10 ms·
Gelsinger and Meyer: Two CPU Designers Who Changed the World
- tcskeptic 14y agoGelsinger spoke at my high school graduation, he was an alumnus of my high school. This would have been around the time that he was running the Pentium Pro project. What really struck me about his story was that he had graduated from the "Vo-Tech" course in my high school, and then bootstrapped himself through the ranks at Intel via hard work and pursuing education outside of work. If I remember, the first thing he did was get an associates from Lincoln Tech or similar.
- wglb 14y agoGood article, but contains a common misunderstanding: On the original 8086 there was no way to context switch. You couldn't run a multiprocessing OS on it. In 1983, Mark Williams Company demonstrated COHERENT, a Unix-workalike, on the IBM PC at the NCC. Later, MWC also produced COHERENT on the Atari ST, whose processor was a 68000 with no memory mapping. Multiprocessing worked fine on both. The PC was actually based on the 8088, which was an 8-bit bus version of the 8086.
- CurtHagenlocher 14y agoThe original version of Minix ran on the 8086 as well. What was really missing, as you suggest, was any kind of support for true process isolation.
- wglb 14y agoRight, as did QNX, i believe.
- dugmartin 14y agoI threw away my Coherent manual during my last move. It was as thick as a big city phone book and was my goto Unix manual for years. According to this it was over 1347 pages long: http://nesssoftware.com/home/mwc/manual.php http://nesssoftware.com/home/mwc/manual.php
- wglb 14y agoThere was a bit of a ceremony at the last MWC get-together--everyone there signed a moderately dog-eared copy of the manual Fred had. Has a good fraction of them, but the folks listed on that page are scattered pretty far and wide now.
- ChuckMcM 14y agoI joined Intel in 1984, in '85 when the '386 was coming out (it worked first pass which was amazing) and started looking at ways to get a 32 bit OS that fell to a guy named Ken Shoemaker and to some extent me as I had been helping keep intelca, a UUCP node, up which was running Xenix (Microsoft's version of UNIX System V). I got to work with Pat who was helping me figure out all of the 'reserved bits' the Microsoft had used in the descriptor tables on the 286 which were causing the 386 to fall over dead. (We were initially sad that Xenix didn't boot in 'compatibility' mode, when we figured out (which Pat's help) why we were rather irritated :-) He was one of my heroes at Intel, someone totally dedicated to his idea of what he wanted to do, and by God Intel could either help or get out of the way.
- yuhong 14y agoLook up "80386 ken shoemaker" on Google Groups for some history. Speaking of the 386, I wrote an article on the MS OS/2 2.0 fiasco here: http://yuhongbao.blogspot.ca/2012/12/about-ms-os2-20-fiasco-px00307-and-dr.html http://yuhongbao.blogspot.ca/2012/12/about-ms-os2-20-fiasco-...
- ableal 14y agoIs the illustrator making some subtle point with those three locked-up gears that cannot possibly move? P.S. Back in the it day was not only IBM that "was leery of a single vendor supplying the vital CPU." Having a second source for parts was considered essential in many cases. P.P.S. s/iAXP432/iAPX432/ as in http://en.wikipedia.org/wiki/Intel_iAPX_432 http://en.wikipedia.org/wiki/Intel_iAPX_432
- zokier 14y ago> With the Athlon, Meyer and AMD caught Intel flatfooted. Athlon introduced a dual core processor, and it moved the memory controller from a separate chip, the front side bus, onto the chip die. Most important, it was 64-bit No, the original Athlons were 32-bit CPUs. Then came Athlon XP -line of CPUs, which still were 32-bit (and still singlecore). After that came Athlon64, which was 64-bit. And singlecore. Athlon 64 X2, which finally were dual-core, came couple years later (and Intels dual-core Pentium D were released within a week of the release of Athlon 64 X2). I doubt that the original Athlons and Athlon64 X2's had very much in common, so I wouldn't really say that Athlon introduced dual-core or 64-bit. edit: I might add that having 64-bit CPU wasn't really relevant on the desktop back then, as until Windows 7 was released there wasn't a viable 64-bit Windows available. 64-bit Windows XP was delayed until Intel got its 64-bit CPUs out, and support for it was always poor. Vista was significant improvement in that front, but it was Vista.
- yuhong 14y agoAFAIK, it became common on new systems soon after Vista SP1 was released (around mid-2008 I think).
- rayiner 14y agoAthlon 64 (K8 core) was a fairly conservative extension of the original Athlon design (K7 core). Very similar, except widened to 64-bits, deeper buffers in a few places, and with the memory controller brought on board. See: http://www.realworldtech.com/looking-forward-2002/3/ http://www.realworldtech.com/looking-forward-2002/3/
- ksec 14y agoWell Gelsinger and Meyer are both already well known for anyone who has followed anything slightly in CPU world. I would be more interested to know insider story why Gelsinger left, or pushed out.