64bit computers

Floating point numbers already are 64 bits wide if you use doubles.

The most important change will be the wider pointers, which make it possible to address a lot more than just 4GB RAM. Sweeney from Epic already said that the content creation tools for the next generation Unreal engine will need so much memory that they will probably require a 64-bit computer.

cu,
Prefect

quote:

---

Original post by Prefect
Floating point numbers already are 64 bits wide if you use doubles.

---

Yes now floating point = 32bit, double = 64bit, but I meant, will with 64bit computers the floats become 64 and the doubles 128?

quote:

---

Original post by Prefect
The most important change will be the wider pointers, which make it possible to address a lot more than just 4GB RAM. Sweeney from Epic already said that the content creation tools for the next generation Unreal engine will need so much memory that they will probably require a 64-bit computer.

cu,
Prefect

---

Cool, I want to see the next generation Unreal engine!

One day we will have a "power glove" that lets us move our pointer through a room based OS. No longer will we store stuff on a "Desktop" it will be in a locker. We will move the pointer on a third plane.

quote:

---

Original post by smiley4
One day we will have a "power glove" that lets us move our pointer through a room based OS. No longer will we store stuff on a "Desktop" it will be in a locker. We will move the pointer on a third plane.

---

And like in the movies, you will hack into someone else's computer by fighting a security guard that blocks the entrance to their file system...

But seriously, about floats. I wonder if they will become 64 bits or stay at 32. It's the same thing that happened when computers went from 16 to 32, and 'int' was 16 and became 32. But, 'short' was 16 and stayed 16. Is 'float' considered the equivalent of short or int?

I guess it would make sense that float stays at 32, for compatibility and efficiency reasons. Otherwise there'd be no way to use 32bit floats anymore...

[edited by - foofightr on May 1, 2003 11:00:55 AM]

The smart thing is to let the developer choose, ie change declarative syntax in languages for which variable size matters (C, C++ and other sufficiently low-level languages) to include the size:

int:32 a;
float:64 f; 

On platforms without the native size required, it could be eliminated (making transparent the use of big integer facilities and thus making such languages more intrinsically amenable to scientific computation). You notice that the example syntax I use above is a perversion of the C/C++ bitfield syntax in structs.

I don''t know if it''s standard, but at least VC.NET has got types like __int32, __int64, and so on. Maybe this is/will be for floats too?

---

For crying out loud, she has fishes coming out of her head on either side. How can you find this hot?!

maybe they''ll make a new data type to accomodate the 128 bit precision, like ''quad'' (for quadruple the precision of a float, or something)

[edited by - fadilthrejk on March 19, 2002 at 4:53 AM]