Is 32bit relevant anymore?

[James Ladd]

Is doing 32bit assembler relevant anymore?
I mean, all my computers are 64bit so why would I code in 32bits.
Just wondering?

[Magnum]

Of course they are relevant.

I have used a 64 bit for a short time which had Vista.

It wasn't one lick faster for what I used it for.

Maybe the chip can be divided up with half being used with a cable to another computer.

Just kidding.  :lol

[dedndave]

i think 32 bits is optimal with current technology
that doesn't mean it won't change in the next 5 or 10 years, though

[James Ladd]

I'm thinking my question/statement may not have been clear, for which I apologise.

Let me try another way.

I have 64bit chips and I want to write in assembler. Is there any point doing anything other than
64bit assembler, since I and possibly most people who might run the app when it is done will be
on 64bit chips?

[dedndave]

there are 2 drawbacks to writing 64-bit code

1) it is clumsey in terms of stack alignment - also cumbersome in terms of data size

2) the market hasn't caught up with the manufacturers
you may limit the use of your code, excluding the millions of 32-bit machines out there
conversely, 32-bit code will run on 64-bit machines

[Gunther]

Is doing 32bit assembler relevant anymore?

I think 32 bit code will be relevant for some years. But what we can see with our very eyes is the change to 64 bit, step by step. That process will probably be similar like the change from 16 bit to 32 bit several years ago.

In the new 64 bit world are some drawbacks, which dedndave pointed out. But there are advantages, too:

• the register width of 64 bit
• a lot more registers
• parameter passing to procedures in HLL via registers

But another serious drawback is that: The ABI (Application Binary Interface) for Windows and Unix are very different. It's always another world. That makes it hard for everyone to write code for both platforms.

Gunther

[dedndave]

regretably, we are "in the middle" of the change-over
it seems, for now, the best approach is to write 2 versions of everything (32 and 64)
if you can find a speed advantage in 64-bit code, then you probably want to support it
i would think that graphics is the main place to find an advantage - i don't really know

at the end of the day, it won't hurt to learn it and be familiar with it
especially if you are young   :bg
i am too old to start learning something new - lol
i'd have to forget something first - to make room for it

[Gunther]

it seems, for now, the best approach is to write 2 versions of everything (32 and 64)

Right. That's a good approach.

i would think that graphics is the main place to find an advantage

Hard number crunching is probably another area.

i'd have to forget something first - to make room for it

But take care not to forget important points.  :wink

Gunther

[Antariy]

Hard number crunching is probably another area.

Yes, very hard number crunching with using of very precise SSE code  :green2



Alex

[Antariy]

since I and possibly most people who might run the app when it is done will be on 64bit chips?

Market of not old notebooks with PII or PIII CPUs have not small niche. Millions of 5 years "Old" AMD and Intel chips. Chips of other vendors (except, for example Nano)...



Alex

[hutch--]

 :bg

James,

Just wait until they produce a 64 bit version of JAVA and you should be in business. How is that dog doing ?

[jj2007]

I think 32 bit code will be relevant for some years. But what we can see with our very eyes is the change to 64 bit, step by step. That process will probably be similar like the change from 16 bit to 32 bit several years ago.

Going from 16 to 32 bits was a big improvement. Going from 32 to 64 bits is a much smaller step, and has some important drawbacks, too.
So I would guess that the transition to 64 bits will take longer (remember 16-bit code is still around, especially in business environments), and will never be fully completed.

[frktons]

Going from 16 to 32 bits was a big improvement. Going from 32 to 64 bits is a much smaller step,
and has some important drawbacks, too.
So I would guess that the transition to 64 bits will take longer (remember 16-bit code is still around,
especially in business environments), and will never be fully completed.

I tend to agree with these points of view, except for the last one:

and will never be fully completed

as Royal Majesty's MI6 Secret Agent says:

...Never say never...  :lol

Frank

[redskull]

The world will only move to 64-bit the day they stop selling 32-bit compatible chips and operating systems.  You have to drag accountants, kicking and screaming, from any software that still "works".  As jj pointed out, people still use 16-bit software simply because Windows still runs it, even twenty years later.  But, on the other hand, if you are writing assembly language then you are probably not too worried about that part of the market.

-r