PC 4GB RAM limit

[Phil Weldon]

Seeing that the first computers I worked on had less main memory that
current Intel and AMD CPUs have L1 cache... I have a different take on
bloat. What's wrong with bloat? A system that handles 4 GBytes well today
costs less that an month's lease on 4 Kbytes memory back when an OS ran 4
KBytes. What else would the use of 4 GBytes for 999 out of 1000 users?

 

[David Maynard]

Seeing that the first computers I worked on had less main memory that
current Intel and AMD CPUs have L1 cache... I have a different take on
bloat.

Hehe. Yeah, me too.

 

[David Maynard]

Judging from the pace at which Windows bloats, it may be sooner than
you think.





4 GB of RAM is obscene. I can remember when a 20 MB HD was considered
large.

I can not only remember when a 12 inch pizza platter hard drive was 1.2 meg
I've got two drives and a dozen packs in the garage.

 

[Andrew Smallshaw]

If "it" is being able to run an EM64T processor in 64 bit mode then the
board supports "it."

If "it" is having an address space larger than 4 gig then the board does
not support "it."

_All_ modern day processors support more than 4Gb - this was introduced in the
386 - although admittedly you had to abandon the 'flat' memory model. For the
386 the absolute limit was 64Gb if memory serves. Therefore the processor is
irrelevant - it's more dependant on the chipset. I never saw a 386 board that
supported so much memory but I remember them being availiable from about the
Pentium I era onwards.

 

[lyon_wonder]

And I thought this reserved for memory hardware crap died out with the
real mode 8086 and it's archaic 640k and reserved for hardware 384k.

 

[Al Dykes]

_All_ modern day processors support more than 4Gb - this was introduced in the
386 - although admittedly you had to abandon the 'flat' memory model. For the
386 the absolute limit was 64Gb if memory serves. Therefore the processor is
irrelevant - it's more dependant on the chipset. I never saw a 386 board that
supported so much memory but I remember them being availiable from about the
Pentium I era onwards.

The issue for applications is maximum proccess size, which is 4GB or
less per process in x86 architecture. For software to use more
requires PAE which does context switching to access multiple
segments. This has a performace cost and requires PAE circiutry on the
mobo.

 

[David Maynard]

_All_ modern day processors support more than 4Gb

That isn't in question.

- this was introduced in the

386

PAE was introduced on the Pentium Pro.

- although admittedly you had to abandon the 'flat' memory model. For the
386 the absolute limit was 64Gb if memory serves. Therefore the processor is
irrelevant

No, the processor isn't irrelevant. x86 architecture uses PAE to get past
4GB whereas 64 bit processors natively address way more than even PAE.
They're entirely different mechanisms with different capabilities.
performance, and limits.

And, of course, there are non x86 processors, although they clearly don't
plug into an x86 architecture motherboard.

- it's more dependant on the chipset.

That is precisely the 'support' we are discussing. Or rather, not only the
chipset but also how much of the chipset's capability is implemented in a
particular motherboard's design.

I never saw a 386 board that
supported so much memory but I remember them being availiable from about the
Pentium I era onwards.

PAE was introduced on the Pentium Pro, however, we're talking about 64 bit
addressing. At least in theory we are. How 64 bit addressing gets
'upgraded' into a board originally designed as a 32 bit system isn't
entirely clear.

 

[David Maynard]

And I thought this reserved for memory hardware crap died out with the
real mode 8086 and it's archaic 640k and reserved for hardware 384k.

Since it never died out, or even gasped a teensy bit, and has been there in
every single x86 system ever made, and expanded by PCI and further expanded
by PCI express, it hardly qualifies as 'archaic'.

In fact, rather than 'die out' it has grow, lived long, and prospered.

 

[Bob]

What's wrong with bloat?

Good question.

Nature seems to like bloat.

Bloat must be existential.


--

Million Mom March For Gun Confiscation
http://home.houston.rr.com/rkba/mmm.html

A liberal is a person who is so open minded
that their brains have fallen out.

 

[Bob]

But bloat rules. It may be existential.

I can remember when a 20 MB HD was considered large.

I can not only remember when a 12 inch pizza platter hard drive was 1.2 meg
I've got two drives and a dozen packs in the garage.

That goes back a bit. I can remember the Physics Dept doing particle
experiments with drum storage. The memory took a good sized room.

If bloat rules, then how come all advances in computers have come
about by smallness?

That's another existential question.


