Mail Archives: opendos/1997/05/14/20:29:20
On Wed, 14 May 1997 14:04:12 -0400 (EDT) Pierre Phaneuf
<pp AT 55-174 DOT hy DOT cgocable DOT ca> writes:
>On Wed, 14 May 1997, Leath Muller wrote:
>
>32 bit isn't double the speed of a 16 bit program! The *main*
>advantage
>that you can access larger areas of memory at once. For example, you
>can
>use a linear frame buffer, instead of using 4 segments and bank
>switching
>to access the video card. This is a *bit* faster since you don't have
A "bit"? Hell yeah!
(_Especially_ considering most cards with LFB's run faster using those
than a0000)
>to
>bank switch, not because of using 32-bit instructions. Of course,
>using
>32-bit instruction will help a bit too. Under NT, an ordinary user
>cannot
>set a priority to real-time BTW.
Of course, there's also BUS width to consider - an RM program can still
use the 32bit string instructions, so you end up with a wierd hybrid of
16bit memory and 32bit op's, so the whole discussion is rather pointless.
I'd like to ask: What parts of the program are 32bit?
>The only two advantages of DirectX is that it gives any program access
>to
>all the features of hardware acceleration (which is interesting) and
>overcome the GDI, which is a GUI-only concern, since there's no GDI to
>overcome in DOS.
I'd like to see a 'Windows Extender':
We can sell it as a speed improver for Windows apps, it'll give direct
access to everything! (Basically just wipes system memory and hands the
CPU over to the program :)
Of course, programs won't directly benefit from it until they're coded
for it, but that's the same as MMX, right?
(Anybody else frequent x86.org? Check out the article about Photoshop
MMX)
...Chambers
- Raw text -