Rick_the_new_guy

What is the deal with game systems being increased 100% and all.
Also, why even numbers?

NES 8 bit
SNES 16 bit
PS 32
Nintendo 64
PS2 128 bits.

I mean what is the deal? Is that the nature of the beast?
Or is it marketing and making sure each 4 years a new system comes out that is way better.

Thanks champs.

__________________
(PO3) Marcinko_R. (BF2 PR .509) Squad Member
(CPO) Marcinko_R. (BF2 PR .509) Squad Leader
(LCDR) Marcinko_R. (BF2 PR .509) Commander

Squad Member pledge to their SL:http://www.tacticalgamer.com/battlef...ad-leader.html
Squad Leader pledge to their team:http://www.tacticalgamer.com/battlef...r-platoon.html
Commander pledge to their SL:http://www.tacticalgamer.com/battlef...d-leaders.html

Lucky Shot

How stuff works carries your answer far better than I would ever be able to explain it...

Full article...
http://computer.howstuffworks.com/bytes.htm
Just on bits...
http://computer.howstuffworks.com/bytes2.htm

All the best!

Lucky Shot

Kerostasis

That explains what bits are, but doesnt do much to answer why Game Platforms (and most other computer systems, for that matter) tend to increase in power by doubling. Why not a new system thats 50% more powerful, for example?

I know a certain amount of it is due to the fact that computer technology tends to increase at an exponential speed--that is, every two years or so, the stuff we have is roughly twice as fast as the stuff from two years ago. But that doesn't explain why game systems in particular are so attached to the "powers of 2" numbers for their system specs.

__________________

Steeler

Because the chips that power them have advanced along that track. Remember, a bit is on or off. 2 states. And they are bundled into 8-bit bytes. The important thing is not that the chips have repeatedly doubled in complexity over the years, it's that they have continually doubled as multiples of 8.

To make a chip that does not conform to multiples of 8 would require a completely new technology and programming system.

/layman's analysis

__________________

leejo

You have to think in binary to understand why we don't have 60 bit or 71 bit systems.

Everytime you add a character to a binary number, you've increased the set of expressible numbers by the power of two.

xx can express 4 numbers
xxx can express 8 numbers
xxxx can express 16 numbers
xxxxx can express 32 numbers
xxxxxx can express 64 numbers
xxxxxxx can express 128 numbers

etc.

Since these devices "think" in binary, you can't go from 32 to, say, 55 bit without also going to 64 bit.

These numbers are all divisible by 8 because 8 = 2^3, therefore any number 2^(3+x) is also divisible by 2^3. All the numbers above 2^4 are also divisible by 32, and so on.

Talk about a thread to unleash the inner nerd.

Kerostasis

I think you're slightly missing my point.

The upgrade from a 32-value number to a 64-value number is the addition of a single character, with nothing in between, like this:
xxxxx --> xxxxxx

However, the upgrade from a 32-BIT number to a 64-bit number is more like this
xxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxx -->
xxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxx

And there should be something in between that, right?

__________________

Switchcraft

No, Leejo explained it right. Once you make room for anything above 32, you have room all the way to 64. There is no in-between in binary.

Drizzid

It also unleashes the inner thought of "damn I shoulda gone to school"

Too much smarts required for this thread.

__________________

Spyder

Sure, but as previously explained, in order to get to anything in between that, you have to get it to 64 anyway.

0 - 0
1 - 1
10 - 2
11 - 3
100 - 4
101 - 5
110 - 6
111 - 7
1000 - 8


That's an example of how a computer counts in binary. In order to get it to think in bigger terms, you have to add another digit. Each digit is the double of the previous digit. Thus, a single upgrade is a double of potential counting.

You start with one digit, that digit can be 0 or 1, you can count 2 values with it
you add a second digit, you can now count 0-3 (4 values), you have doubled. add a third digit and you can count 0-7 (8 values) and you have doubled again. So on and so forth. Sure you could handcuff yourself by only counting 0-2, but why bother?

It's also why ram chips go 128, 256, 512. 1GB is a round-off, the true value is 1024mb, but it doesn't sound as pretty.

__________________
a.k.a. NinjaPirateAssassin
Celibacy is not Hereditary.
Everybody should believe in something - I believe I'll have another drink.
Happiness is like wetting your pants, everyone can see it but only you can feel the warmth.
Flying is easy, Just throw yourself at the ground and miss!

