Replies to post #135105 on Intel Corp (INTC)

[chipguy]

It will eventually have to produce ARM chips

Why? We already know you have a hard-on for ARM but could
you at least try to articulate a few objective facts why you think
that the ARM architecture has some secret sauce that will allow
it to succeed against x86 in general purpose computing when all
other RISCs are either dead or in rapid retreat into tiny proprietary
niches. The major historical trend of the computer industry is quite
clear yet you argue against it.

[Tim May]

Intel is doing fine with x86, will do fine with ARM

ARM may stand for "Acorn RISC Machine," or at least once did (probably back-formed by marketing droids into something more marketable by now), but the RISC part is no longer meaningful. As we all know, modern CPUs are multicore, superscalar, out of order/register renaming, speculative execution, etc. The P6, aka Pentium Pro, had a little tiny core of x86 instructions which then expanded-out across the entire chip area. The development team even matched the physical chip layout in terms of where they were located, vividly described by Bob Colwell in his book about his experiences at Intel.

So, as certain ethnic groups like to say, "major props to ARM." It's a respectable competitor to Intel and should not be casually dismissed (or "dissed," as those same groups like to say).

As an owner of Intel shares for 40 years now, I'm glad it's not as bad as the days when some Really Serious Competition existed. Remember the 68000? The Apple/IBM/Motorola alliance to develop an Intel-killer, the PowerPC? The SPARC? And I remember the days when Intel was facing the Ten Feet Tall Japanese--Hitachi, NEC, Toshiba, Mitsubishi, Fujitsu. And MITI and the Fifth Generation project.

Intel is thriving in the PC and server markets. Heck, even Apple is building ALL of its laptops and desktops using Intel CPUs. (I just bought a Mac Pro, the cylinder, which is based on the Xeon. I got the 6-core, 12-thread model, 512 GB of fast SSD on a fast PCIe bus....a lot of Intel technology goes into optimizing this very fast desktop).

Not all is roses, though. My concerns are about the transition, and its horrendous development and fab costs, to sub-10 nm dimensions. Frankly, I doubt whether _anybody_ can afford this. This is sometimes called Gelbach's Law, in parallel with Moore's Law, about the doubling or tripling of fab facility costs. Do the math.

And we enter uncharted lithography territory when dimensions are 5 nm but interatomic spacing is comparable to this. I'm glad I'm retired lo these many years.


--Tim May