On 7/6/2011 14:45, Torbjorn Granlund wrote: > JonY writes: > > GMP is misusing the --host option for detecting CPU optimization levels > > I don't think it is doing that. It used -O2 for all CPUs, I think. >
I mean for ASM codes, adding -O2 implicitly isn't an issue. > (Why not just use --with-arch=XXX?). Can anybody explain why is it doing > something like that? > > I am not sure I follow you. GMP detects the CPU, by means of > config.guess, and then uses the detected architecture for selecting some > special code, and architecture specific compiler flags. > This is a horrible hack, --host is for selecting cross compiler toolchains, not CPU ASM optimization levels, use another option for that. How do you specify the toolchain to use if the value is supposed to come from config.guess? Am I supposed to invent a new triplet and build an entirely new toolchain *just* to satisfy GMP's config.guess abuse? > I had to do this along with --host=none-none-none ABI=longlong to make > gmp work without ASM. > > I think GMP 5 will not require such workarounds (but I am not sure what > "do this" above refers to). > "Do this" refers to using --host=none-none-none and setting CC/CXX/AR/RANLIB manually to disable ASM when configure is too smart for its own good.
0xED74C077.asc
Description: application/pgp-keys
signature.asc
Description: OpenPGP digital signature
------------------------------------------------------------------------------ All of the data generated in your IT infrastructure is seriously valuable. Why? It contains a definitive record of application performance, security threats, fraudulent activity, and more. Splunk takes this data and makes sense of it. IT sense. And common sense. http://p.sf.net/sfu/splunk-d2d-c2
_______________________________________________ Mingw-w64-public mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
