That's completely wrong, of course. On the high end, it is x86 that is
squeezing out the traditional RISC processors. Unless your world
definition of "squeeze" is much more loose, and includes such examples
as a rubber ball /squeezing/ the surface of a road, when it hits it: you
can argue that the road surface might flex microscopically, but the
rubber ball will show much more flexure.
Stick to bidding clubs, diamonds, hearts, spades, and no trump,
because sarcasm isn't your suit. It's very hard to be convincingly
sarcastic when you've just missed the point.
Besides, "high-end" isn't even
limited to just HPC work, it's the entire datacentre thing that needs to
be looked at. The datacentre is increasingly going x86.
I'm thinking of high and and low end in terms of processor complexity
and capability. It wasn't too long ago that someone in one of these
forums (and perhaps you) was stating with sarcastic certainty that Via
chips would never appear in servers. It's been known for a long time
that the be-everything-to-everybody chip that "high-end" x86 has
become is far from ideal for servers, both in terms of power
consumption and real estate. "Low-end" x86 will increasingly find
applications where "high-end" chips once ruled, and, once you are
trying to design for minimum power consumption and server footprint,
it isn't clear that x86 will be the long-run winner at all.
For those applications where the CPU is performing more compute-
intensive tasks, it's far from obvious how much of the work will wind
up in processors that bear little resemblance to, say, core i7. It's
like the early 90's again with respect to parallelism, except that,
then, the winner (often even for applications like databases) was
clusters of commodity processors--at a time when commodity processors
were much less complex than they are now.. Is it better to have a
computer full of atom or via or arm, or is it better to a have a
computer where most of the work is done by processors that look more
like GPU's? At this point, I don't think anyone knows. How much will
be left for the von Neumann bottleneck of yesteryear? Not much.
Similarly, it isn't clear that x86 will be the long-term winner in a
mobile space that looks more and more like what used to be called the
embedded market.
Calling specialized HPC stream processors a high-end threat, is to even
insult the very definition of a high-end processor. You can't even run
applications on a stream processor without assist from a CPU, and
usually that CPU is an x86 rather than a RISC.
You can run full-up Linux or BSD on stream processors. Even were it
convenient to have a more conventional CPU around to to things like
cope with I/O and interrupts, that CPU doesn't have to be particularly
sophisticated or capable.
And actually, it's the
threat from a CPU and a stream processor combo that's also creating a
bigger threat to high-end RISC: a stream processor is more specialized
than a RISC, while an x86 is more generalized than a RISC, together it's
a threat.
But the "host processor," whatever it is, doesn't have to be much,
because the muscle will be in the stream processor.
I have no idea whether you're insulting Opteron or Itanium or IBM.
Sometimes you're just a big sarcasm shotgun with no fixed target.
Intel and Microsoft are competitors and allies-of-opportunity for
IBM. AMD is just an ally.
That's right, after nine years of investigation (complaint filed
originally in 2000), plenty of opportunity for Intel to respond but
failing to do so, in the end the EU was just an emotional wreck and
chose based on "feelings". The same goes for Japan and South Korea
before it, who also found Intel guilty. Three for three, all these
countries have it in for Intel.
The question is, guilty of what. Did Intel make life hard for AMD?
Without a doubt. Did it do so with malice aforethought? Hard to
believe the answer is no.
But did they harm consumers? The EU's position (and yours and the
position of many others here) is that consumers are harmed if there is
market domination by one company. There's no proof of that. It's
conjecture or mostly an article of religious belief on your part.
Anyways, VIA's entry into the Dell low-end server portfolio is an
example of all boats now being lifted as a result of AMD's takedown of
Intel. It's probably not a coincidence that Dell felt confident enough
to introduce a VIA-based server just a couple of days after Intel lost.
The VIA deal has been in the works for a long time. It's true that
Dell is probably less afraid of Intel than it once was. In any case,
what you are arguing now vitiates your initial claim that nothing of
material importance has happened.
You're still invoking the stock market as an indicator of anything, even
now?!? It's plain to see that the stock market was _at best_ a Ponzi
scheme, and at worst a rigged casino. The only thing it was an indicator
of was its own greed. If the best you can call it is a Ponzi scheme, if
that's the moral high-ground, then that's like comparing the seven
levels of Hell: you don't want to enter any of the levels.
If you'd said that capitalism was a Ponzi scheme, I'd have agreed with
you. Markets actually work pretty well, except when they don't. When
don't markets work well? They don't work well when they are gamed or
rigged, which is part of what has gone so badly wrong. It was debt
and not equity markets, though, that were rigged through manipulation
of bond ratings. Markets also don't work when everyone starts doing
the same thing, which is what happened with valuation models being
used by Wall Street. Again, it was debt and not equity markets that
were affected.
The odd thing about Ponzi schemes is that, locally, they work just
fine. Investors look at past performance and bet on what they take to
be winners. The problem with all Ponzi schemes (including capitalism)
is the failure to recognize global constraints: there isn't an endless
supply of new investors, the economy cannot grow at an exponential
rate forever, housing prices cannot go up forever at a rate that
exceeds the general rate of inflation, and there is not an endless
supply of cheap labor and natural resources to exploit.
If you listened to the stock market, then it's betting was that AMD was
already dead two years ago. In fact, one hack, Rick Hodgin, even wrote
an article wondering aloud why his prediction that AMD would be gone
didn't come true:
Which is why I pointed out the recent performance of AMD. Clearly,
something is up.
What are you talking about?
If you go to an investor and say, "If you give me more money, I can
turn my money-losing business," the investor will say, "How?"
The "How" in this case is in part government intervention in markets.
Robert.
