Archived from groups: alt.comp.hardware.pc-homebuilt (
More info?)
JS wrote:
> David Maynard <dNOTmayn@ev1.net> wrote in
> news:10ouspsa78ckcc3@corp.supernews.com:
>
> Let me just say at the outset that I don't disagree with your comments. I
> would like to add some additional thoughts peculiar to the semiconductor
> business. Again, some of this is not going to be 100% relevant to CPU's (or
> to other businesses for that matter).
>
>
>>JS wrote:
>>
>>
>>>"Mr Koko" <mrkoko@comcast.FISHnet> wrote in
>>>news:m9CdnSi81sYN7xDcRVn-rw@comcast.com:
>>>
>>>
>>>
>>>>There is something that I just don't get.
>>>>
>>>>When price shopping the older .13 micron cpu w/512kb L2 costs
>>>>slightly more than the newer 90 nanometer cpu w/1mb L2. Both being
>>>>the same speed.
>>>>
>>>>Ex:
>>>>Intel Pentium 4 2.8GHz, 512K 800 FSB Socket 478 Hyper Threading
>>>>Boxed
>>>>Processor $179
>>>>Intel Pentium 4 2.8GHz E , 1MB 800 FSB Socket 478 Prescott Hyper
>>>>Threading
>>>>Boxed Processor $178
>>>>
>>>>Not that the price differs much but, why isn't the newer/ more cache
>>>>cpu more?
>>>>The only downside that I found out about the 90 nm cpu is it runs
>>>>hotter.
>>>>
>>>>Input Needed,
>>>>Mr Koko
>>>>
>>>>
>>>
>>>
>>>In general, for semiconductors, as a part reaches end of (production)
>>>life many manufacturers will increase its price to discourage
>>>continued use of the part.
>
>
> This is probably rare but has happened in the past. A different tactic is
> used more regularly nowadays. (See Below)
>
>
>>Which, in general, would simply drive purchasers of that part to a
>>competitor.
>
>
> This is a potential issue for any business, but don't forget that I'm
> talking about 'end of life' parts and specifically about the semiconductor
> business. There is the complication that continued support of a given part
> may require continued support of an old technology. Supporting old
> technologies may be more expensive, both in terms of money and resources,
> than is economically feasible or prudent. Manufacturers frequently have to
> move to newer technologies to keep up with customer requirements and
> competitors technological advances. In fact, this happens frequently in the
> semiconductor business. So much so, that manufacturers regularly notify
> customers of impending manufacturing 'discontinuances'. These
> 'discontinuances' are often referred to as 'End of Life' notices. But your
> correct in that they typically do not involve price increases per se.
> (Customers are required to place final orders though.)
You're using a lot more words to say the same thing I did. Market volume
shifts to the new technology rendering the old unprofitable from lack of
demand.
> Another complication
> is that, in some instances, there is no competitor to turn to when the 'End
> of Life' notice is issued.
There's always the 'last guy' to give it up.
> Furthermore, other manufactures may also want to
> quit supporting old technologies so that they will issue 'End of Life'
> notices on their equivalent parts as well.
If everyone is dropping it then you can be dern sure there's little or no
demand for it or else someone would get the bright idea to keep the line
and rake in the volume and profits from all those other 'idiots' who EOLed
theirs.
> This is relevant here because I
> believe that Intel will not attempt to support both cores for an extended
> period.
Of course not. Just as they don't want to make P133 cores anymore. There's
no demand for it.
> Intel wants to move all it's production capacity to the new
> processes and will stop producing Northwood cores as a result of this
> desire.
You beg the issue, though, by simply calling this a 'desire'. What I was
getting at is the *reason* for the 'desire' and that it isn't just a
cavalier decision or some pervert who thinks it would be 'fun' to 'screw'
the buyers.
> Prescott, and the associated newer manufacturing processes, is
> their future
Bingo. And 'why' is it the future? For the reasons I explained.
> and continued production of Northwood cores wood use resources
> better applied to making Prescott cores.
Which is the manufacturing transition period I spoke of.
> So, you can see that making
> Prescott cores less expensive may be viewed by Intel as an enticement to
> get their customers to use Prescott cores.
Someone may 'view' it that way but it isn't what's driving the process.
And if it is, in fact, being done it is one of those 'bumps' in the
transition I spoke of; where the business plan's estimates of when and how
fast the market would shift to the new product isn't playing out as
expected (or maybe you're competitor is taking market share for some
reason, or the market itself has shifted due to economic pressures). But it
is, again, a market supply/demand issue. You've prematurely ramped up
production on the new product and are facing either excess inventory or a
price cut to move it.
> Since there are no second
> sources for Northwood cores that is not an issue here. (I think they
> probably disregard losses to AMD processors since they probably believe
> they have a good alternative to Northwood core CPU's, namely the Prescott
> core CPU's and so losses would be insignificant or none existent.)
If the Prescott is, indeed, a 'better' processor than the Northwood then
the issue is moot because it *will* be transitioned to Prescotts and we
simply discussing if their production timing is well planned. If it's not,
which seems to be the implied 'accusation', then AMD would indeed be an
alternative since Intel would have made a catastrophic error and 'missed
the boat' on the next generation of competitive products.
But we've strayed from your original premise of 'the semiconductor business
in general' in dealing with a proprietary, single source, product.
