Re: A89: Re: Game Programming -- Pure Assembly vs. TI-GCC?
[Prev][Next][Index][Thread]
Re: A89: Re: Game Programming -- Pure Assembly vs. TI-GCC?
Zeljko Juric wrote:
>
> TIGCC internally first translates C into ASM, then assembles
> it. So, in C may be even faster than program produced in
> assembler by some not so good ASM coder. TIGCC generates
> quite good ASM code but not ideal. My experience is that
> C programs is tipically 15-30% slower that ASM programs
> created by tipical ASM coder. But, I don't believe that
> even the best ASM coder will made a program which will be
> more than twice faster than in C (of course, I talk
> about C programs written by _good_ C coders too).
Hmm, OK... I'll probably try using TI-GCC then & see how it works.
> You ask about DoorsOS libraries etc? Usage of routines
> which use stack for passing parameters is straightforward.
> Routines which use registers for passing parameters needs
> small interface in asm for addapting calling convention
> to C. By the way, there is a special collection of
> routines (written by me) called tigcclib with more than
> 300 routines for usage with tigcc. Current version is 1.1,
> but release 1.5 will be ready in next 10-15 days (which
> will include floating point support, and probably, full
> interface code with Doors libraries).
Sounds kewl! Hey, thanks a bunch for the great info :)
BeaT (Seth Peelle)
beat@drigon.com
http://www.drigon.com/~beat
Follow-Ups:
References: