2013/6/4 dw <limegreenso...@yahoo.com>: > >>> I think option #3 is that one I prefer too. Just one thing about >>> SSE-instruction. For 64-bit we can assume that SSE has to be present >>> in any case. Just for 32-bit we should check in headers for the >>> __SSE__ macro, and in the intrinsic-function (none-inline) we should >>> default to none-SSE version. >> __SSE__ should always be defined with x86_64 so just checking for >> __SSE__ might be simpler and will give the fastest version. >> >> Thus the same definition could be used for both inline and intrinsic. > > Per MS docs, __faststorefence is only defined for x64. While I assume > it would work, the current winnt.h doesn't define it for x86 and I > wasn't planning to add it.
That's a fair point. > Being able to just assume that __SSE__ is defined certainly makes the > code cleaner. The reason I didn't was that it is possible that someone > could be compiling with -mno-sse for some reason. Ignore this? I think we should ignore that somebody uses -mno-sse for x64. First this would be clearly an ABI-breakage, due SSE is mandatory in calling-convention. And secondly, on x64 CPUs base SSE feature-sets we can assume always. > There is a similar question for MemoryBarrier(). The mfence instruction > is SSE2. Check for __SSE2__ or not? Well, not sure here. AFAIR is SSE2 present for all more modern x64 CPUs, but well, here this check might make sense. > dw Kai ------------------------------------------------------------------------------ How ServiceNow helps IT people transform IT departments: 1. A cloud service to automate IT design, transition and operations 2. Dashboards that offer high-level views of enterprise services 3. A single system of record for all IT processes http://p.sf.net/sfu/servicenow-d2d-j _______________________________________________ Mingw-w64-public mailing list Mingw-w64-public@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