--

Million Mom March For Gun Confiscation
http://home.houston.rr.com/rkba/mmm.html

A liberal is a person who is so open minded
that their brains have fallen out.

 

[Phil Weldon]

Software expands to fill availble space.

But bloat rules. It may be existential.

 

[Mxsmanic]

What's wrong with bloat?

It erases hardware gains. Today's PCs don't run much faster than PCs
twenty years ago, in terms of response time for users; all the
additional hardware horsepower has been absorbed by bloat.

 

[Mxsmanic]

Since it never died out, or even gasped a teensy bit, and has been there in
every single x86 system ever made, and expanded by PCI and further expanded
by PCI express, it hardly qualifies as 'archaic'.

It's still archaic; it's a really stupid way to do things. And it just
amazes me how much membory is squandered on PCI Express; indeed, almost
all the memory used is used for PCI Express.

In fact, rather than 'die out' it has grow, lived long, and prospered.

Festered would be a better word.

 

[David Maynard]

David Maynard writes:




It's still archaic; it's a really stupid way to do things.

That it's almost universally popular is defacto proof it's not just "a
really stupid way to do things."

And it just
amazes me how much membory is squandered on PCI Express; indeed, almost
all the memory used is used for PCI Express.

Maybe if you put more effort into understanding why it's done that way it
wouldn't be such a mystery.

Festered would be a better word.

There are a lot more considerations to computer design than making it
trivially transparent for the unskilled to pop in an indeterminate amount
of RAM without being confronted by the technical.

 

[David Maynard]

Phil Weldon writes:




It erases hardware gains. Today's PCs don't run much faster than PCs
twenty years ago, in terms of response time for users; all the
additional hardware horsepower has been absorbed by bloat.

And all the advancements in automobiles over the past 100 years have been
'wasted' because one still can't go faster than 35 MPH in a 35 MPH speed zone.

Point is, one can make any irrational claim by picking the appropriately
inappropriate 'measurement' criteria.

And since you think "all the additional hardware horsepower has been
absorbed by bloat" then why don't you run DOS on a 386 and do your video
editing with it?

 

[David Maynard]

Software expands to fill availble space.

It certainly can't expand into unavailable space

 

[David Maynard]

But bloat rules. It may be existential.




That goes back a bit.

I did pull them from a dumpster, along with the mini-computer they went
with, so they were old when I got 'em, but many were still in service. One
worked but I had to fix the other.

I can remember the Physics Dept doing particle
experiments with drum storage. The memory took a good sized room.

Oh yeah, drums. The ones with heads all over the place were impressive, and
expensive as all get out.

The strangest 'disk drive' I ran across was a real old one, still in
service, that was a huge 30 inch, or so, diameter aluminum disc mounted
vertically. Capacity was something like 250K.

If bloat rules, then how come all advances in computers have come
about by smallness?

Miniaturization enables more bloat in the same space

 

[kony]

And all the advancements in automobiles over the past 100 years have been
'wasted' because one still can't go faster than 35 MPH in a 35 MPH speed zone.

Point is, one can make any irrational claim by picking the appropriately
inappropriate 'measurement' criteria.

It'd be a poor point then, because it's a quite rational
claim that hardware bloat is ridiculous. I'm not arguing
that "all additional hardware horsepower has been absorbed"
though, rather that the developers seem to have little to no
concern about the escalating storage requirements nor memory
to run applications. Just because memory is far cheaper
than it used to be, that doesn't mean I find it acceptible
for a developer to take a view that they don't have to
follow good practices.

A better argument relating to automobiles is, what do I care
if i haul around 200 lbs. of bricks in my truck everywhere
even though I have no need for them, since my engine has the
extra power and efficiency over one made 40 years ago.
While it's a shame the car dealer couldn't be bothered to
take the bricks out of the trunk when it was sold to me, I
can still drive around therefore all is right in the world.

And since you think "all the additional hardware horsepower has been
absorbed by bloat" then why don't you run DOS on a 386 and do your video
editing with it?

That may be a good point, or may not.
Suppose the video editing app had become more and more
bloated onto the point of being less efficient than it
should be. Suppose it's 10% slower as a result. 10% could
be considered the price different between two different
models of CPU, are you happy to pay more for the faster CPU
so the developer can profit more by not making the effort to
code better? Passing the buck is ok as long as it doesn't
stop here.

 

[kony]

It certainly can't expand into unavailable space

Pagefile?

 

[David Maynard]

Pagefile?

That's available.