Kerostasis

....

I'm not talking about doubling the number of VALUES!!! I'm talking about doubling the number of DIGITS! Adding ONE digit doubles your potential values, but they never add just one digit. They always add as many digits as they already had, doubling your digits, which squares your values.

As in:
NES--8 bit system--256 value range
SNES--16 bit system--65536 value range
PS1--32 bit system--4.3 billion value range

Does anyone get what I'm talking about?

__________________

icky

Any true nerd knows that these wannabe nerds don't know what they're talking about. I've programmed things which used both 20-bit and 24-bit addressing, and worked on stuff with 12-bit instructions among other things. I think I've seen a 24-bit data bus as well, but I can't remember for sure. I can guarantee that such things do exist though.

It has everything to do with speed and space. The data bus size needs to double when increased in size if you want to continue to handle all data sizes with maximum efficiency. For simplicity, let's just say that it started at 8-bits, and the only sensible step forward in each case was to double that size.

There are many factors which define the power of a system. These numbers are mostly used for marketing. They grab the biggest "bit" number they can find in the system and advertise it.

__________________
Peace through fear... since 1947!

sordavie

So what does 8 bit refer to when talking about the NEC anyway? Does it refer to the max size of registers, memory addresses, applications, data sizes, or what? I think some people are missing Kerostasis' point. If we're talking about CPU architecture (even then it's not clear what we're talking about), it's not the case that there's nothing between 32 and 64 bit architectures. For example, IBM had a chip based on a 48 bit CPU architecture. Generally, you can have any multiple of 8, since the standard memory unit for computers, a byte, is 8 bits.

You guys are pointing out the largest number one can express given limit on how many characters you can use. That's like saying '99' is the largest number I can express in base 10 using two characters. However, when we're talking about bit size of CPU architecture or data sizes we aren't referring to the largest number one can express given a certain number of charaters. That is, if we say a data size is 64 bit, we are not saying that '64' is the maximum number expressible by the number of characters that data size can hold. That is, we're not saying that that data size is limited to 7 characters, able to express a maximum number of 127, and that 32 bit is one that's limited to 6 characters expressing a max of 63. We're not talking about adding one more digit to binary numbers when going from 32 to 64 bit. We're talking about how many bits something is.

Doh, Icky got here as I was typing.

__________________

leejo

I think we're all basically saying the same thing but coming at it from several different directsions. There is a little confusion between computational power and how physical memory is accessed and loaded into memory. The 20-bit buses, etc., exist, i think, because a 20-bit bus will have processor memory in use to act as a traffic cop while the physical memory is accessed and loaded into processor memory. I could be flat wrong, but I believe there is some mapping that's occurring.

There's plenty about all this mumbu-jumbo online for the searching. Upon further review, I think Steeler was basically right: the 8086 architecture is built on octets.

icky

I'm too lazy to search online for the answer, but I'm pretty sure it is the data bus size.

Ok, I'm not too lazy. I looked it up and found that it uses a 6502 processor with an 8-bit data bus and a 16-bit address bus. I don't think even marketing folk would try to use address bus size to promote a system, since it has almost nothing to do with performance.

The PS2 uses a mixed 128-bit and 64-bit data bus. I believe the main-memory data bus is 64-bit and hangs off of the main 128-bit video bus. I don't know the details too well, and I really am too lazy to dig into that stuff.

For those who are lost at the word "bus", it is literally a bunch of wires which carry bits from one place to another. If you've ever connected an IDE cable to a drive, you were connecting a 32-bit bus (32 wires) and a few control wires.

__________________
Peace through fear... since 1947!

El_Gringo_Grande

I always thought the a larger bus size was a very important factor in performance. Maybey not the data bus (speed of light being all that it is) but the address bus should effect how easy it is to access all the juicy ram that you have.

__________________
I’m not racists, I have republican friends. Radio show host.
- "The essence of tyranny is the denial of complexity". -Jacob Burkhardt
- "A foolish consistency is the hobgoblin of little minds" - Emerson
- "People should not be afraid of it's government, government should be afraid of it's People." - Line from V for Vendetta
- If software were as unreliable as economic theory, there wouldn't be a plane made of anything other than paper that could get off the ground. Jim Fawcette