PC 4GB RAM limit

[David Maynard]

David Maynard writes:




Not really. It only takes a single use of a single instruction that
doesn't exist on the 486 to make the software incompatible,

It might "only takes a single use" in theory but that isn't the reality of it.

and it
doesn't surprise me that many Linux distributions might be this poorly
written.

That's like saying anything using capabilities past an 8080 is "poorly
written."

Technology moves on.

 

[David Maynard]

David Maynard writes:




Nobody has pointed a gun at my head.

I didn't really think so.

But changing anything on a
computer today often releases an avalanche of update and upgrade cycles
that force many software products to be changed whether one wishes to do
so or not. In some cases new hardware is required as well. And even
trivial changes can do this.

But there's no reason to change since you argue there are no new useful
features and no advantage to faster processors.

Exactly.

Then use it.

Those who use them are often a minority.

I'll be happy to see your poll numbers on that, not that it matters even if
true. All that matters is if the number is significant enough to justify
the expense of incorporating the feature.

No, I'm simply arguing that there is too much running on the system.

But the argument is based on nothing.

I 'feel' 45 is right. No wait, make that 67. How about 29? 241?

 

[DevilsPGD]

In message <[email protected]> Mxsmanic

Not really. It only takes a single use of a single instruction that
doesn't exist on the 486 to make the software incompatible, and it
doesn't surprise me that many Linux distributions might be this poorly
written.

I'm about as far from a linux zealot as you can get (I run Windows both
on desktops and servers), but this isn't a matter of being poorly
written.

As long as the system requirements are documented, it's not poorly
written, it's design goals which don't meet your needs, and I'd
recommend you purchase another product.

 

[David Maynard]

David Maynard writes:




I was including everything in it, yes.

I take it you mean 'everything' except the more than 12 items you snipped out.

 

[David Maynard]

David Maynard writes:




Most of it was profit as well. The mainframe business is a major cash
cow.

myth

 

[CBFalconer]

.... snip ...

"...make the correspondence more useful to all...." is certainly a
worthy task. Top posting is a great aid to skimming posts because
a preview pane shows new material immediately rather than after a
fitful scroll. Snippy remarks, on the other hand, don't help 'make
the correspondence more useful to all. When adding to a cross-posted
thread civility is even more important than usual.

OTOH a properly snipped article will normally require no action to
examine for skimming purposes, and will maintain proper context if
not topposted. Some useful links below. Crossposting should not
exist past the initial query, which should have set follow-ups. I
am answering in a.c.h, and am quite willing that the fups eliminate
all else.

--
Some informative links:
http://www.geocities.com/nnqweb/

http://www.caliburn.nl/topposting.html
http://www.netmeister.org/news/learn2quote.html

 

[CBFalconer]

I didn't say they were. But that some Linux variants will not even
run on 486s suggests that the differences are not as 'minor' as
claimed.

If the binaries have been compiled to use the added instructions,
of course they won't execute. There are other variations between
Intel and AMD flavors. However I maintain that a compilation that
does not use the extensions will have, at most, a minor performance
penalty. One is the instruction that measures cpu cycles used,
which is purely informative. I forget just what it is. However it
is trivial to write code which checks for cputype on startup, and
does the appropriate things.

--
Some informative links:
http://www.geocities.com/nnqweb/

http://www.caliburn.nl/topposting.html
http://www.netmeister.org/news/learn2quote.html

 

[David Maynard]

David Maynard writes:




Not today. The hardware is not expensive to build.

It's more expensive than a PC.

Only marginally.

Then go buy a mainframe.

They can be on the same board, although they probably
won't be.




I was not suggesting that they replace PCs.

You're arguing a 'PC' should use the same architecture, which also means
expense, and that is equivalent to 'replacing' it.

A rose is a rose.

 

[David Maynard]

David Maynard writes:




Not in the average desktop or laptop.

I'd love to see your poll numbers. But it doesn't really matter if it's an
average desktop or laptop, an above average desktop or laptop, or a below
average desktop or laptop. They're still used in 'PCs'.

