Sammy, I am beginning to come around to your way of thinking

On Jun 4, 7:28 am, Mxsmanic <mxsma....DeleteThis@gmail.com> wrote:
> I've already explained the danger of encoding information in a linear space
> that controls an exponential space.

You've babbled about cutting up the address space into pieces that are
not the same size and that are sometimes too large for their specific
use. That's all.

> > When they chunk up an address space they do it for a reason and they
> > fully realize the implications.
>
> They do it to satisfy their egos, and they never have a clue about the
> implications, which leads to endless problems down the line. Engineers can't
> predict the future better than anyone else, but they think they can.

Ah I see the big bad world is just full of egotistical fools....all
except you.

In building a new protocol or address space you do have to predict how
it's going to be used. I think they did remarkably well with IP
addressing. That you can point out flaws 30 years later just means it
outgrew its original design. What would you propose? That engineers
NOT try to predict how a product is going to be used and cater for it?

> They were doing the same thing twenty years ago, and they'll be doing the same
> thing twenty years from now. They were doing it before I was born.

You mean that people will continue to make mistakes? Yes. That the
same mistakes will continually be made? No. Perhaps the same tradeoffs
will need to be considered. Perhaps the political process will
interfeer with the engineering process and you won't get an ideal
protocol. That does happen. It doesn't mean computer engineers are
stupid and don't understand the difference between exponential and
linear, nor does it mean that they are incapable of working out the
size of an address subspace allocation.

If you're so good at so many damned things, why don't I see your name
in the computer journals? Why aren't you doing things better? Why
aren't you rich and famous.
 >> Stay informed about: Sammy, I am beginning to come around to your way of thinking 

On Jun 1, 4:26 am, Mxsmanic <mxsma....DeleteThis@gmail.com> wrote:
> > You really think you'd win that argument?
> I'm pretty good at winning arguments.

Yes amazing how good a person can be at winning arguments when they're
happy to ignore the facts, make gross generalizations and be so
imprecise about how they word things. Thing is you're only winning
arguments with yourself that way.
 >> Stay informed about: Sammy, I am beginning to come around to your way of thinking 

On Jun 4, 7:26 am, Mxsmanic <mxsma....TakeThisOut@gmail.com> wrote:
> Emotion can cloud your judgement, so beware.

Yes Mr Spock. Live long and prosper.

You've certainly got his people skills down pat.

> It really doesn't that much intelligence. Yes, the increases have been close
> to linear. PCs still cost about the same as they did years ago, and they
> still work about as slowly, and they still allow about the same amount of
> productivity. The software bloats to erase every hardware gain made.

Again with your misuse of the word linear. Linear has a precise
meaning. the equation needs to fit f(x)=mx+c where m and c are
constants.

What you're trying to say is that "things haven't got that much
better" looking at the speed of your computer and conveniently
forgetting that a lot of today's software does much more than
yesterday's. Just look at the detail in 3D gaming between the oiginal
Wolfenstein and today's 3D marvels, or at a flight sim. Sure
framerates don't skyrocket but that's because each evolution of the
software offers a lot more detail (computationally speaking).

Have a look at any set of benchmarks and you'll see that isn't
true.Would you like me to present you with graphs that show these
trends? There's more to computing than Microsoft Office and the start
up time for your operating system.

Go tell that to anyone that does commercial 3D rendering and
animation. Or tell it to someone analysing large data sets.

> I have done better in the past. I'm one of those people who still likes
> assembly language.

Please do tell me what products you've worked on. If you're so
wonderful there must be dozens of excellent products with your name
attached. Put up or shut up. Compiling a few scripts out of an
assembly manual doesn't make you an expert.

> > Ever heard of Moore's law?
>
> Yes. It doesn't actually apply to computers per se, but to component
> densities.

Well if you're such an expert, tell me is that linear growth or
exponential? You claim computers are progressing linearly.

> Essentially all the architectures used today already existed in the 1960s.
> Microcomputer technicians think they are being inventive, but it has all been
> done before, and in some cases, better.

Ah wonderful so you're completely dismissing new optimization
techniques, and changes to architecture. It must be true because you
say so.

> No, because you're afraid that I'll sound smarter than you if you don't reply
> somehow. Unfortunately, personal attacks do not circumvent that possibility.

It doesn't matter how I reply to you, you still come up with weird and
whacky unsubstantiated cryptic rubbish and attack me whether I respond
or not. Your misuse of terminology, and your belief that your thinking
is superior to everyone else's isn't of my making.

> You're creating your own situation. Many of your conversations end in the
> same way, based on my observations thereof here. So it isn't just me.

Yes I'm creating my own situation in the limited sense that I choose
to argue with fools (plural).

Have a nice day, troll. Should be an easy accomplishment for someone
who believes they know better than everyone else.
 >> Stay informed about: Sammy, I am beginning to come around to your way of thinking 

Canuck wrote:
> Oi, leave my thread alone you hijackers!
>


Forget it Canuck, it's in the death spiral......
 >> Stay informed about: Sammy, I am beginning to come around to your way of thinking 

On Jun 4, 10:54 am, Mxsmanic <mxsma... RemoveThis @gmail.com> wrote:
> Sammy writes:
> > You don't sacrifice it. You allocate a special purpose use to it.
>
> No, you sacrifice it.

