[Info-vax] "x86 has only a few years left in the market place"

[Dave Froble]

On 5/21/2018 2:32 PM, seasoned_geek wrote:
> On Monday, May 21, 2018 at 11:29:18 AM UTC-5, Johnny Billquist wrote:
>>
>> But yes, DEC did do big iron back in the past, and then pissed all of
>> those customers off by killing the architecture after promises of new
>> machines again and again. And insisting that the customers migrate to
>> VMS, which was rather inferior compared to TOPS-20 (and refusing to
>> implement most of the good stuff from TOPS-20 on VMS). Almost none of
>> those customers migrated to VAX/VMS, and most of them swore to never use
>> DEC again.
>>
>> That was bad for their big machine business market. After that, DEC was
>> pretty much restricted to mostly existing customers and the mini market,
>> and for a while networking which was better than any competition. And
>> that sold VMS quite well for a while. But at the price DEC wanted,
>> workstations and later PCs pretty much ate DEC up from beneath.
>>
>
> It's up or out in IT.
>
> The absolute bottom of the industry will rise up to consume the food
> above it. We've watched this time and again. Wang clung to word
> processing and basically got destroyed by Word Perfect. Same has
> happened to all one trick ponies _except_ big iron. The bottom of the
> market cannot hold a candle to the I/O throughput of those machines.
> Not even in their most drug induced fantasies.
>
> Today ARM is the absolute bottom of the market. It's primary selling
> point is 85% reduction in power consumption with roughly the same
> performance. Yes, someone is going to fish out a chart comparing some
> version of ARM to some other Xeon or whatever, but it doesn't matter.
> Unless you run at 100% of capacity 100% of the time they are borris
> for most, especially when the TCO is dramatically lower. For most
> companies, even if INTEL gave them servers for free it would still
> make financial sense to move to ARM.
>
> How many of you worked in shops which brought in a VAX to "try it
> out" only to find they could cluster a few of them together to obtain
> 90-100% of what they needed for 1/3 the money? To "cut costs" they
> gave up the extra throughput and whatever.
>
> ARM is moving up. We are once again facing a tipping point _exactly_
> like CPM and DOS. x86 has an installed base, but, it is a chip with a
> short life. Never one known for "up times measured in decades."
>
> You know, my Raspberry Pi is over in a plastic file box because I had
> to clean off that table for another client project, but, all of this
> talk about them not taking over the desktop consumer market is B.S.
> I've used it to surf the Web, post here, answer email, even wrote a
> couple of chapters of a book using mine. Heck, I've compiled all of
> Qt on mine (well, all but the Web stuff) and built applications with
> it. You just have to use a class 10 SD instead of the cheapest one on
> the market.
>
> So, the absolute bottom of the market is climbing forward at Internet
> pace. The very tippy top of the server world is starting to be
> consumed by the Z-Box leaving INTEL with a piece of pie shrinking in
> size at an exponential rate.

Well, it was ever thus ....

When mainframes, such as they were then, were the only game in town, 
everybody used them.  Then came the minis, and they could handle some of 
the jobs better than the mainframe, at less cost, and there were new 
apps that didn't exist.  Then came the micros, and they could handle 
some of the jobs better than the mini.  Along with more new apps.  Then 
came notebooks, tablets, and phones, and yes, that Pi.

It's just technology advancing, and some of the jobs they can do will 
use them, for cost and other benefits.  But always, the smaller did not 
do all the jobs well, and is unsuited for such jobs.  The key thing is, 
the jobs unsuited for downsizing are still there, and being performed on 
the micros, minis, and mainframes.  Yes, some jobs were lost to smaller, 
cheaper, better, but they were never well suited to the larger, 
costlier, ....

Now, you are partially correct, x86 volume will decrease, and in so 
doing, lose some economies of scale.  This has been happening for some 
time now.  Perhaps Intel will not find the chips so profitable, nor do I 
see the price going up, AMD will see to that.

Key thing, not once has the newer, smaller, cheaper, totally replaced 
the technology preceding it.  I doubt it ever will.