Microsoft has never faced pressures like that. Windows is not something you can simply swap out. Switching from Intel to AMD was simple, your software would hardly notice, but from Windows to another OS has always been a massive headache.
AMD usually played catch up, usually tot for tat making less money, got ahead big on 64bit Opteron, and then Intel caught up. Today, Intel bough Altera while people talk up Xilinx buying AMD. World of financial difference first mover, IP, and lockin can make, eh?
https://web.stanford.edu/class/e145/2007_fall/materials/noyc...