[Coco] Re: gcc-coco revisited

[John E. Malmberg]

Nathan Woods wrote:
> 
> LOL.  In any case, I can definitely say that MESS (and MAME) will never 
> be as fast as a standalone emulator.

True.

> Jeff's emulator is written in 
> highly optimized x86 assembler and MESS is written in portable C.

Not a reason.  A good C optimizing compiler with well written C code can 
produce very good and fast assembler code.

 From what I have seen of Open Source C code, most of the programmers 
have not learned why using the const qualifier is important.  But that 
may just be my sample.

> Portability is also another issue; there is an internal layer that the
> drivers write to which gets blitted to the screen, whereas Jeff's emulator
> can write directly to the PC's video RAM.

There is also the issue is that according to graphics driver programmers 
that I know, the video chip manufacturers only supply their fastest 
drivers for the Microsoft Windows platform, and will not release the 
details needed for other platforms to take advantage of all of the 
hardware accelerations available.  So on the same high end hardware, 
Microsoft Windows will be able to render graphics faster than LINUX.

The COCO hardware having polled I/O also presents a challenge for any 
multi-user time sharing platform.

But the big thing that will probably affecting the speed would be 
graphics, and then any place where the application running in emulation 
is in a polled I/O loop.

-John
wb8tyw at qsl.net
Personal Opinion Only