Re: PowerPC desktop -- asking for trouble?
On Thu, Jan 06, 2005 at 11:53:01PM +0200, Leandro Guimarães Faria Corcete Dutra wrote:
> > Yeah, indeed... it really seems that IBM doesn't want Power(PC) to be a
> > successor.
> I assume you meant simply success, as you didn't point successor to
> what -- I'd assume successor to the IA-32.
Yes, of course success... :)
> > But that's somewhat IBM typical. Remember OS/2. There's a long
> > term support for OS/2, but over all the years, IBM failed to really make it
> > popular.
> Not only OS/2, but also PS/2 and various generations of POWER.
> Remember the IBM PowerPC workstations and notebooks running OS/2?
Yes, there are several examples how bad IBM marketing sometimes is - if it
exists at all! ;)
> > Freescale aims at a different market than IBM. IBM goes for big iron systems
> > and midrange servers, Freescale for embedded systems and low end.
> Yeah, but these Genesi systems really look underpowered. I'd still buy
> them if I could -- would have done if I hadn't lost my job in
> Switzerland --, but they really offer no competition to mainstream PCs.
Yes, the peg2 boards are somewhat behind in some regards. There's still the
possibility to replace the CPU module by a faster one. Sadly, there is none
available. It would be nice if some third party vendors would bring some to
market - but that market is too small yet.
Furthermore it would IMHO make more sense to put the RAM modules as well on
the module and just keep the basic things on the mainboard such as IDE, PCI
and other IO ports. Dunno if this is possible with all those North- und
Southbridge thingies...
> And IBM simply won't go cheap enough.
Indeed. So there's just Apple and Genesi and Hyperion out there.
> > Whereas it would be better for the users when there would be a competetive
> > market with several vendors to make the prices drop, it might not be that
> > good for the vendors (and in the end for the users, too).
> Hm, looks like a disguised argument for anticompetitive, economically
> inefficient behaviour...
Not really. I would like to see a good competition bringing down the costs
and keeping the high quality. But from my experience and what I've seen, the
costs goes down first, followed by the quality and then, sometimes, by the
vendor going out of business.
> > I've some serious 3D graphics background and used Alias PowerAnimator on
> > SGIs and made the transition to Maya. [...] CEO driven expedition into x86 land and
> > returned to their core business. Luckily Beluzzo left SGI to join MS
> > somewhen. But the workstation market is now gone. The sales driving need for
> > Maya has gone for ever.
> I fear this is not relevant to the discussion at hand... what you told
> is just a tale of corporate stupidity, or perhaps not; you didn't
> mention what perhaps drove SGI to that course of action, namely
> diminishing competitive advantage. Granted it is stupid to react to
> diminishing advantages by shedding any advantages (substituting Wintel
> for Unix and RISC).
I think that tale tells what might happen to somewhat "closed" markets when
they try to compete with the mainstream (PC market) and to open their niche
market. It's most likely that this will end in a loose-loose situation for
both: the vendor and the userbase.
There are many examples where not the best solution won the competition but
the cheapest: VHS vs. Beta, Amiga vs PC, Mac vs PC, SCSI vs IDE, ...
Good things are more expensive than cheap (cheap on price and
quality/performance) things. The Joe Average User tend to save 5 bucks on
buying something, instead of buying good stuff.
> > So, overall it might be a wise decision of Apple and IBM to protect their
> > niche market. They have a sort of fanatic userbase that are willing to pay
> > higher prices which you need to develop a low-number-sales arch.
> > On the first look this high price politic seem to be bad for the user, on a
> > second look it's good for the user because it ensures that the vendor stays
> > in the market and can afford development of new hard- and software.
> But how bad would be to IBM workstations and Apple Macintoshes if we
> could buy GNU/Linux, white box G5s if they already have their fanatical
> user bases?
A part of that fanatical userbase is that they have the feeling of being
something special or have special hardware. If anyone have that piece of
hardware those individualists will go and search for something new special.
There are Apple users that are Apple users because Apple is Apple and not
just an ordinary PC.
> Apple will still have their integrated, proprietary Mac OS X and
> Macintosh. IBM would still have AIX, DB2 etc. But we'd have cheaper,
> more powerful, more energy-efficient machines, and both Freescale and
G5 and energy-efficient?
> IBM would have a much easier time selling chips, patents licences,
> whatever. The whole platform gets validated, the tide rises all ships.
I doubt that this would be that simple. If it were, I believe IBM would be
first to take that chance to make money fast. But I think this could be a
short-sighted view.
> See how Intel has so much success even with absolutely mediocre, even
> antieconomical designs. It all comes down to scale, and one reaches
> scale only by having a healthy economical system with good critical
> mass. The PC system is so strong today that it would survive even Intel
> floundering, while any player in the POWER space would be a huge miss.
You need high sales numbers to get your ROI for development in a low-price
and high competition market. In a quite protected niche market you can (and
of course have to) have higher margins when selling hardware which results
in higher prices for the end user of course. But usually you can count on
some sales number predictions, whereas you can't rely on your predictions in
a multi-vendor high-competition market, because there always might be
someone else that is 10 cents cheaper as you.
For Apple this could mean: "Hey, we're about to ship a new machine... the
fans out there are that curious about that machine, that we can expect
100.000 units within the first 3 months. So, let's calculate the price to
get the breakeven point for ROI within that 3 months! Anything above that
100.000 unit mark will be our profit!"
In a PC-like market you might have the chance to sell 5 Mio units within the
first 3 month - or only 10.000 because someone else has a better/cheaper
product or even worse: a worse and more expensive product, but better
advertising.
Being a vendor, what way would you go?
> > Having cheap hardware is not always a benefit to the user. Sure, they can
> > obtain hardware quite cheap, but on the long run PC market showed that the
> > quality begins to drop. Quality assurance of products is expensive and can't
> > be done anymore when you compete on a below dollar per component sales
> > margin.
> This is an elitist argument -- but anyway it is false. You do can buy
> high quality PCs, with ECC RAM, dual SCSI buses, multiprocessed Xeons or
> Opterons and what not, and these will be expensive. So the need for
> high quality stuff is no excuse to block the existence of cheap stuff.
Of course you can buy high-quality x86 based hardware - at higher costs.
Go and buy some servers from Dell or Compaq or whomever. But that's not the
typical PC market where profit margins of single machines are very small.
For a server it's more important to have good performance and good
reliability, so most servers will run with SCSI drives instead of IDE disks
for a reason. But no computer discounter will sell machines with SCSI drives
at a competitive price, because it's simply not possible.
> > And because such a high competition market as the PC market is, you need
> > fast release cycles and always be present with new and shiny products, so
> > that the users are sometimes (increasingly) punished with buggy and not well
> > tested devices.
> Again, not quite true. If users are punished, it is their fault for
> believing the MHz hype.
Rule #1 of computing: users are dumb ;)
Of course they believe that MHz hype and of course they thing they will
always get the best hardware for as less money as possible.
And this competition leads to quality problems. Remember when IBM had
problems with the quality of their IDE drives? I had some of those fscking
drives. Within two days a new IBM drive turned from being new to being
defect garbage. Some other users reported to me that they had IBM drives
that made that turn within two hours. That was a real low price, low quality
problem that IBM had back then.
All this price competition leads to a situation where vendors, that can
compete anymore, are bought out by others. Quantum was bought by someone,
IBM sold its drive storage division to Hitachi and so on... This results in
a a turnover from a polypoly (sp?) into a oligopoly or even a monopoly structure
of the vendor market. Definitely a polypolical market is quite good for the
consumer, because s/he can choose from many vendors. A monopoly is not good.
See MicroSoft.
> But true competition would even enable nobler
> competition, because different architectures would make
> performance-based measurements the norm -- even Intel felt the need when
> introducing the Itanic -- and open standards would allow us to focus on
> reliability and performance instead of just keeping pace with Redmond's
> latest.
Of course that would be the idealistic goal. Unfortunately not every arch of
an multi-arch market can be produced at the same cost. IBM used its market
power back then to establish the PC market. MS and Intel are the winners of
that history. Commodore, Apple and Atari are for example the losers,
although that had the far more advanced machines at that time.
> > But because I don't like all these Apple bubble plastik designs, I'm happy
> > that there are now other vendors that enable me to migrate from x86 based
> > computing to PPC based computing. And I don't need that high speed big iron
> > CPUs of IBM for my machines at home. Even the G3 is idling at over 90%
> > during the day. Why should I then have a higher speed CPU that idles even
> > more but consumes much more electricity? :-)
> What about having a dual G5 at the basement to serve a terminal or two
> at each room, including the kitchen?
Nothing's wrong with this. Just go ahead and buy a dual Apple Xserve and
some sort of terminal multiplexer or multihead grafics pipe, if you can
afford. Other people might have requirements and imaginations of how they
want to solve a problem. ;) But I doubt you need the power of two G5 to know
that your fridge is empty. Maybe just open your fridge and look into it will
be more convenient for most households... ;^)
> And it's not about my personal needs, but the whole market. Great
> performance is needed to validate a platform for gamers, programmers,
> scientists and engineers.
Of course there is a market for those. But the market is not flat and plain,
it differentiates into many niches. Apples high price, high-protecive niche
market is just one of them. IBMs high performance computing is another
niche as it is for SGI to sell Altix super computers - or Genesi/Hyperion
for serving the still existing Amiga fan market and expanding it to people
who are just sick of x86 hardware and want something special. :)
> > Long posting, short summary: I'm quite happy about the current situation of
> > the PPC market. Let Apple/IBM do the G5 stuff and Genesi/Freescale the rest.
> > :)
> Still too few sources, and too limited combinations. There need be
> more sources and more distinct products -- or at least configurations of
> them.
Yes, more sources, more vendors. Therefore I bought my peg2 - because I know
that the money I pay will go to a german manufacturer instead of some
fareast, multimillion dollar company in Asia.
Of course you can complain about that high prices - but then you can't
complain about the diminish of vendors. Do you remember the time in the 80s
and 90s when there were everywhere small computer stores in the neighborhood
with some strange looking geeks that had more similarity to vampirs than
human beings because they were hacking and working the whole day and night,
living and sleeping in their shops? Where have they gone?
They were forced to close because there were big companies selling the
hardware for some less bucks than those geeks.
If you want more competitions and more vendors, you have not to buy the
cheapest pieces of hardware. Support your local vendors and shops!
I've seen several small shops and vendors passing by (although they
delivered good work and high quality) just because there were someone else
that was 5 bucks cheaper in selling pieces of crap.
So, yes, Apple and IBM are more expensive and more powerful with their
machines, but with supporting (i.e. buying) Genesi and others you have the
chance in your own hands that there will be a growing competition on some
day.
And yes, I would like to have a supercomputer at home, but not at the cost
of higher power bills for electricity. In the last few years, the costs for
electricity raised dramatically in Germany. A G5 that consumes about 100W is
just too much. A 8W consuming G3 is sufficient for DSL, routing, firewalling
and file serving.
--
Ciao... //
Ingo \X/
Please note that year 2004 has come to an end and
the year 2005 is now - even in my mail address!
Reply to: