Hello Uzi,

Wednesday, August 04, 1999, 2:14:48 AM, you wrote:

>>   32-bit colour is generally 8 bits each of red/green/blue (that's 24 --
>> that's about as fine a gradation of colours as the human eye can
>> distinguish), but pixels are aligned on 32-bit boundaries rather than
>> 24-bit boundaries -- this greatly simplifies I/O between CPU and video
>> buffer, improving performance of the video subsystem.

UP> Now, according to what you say, on standard systems, it will not
UP> improve quality, but will improve speed ! (I interpret performance as
UP> speed) - almost the opposite from the naive interpretation.

This is true. Here, the idea that computers best work with power of 2
is applied. Computing sizes of 3-byte areas requires some
multiplication; 4-byte areas can be computed by shifting left by 2 - a much
faster procedure. (Even better, good compilers will always replace
2-powered multiplication with shifting!)

Another factor is that pixels are always aligned on word boundary.
This should fasten the access time.

And the last factor I can think of that 386-s and up are 32-bit
CPU's and accordingly work best with 32-bit data; and this is exactly what
4-byte word is.

+=-.
| Max Timchenko [MaxVT]
| [log in to unmask], [log in to unmask]
| (ICQ) 238-6792
|
| Freelance website and graphics designer
| Max Webdesign at http://maxwd.hypermart.net
+=-.

                Curious about the people moderating your
                   messages? Visit our staff web site:
                     http://nospin.com/pc/staff.html