I'm describing operating systems and applications, not myself.

No you're not. You're describing how you 'feel' about them.

Try typing that number of pages. It helps create an appreciation for
the real size of memory.

Four gigabytes is 1.6 million pages of text. Try typing that.

Why? I've already intimated I do more with my PC than just type text.

And I've been real kind in not pointing out there should be a 'program' in
there somewhere.

 

[David Maynard]

If the binaries have been compiled to use the added instructions,
of course they won't execute.
Right

There are other variations between
Intel and AMD flavors.

Distributions are generally not manufacturer specific, they're architecture
and instruction set specific. For the x86 family that would be the 386,
486, 586, and 686 kernels, each of which pretty much precludes use of the
earlier processor types. (nothing prior to 386 is supported)

Suse has moved to the 686 kernel exclusively but you can get any of them
via Debian.

Of course, more than 4 gig (the origin of this thread) precludes anything
prior to the 686 (in 32 bit) since PAE didn't exist before then.

However I maintain that a compilation that
does not use the extensions will have, at most, a minor performance
penalty.

Makes one wonder why they ever bother with improvements.

One is the instruction that measures cpu cycles used,
which is purely informative.

Information is a useful thing.

I forget just what it is. However it
is trivial to write code which checks for cputype on startup, and
does the appropriate things.

Talk about 'bloat'. Why would I want the system to keep multiple copies
just so it can perpetually rediscover I've got the same processor and then
waste lord knows how many resources hopping around the system informing and
reconfiguring everything?

 

[David Maynard]

Oops, hit the previous track button.

Darn. And here I thought you were infallible. Oh well, another myth
shattered

 

[David Maynard]

In message <[email protected]> Mxsmanic



I'm about as far from a linux zealot as you can get (I run Windows both
on desktops and servers), but this isn't a matter of being poorly
written.

As long as the system requirements are documented, it's not poorly
written, it's design goals which don't meet your needs, and I'd
recommend you purchase another product.

Good point and well said.

 

[Conor]

No, I'm simply arguing that there is too much running on the system.

**** me, don't want to use Linux then. Out of the box installation of
MDK10/Suse9.3 has over 30 processes, including servers, running after
setup.

 

[Conor]

Not really. It only takes a single use of a single instruction that
doesn't exist on the 486 to make the software incompatible, and it
doesn't surprise me that many Linux distributions might be this poorly
written.

Wrong. The only time you'll get this is if you try and use code
compiled for, say, 686 on a 486.

 

[Conor]

Most of it was profit as well. The mainframe business is a major cash
cow.

Only if you're stupid enough to think that you need a quad Xeon with
4TB of RAM just to serve files.

 

[Conor]

Yep, but then you have to go find the CD and put it in. *&%&$# Microsoft.

And? Can't have it both ways.

The 'easy way'. On the command line. You do grok grep don't you?

ROFLMAO. Way to go. Lets just undo 20 years of advancements. Hell, even
back in the days of DOS the CLI was getting to be such a PITA that many
DOS based GUIs started to appear.

 

[Mxsmanic]

It might "only takes a single use" in theory but that isn't the reality of it.

If that's the instruction the software needs to execute, it's a reality.

That's like saying anything using capabilities past an 8080 is "poorly
written."

No, it's saying that coding for specific hardware platforms dramatically
impairs portability and is thus usually a poor design decision.

 

[Mxsmanic]

But there's no reason to change since you argue there are no new useful
features and no advantage to faster processors.

No, I argue that software bloat consumes the horsepower of faster
processors. Almost all the improvements in hardware are absorbed by
software bloat. As a result, newer systems perform about the same as
older systems, from an end user's standpoint. If software were not so
bloated, new systems would have performance thousands of times better
than old systems.

Then use it.

It doesn't support my hardware.

 

[Mxsmanic]

**** me, don't want to use Linux then. Out of the box installation of
MDK10/Suse9.3 has over 30 processes, including servers, running after
setup.

Maybe running stuff out of the box isn't a good idea.

 

[Mxsmanic]

myth

Not in the mainframe companies I worked for.