No you don't.

> Suppose you have an eight-bit field. An 8-bit field will hold 256 addresses.
> If you assign them serially, you will have a capacity of 256

You're talking about compartmentalizing the address space again.
Assigning memory means assigning a value within memory.

>. Now, if you
> divide the address field into two parts, say "vendor" and "product," of four
> bits each, you've divided the field by a factor of two, but you've reduced the
> address space by a factor of 16.

Except that you don't divide address space in a special purpose way
like this unless you have to.

It's a terrible example. I'd much rather talk about TCP/IP packet
headers or something more concrete so as to pin down your definitions
but I'll try.

If you only need a single byte to hold the vendor code, your only
reason for allocate more than that is for efficiency. That doesn't
happen for vendors and product codes. Depending on the number of
vendors or products you have, and on the architecture you might
allocate anywhere between 2 and 4 bytes to hold each field. YOU DON'T
allocate address spaces for something like this.

I've already pointed out that the limitations for virtual memory on 32
bit windows are 2Gb data and 4Gb total The address space here is cut
into two. It makes good sense to divide up your code (instructions)
from data (despite the limitation against self-modifying code - that's
a good limitation). Clearly the "mistake" of breaking up the address
space into many little pieces along byte boundaries wasn't repeated
here.

You wouldn't allocate 4 bits of the address space to vendor unless you
needed the 16 bits to describe a possible 16 vendors, in which case
the address space isn't wasted. So why do we do it for a network...or
rather why was it done for TCP/IP? When this was developed computers
were very slow, but they're faster at spliting things up on the byte
boundaries and using bitwise operations to manipulate them
individually. So to get the hardware to perform at a reasonable speed
routing packets they put things on the boundaries and that did result
in segmentation. Today the hardware is exponentially faster but the
demands on it have also skyrocketted (at least for top end equipment,
the likes of which you might see on a backbone). The efficiency you
gain in speed must be traded off against the way you use space (and
which you claim is wasted). In other words it's the price you pay for
not having TCP packets take many seconds to be switched through the
network.

> Unless you have exactly sixteen vendors with
> exactly sixteen products each, you will waste a tremendous amount of the
> address space. By dividing the field in two, you encode information in the
> address, and a linear increase in encoding produces an exponential loss in
> address space.

Your analogy is so bad I do wonder if you've ever programmed.

> > Yes you know much better and are a true visionary.
> That's an exaggeration, but I am surprised by how clueless so many engineers
> are (or perhaps they are too stubborn to consider their own mistakes).

Okay then Mr visionary. Prey tell how on earth would you allocate
memory for addressing, particularly if as in the networking example
you are harping on, efficiency is of the essence. If you suggest
holding memory addresses "linearly" ie one byte per address I'm going
to have a good time explaining why this is a bad idea.
 >> Stay informed about: Sammy, I am beginning to come around to your way of thinking 

Sammy writes:

> What a complete and utter load of bullshit. Your grandiose posturing
> just pisses me off.

Emotion can cloud your judgement, so beware.

> Yet you think you're so intelligent that you dismiss all of this and
> suggest that our increases have been "linear".

It really doesn't that much intelligence. Yes, the increases have been close
to linear. PCs still cost about the same as they did years ago, and they
still work about as slowly, and they still allow about the same amount of
productivity. The software bloats to erase every hardware gain made.

> It's true that there
> have been tradeoffs and that some software is bloated and/or badly
> designed, but you could certainly do no better.

I have done better in the past. I'm one of those people who still likes
assembly language.

> Ever heard of Moore's law?

Yes. It doesn't actually apply to computers per se, but to component
densities.

> It's still done that way for architectures developed in the 60s yes.
> These architectures complete with their inefficiencies are the stable
> base without which the Internet could never have taken shape. Not
> every address space has been subdivided in the way you're discussing.

Essentially all the architectures used today already existed in the 1960s.
Microcomputer technicians think they are being inventive, but it has all been
done before, and in some cases, better.

> Yes because you keep saying things in an attempt to sound smart that
> get more and more preposterous.

No, because you're afraid that I'll sound smarter than you if you don't reply
somehow. Unfortunately, personal attacks do not circumvent that possibility.

> In other
> words you're trying to create a situation where I feel I have to
> respond despite not wanting to waste the time so that you can attack
> that.

You're creating your own situation. Many of your conversations end in the
same way, based on my observations thereof here. So it isn't just me.
 >> Stay informed about: Sammy, I am beginning to come around to your way of thinking 

Sammy writes:

> Yes Mx, you know much much better than all these people and should be
> put in charge of the world

Not all of them, but clearly better than some of them.

