Author: Dieter Buerssner
Date: 12:49:52 12/20/02
Go up one level in this thread
On December 19, 2002 at 17:59:46, Eugene Nalimov wrote: >The following is part of the ANSI/ISO C++ Standard (section 3.10, clause 15): [...] >I don't have C standard nearby, but it contains something similar. Yes. >Unfortunately, lot of real-world C/C++ code violate that part of the standard. Sure. I hope, my code doesn't use such things. My understanding was that the "assume no alias optimization" switches of Compilers will also apply to Standard conform usage of the language. A gave a very small and primitive example, in the answer to Bob. ON CPUs with many registers, sometimes Fortran numerical code is better optimized than C code. The Fortran compiler can keep things in registers, which the C-Compiler can't, when pointers are involved, because it in general will not know, if some writing to an adress will invalidate the register. Of course, in many cases, this can be worked around by some temporary variables. Or, in C99 probably with restricted. My understanding was, that with those optimization switches, you allow the C-compiler using similar optimizations as a Fortran compiler. And, for some code, this would certainly fail (any memmove alike code, for example). Regards, Dieter
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.