From: Hans-Bernhard Broeker Newsgroups: comp.os.msdos.djgpp Subject: Re: DJGPP Compiler with output for x86-64 Date: 19 Feb 2001 14:05:05 GMT Organization: Aachen University of Technology (RWTH) Lines: 32 Message-ID: <96r96h$q76$1@nets3.rz.RWTH-Aachen.DE> References: NNTP-Posting-Host: acp3bf.physik.rwth-aachen.de X-Trace: nets3.rz.RWTH-Aachen.DE 982591505 26854 137.226.32.75 (19 Feb 2001 14:05:05 GMT) X-Complaints-To: abuse AT rwth-aachen DOT de NNTP-Posting-Date: 19 Feb 2001 14:05:05 GMT Originator: broeker@ To: djgpp AT delorie DOT com DJ-Gateway: from newsgroup comp.os.msdos.djgpp Reply-To: djgpp AT delorie DOT com Eli Zaretskii wrote: > On Sun, 18 Feb 2001, G=FCnter Reusing wrote: >> Does anyone know a DJGPP compiler (DOS/Windows) for output of x86-64 or >> 3D-now code? > What's x86-64? He probably means AMD's latest CPU development, also known as 'Sledgehammer': a new design with x86-like structure of opcodes and registers, but everything that used to be 32 bits wide is now 64bits wide. A repetition of the step up from 286 to 386, sort of. > And under what circumstances would you like GCC to emit 3D-now > instructions? I second the implicit doubt you express here, Eli. As far as I know, MMX and 3DNow! are really almost completely unusable for compiled languages. Very different datatypes and access method from the CPU and traditional FPU, and you have to save the whole FPU state before and after, or be certain that the FPU contains no valuable information, to be able to use these extensions. It's basically an all-or-nothing decision: either FPU or MMX. Mixing them is hard, if not impossible, for the compiler. But C assumes the presence of the FPU. This effectively limits the use of these technolgies to hand-coded, optionally inlined, assembly routines. -- Hans-Bernhard Broeker (broeker AT physik DOT rwth-aachen DOT de) Even if all the snow were burnt, ashes would remain.