> The same gibberish wording for the simple idea that they tend divide
> an address into segments based on a single bit in the address, which
> means that the subdivisions of the address can be large. That doesn't
> decrease the amount of memory available unless you choose to put these
> rules around it. Do you even realize that the smallest subdivision
> based on the highest bit in an address only cuts it in half? Half of
> the 64 address space is still very very much bigger than the whole 32
> bit address space.

I've already explained the danger of encoding information in a linear space
that controls an exponential space.

> When they chunk up an address space they do it for a reason and they
> fully realize the implications.

They do it to satisfy their egos, and they never have a clue about the
implications, which leads to endless problems down the line. Engineers can't
predict the future better than anyone else, but they think they can.

> You are amazing in that you seem to
> think you can predict 20 years into the future what people will be
> doing in the computer industry and should be put in charge of these
> stupid engineers that make the same mistakes so that your wisdom might
> prevent them.

They were doing the same thing twenty years ago, and they'll be doing the same
thing twenty years from now. They were doing it before I was born.
 >> Stay informed about: Sammy, I am beginning to come around to your way of thinking 

Sammy writes:

> ...and some of these have retired.

So?

> Can I borrow your crystal ball please?

I don't have a crystal ball, but I know that history tends to repeat itself.

> I'd be very surprised if this were true.

Most engineers end up surprised.
 >> Stay informed about: Sammy, I am beginning to come around to your way of thinking 

Sammy writes:

> You don't sacrifice it. You allocate a special purpose use to it.

No, you sacrifice it.

Suppose you have an eight-bit field. An 8-bit field will hold 256 addresses.
If you assign them serially, you will have a capacity of 256. Now, if you
divide the address field into two parts, say "vendor" and "product," of four
bits each, you've divided the field by a factor of two, but you've reduced the
address space by a factor of 16. Unless you have exactly sixteen vendors with
exactly sixteen products each, you will waste a tremendous amount of the
address space. By dividing the field in two, you encode information in the
address, and a linear increase in encoding produces an exponential loss in
address space.

> Yes you know much better and are a true visionary.

That's an exaggeration, but I am surprised by how clueless so many engineers
are (or perhaps they are too stubborn to consider their own mistakes).
 >> Stay informed about: Sammy, I am beginning to come around to your way of thinking 

Sammy writes:

> In building a new protocol or address space you do have to predict how
> it's going to be used.

Not entirely, and it's foolish to try to predict everything in advance.

> I think they did remarkably well with IP addressing.

They did so-so, and they made some very serious early mistakes.

> That you can point out flaws 30 years later just means it
> outgrew its original design.

If its original design were open-ended, it would not be outgrown.

> What would you propose?

See above.

> That engineers NOT try to predict how a product is going to be
> used and cater for it?

In part, yes.

> You mean that people will continue to make mistakes?

Yes. The same mistakes, in many cases.

> If you're so good at so many damned things, why don't I see your name
> in the computer journals?

You haven't read them all.

> Why aren't you doing things better? Why aren't you rich and famous.

Solving grand problems usually doesn't provide a salary. Most people just
want the current fires to be extinguished, and won't pay for anything else.
 >> Stay informed about: Sammy, I am beginning to come around to your way of thinking 

Sammy writes:

> So contrary to your claims their address space creations outlived
> them, at least in a professional sense.

Just because they've retired doesn't mean they've disappeared from the
profession.

> History repeats? How cliche.

The truth hurts.

> Today's computers aren't just simple extensions
> of yesterday's.

Actually they look pretty much the same. PCs of today use the same
architecture they used nearly 30 years ago; and that in turn was based on
architecture that is around half a century old already.

> Nor is the software. It has all gotten quite complex
> when compared to the computers of the 1980s I grew up with. The
> changes made have been revolutionary as well as evolutionary.

It has bloated, but it has not substantially improved.

> Another sensationalist claim that "most" experts in a field make
> trivial mistakes and are surprised by them when you have clarity.

Most non-experts, that is. There are very few experts, and not enough to go
around.
 >> Stay informed about: Sammy, I am beginning to come around to your way of thinking 

Sammy writes:

> Yes amazing how good a person can be at winning arguments when they're
> happy to ignore the facts, make gross generalizations and be so
> imprecise about how they word things. Thing is you're only winning
> arguments with yourself that way.

I'm not arguing, I'm playing.
 >> Stay informed about: Sammy, I am beginning to come around to your way of thinking 

Nice Bob. Your entire aim is to antagonize me. Not say that something
I've said is wrong or even crazy. Just to specifically antagonize.
Real nice. Do you wonder why I call you a child or a moron?

You've fallen in the same class as Mx. A troll that doesn't know fact
from fiction and is happy to bignote himself while putting others.
Specific people you don't agree with (not a class of people). 11+
pages of ranting and threads started for no other reason than abuse
shows that up really nicely. Grow up, or get back on your meds, or
whatever you need to do to stop being such a nasty man. I guarantee if
you're acting like this offline you've dug yourself into one hell of a
hole.
 >> Stay informed about: Sammy, I am beginning to come around to your way of thinking