>>>Sometimes, distributors, and other profiteers, will buy up supply
>>>at the wholesale level and reap the benefits as manufacturing ramps
>>>down (supply dwindles while demand does not) and prices go up.
>>
>>The problem with your theory "supply dwindles while demand does not"
>>is that no one in their right mind is going to ramp down production
>>while demand remains robust and profitable. It flies directly in the
>>face of "supply and demand."
>
>
> Very true, but again, in the semiconductor industry, supporting an old
> technology may be more expensive than not. Particularly if a manufacturer
> sees more profit in another (newer?) technology or part.
Again, you're begging the issue by not dealing with *why* supporting an
'old' technology is not profitable. Little or no demand.
> In this case, I
> believe that a manufacturer may feel that resources should be moved to the
> newer technology or part.
Of course. Because that's where the demand is.
> I do believe however that your argument is valid
> in that some 'end of life' notices, and therefore, 'end of technology'
> notices have been delayed, perhaps for years. In some cases industry
> customers have managed to exert great influence on manufactures, by keeping
> demand high and being very vocal in their discontent, on some very mature
> technologies. Many manufacturers and industry reporters have been proven
> wrong, time and again, for the very reason you state,
Right. I'm talking of how the market works and how a manufacturer deals
with it, if they're *right*. If they screw up then you have the kind of
situation you just described: the market forces an adjustment.
> but many mature
> technologies have gone the way of the Dodo as soon as manufacturers could
> get away with it.
The way you word that is what I have a problem with, "if they can get away
with it", because it ignores the reasons why they give a tinker's dam. As I
said, they'd gladly make buggy whips if enough people would BUY the things.
What they want to 'get away with' is "Oh, please don't try to make me keep
a 500 million dollar plant online just to make the 100 parts a year you
want to buy at 50 cents each."
>>The classic notion of supply vs demand price fluctuation is fine, and
>>appropriate, as long as everything else remains the same (and the
>>market is elastic) but, in the real world, they seldom, if ever, do.
>>In particular, prices can go up in a 'perfectly matched' supply vs
>>demand market if demand (and the matching supply) is decreasing
>>because the economies of scale are lost. And, just as economy of scale
>>often increases demand because more can buy at a lower price, losing
>>it can depress demand as the price increases; which is probably what
>>leads to the fanciful notion that it's 'purpose', rather than natural
>>result, is to "discourage continued use."
>
> Good point, but how natural is it when a manufacturer decides to ramp down
> production in the face of steady demand.
I dispute the premise of "steady demand."
> As I said previously, Intel does
> not want to support both Prescott and Northwood at the same time. Northwood
> is last years model and as such is a resource drain and a hindrance to
> forward progress.
The demand for Northwood will *not* "remain steady." Whether they've timed
their process transition 'perfectly' is a different matter.
> But realistically, I don't know if its Intel or the
> retail vendors pricing choices causing the Northwood prices to be higher
> than the Prescott prices. I think that it is probably the retailers
> smelling a little extra profit to be had.
Intel may be off a bit on their process ramp schedule. Or it could be that
Northwoods, being a 'larger' process and, hence, fewer dies per bake, are
inherently more expensive to make.
>>To wit, manufacturers would love to build the same thing forever
>>because it's predictable. You know how it works, how to make it, what
>>the costs are and all the rest whereas the 'new' thing is full of
>>development costs and unpredictable problems. But you can't just keep
>>making the same old thing because Mr. Competitor will come out with
>>his 'new and improved' version to take market share so you have to
>>have your 'new and improved' version in the works to keep even, or
>>maybe take HIS market share. And that means, eventually, your
>>manufacturing is going to transition to the 'new and improved'
>>version
>>because that's what people will be buying (unless you screwed up). In
>>the process you're losing sales volume (people want the 'new and
>>improved') and economies of scale on the fading product till it just
>>isn't worth making any more. In practice you try to predict the
>>timing
>>of that transition in your business plan so, as Murphy would have it,
>>there may be unusual, transitional, 'bumps' along the way, resulting
>>in price oddities, if things don't pan out precisely as planned but
>>you don't just arbitrarily dump a profitable product.
>
>
> Of course not, but sometimes the march of technology demands resources, and
> resources have to come from somewhere. You can't just keep adding a new fab
> every time something new comes out;
That's why I speak of the process transition.
> and again, Intel wants Northwood to go
> away. Again, more profit may lay with the newer technologies than the old
> so, 'Out with the old and in with the new.'
Begs the question again. *Why* does 'more profit' lay with the new? Because
market demand for the old dies.
('More profit' is 'more' than staying with a dying product but it doesn't
necessarily mean 'more' than you're doing now. You have to keep improving
product just to stay even.)
>>Now, if you folks would quit buying the 'new and improved' versions
>>and be satisfied with P233MMXs and 32 Meg of RAM SIMMS running Windows
>>for Workgroups 3.11 we wouldn't have this problem
>
>
> You are so right, but I must admit I have yet to use a computer that I
> thought was fast enough. Microsoft will probably make it impossible for
> that to happen, bloat ware and all. At least I have stopped at a PIII-733,
> for now anyway.
Considering that clock speeds are now in the 5 times faster range you've
got some headroom there. <g>