Yeah the Win64 function call ABI is used and but the longs are 64 bits not 32.
This probably shouldn't affect things too much for the reasons you state. I don't have any special insight except that maybe it is not detecting the processor correctly on Cygwin64, or there's some bottleneck which has been sped up in GMP and not MPIR. The configuration information you requested should help us get to the bottom of that one (as will the performance improvements I have been making...). Bill. On 27 February 2014 15:13, Jean-Pierre Flori <jpfl...@gmail.com> wrote: > > > On Thursday, February 27, 2014 2:31:44 PM UTC+1, Bill Hart wrote: >> >> Hi fwjmath, >> >> >> Cygwin uses a different ABI to msys. So I expect the times to be >> different. >> > > I think they both use the Win64 function call ABI (oh yeah Microsoft was > nice enough not to use the AMD64 ABI...). > There is no real doc about that but you can find some clue on the Cygwin > mailing list. > In particular fewer args are passed through register. > > The main difference is that the Win64 ABI is LLP64 (so longs are only 4 > bytes), mingw64 is LLP64 and cygwin64 is LP64 (as in the AMD64 ABI). > Note that MPIR and GMP should be smart enough to define the limb type > which is used most of the time to be 8 bytes long on mingw64 and on cygwin64 > -- You received this message because you are subscribed to the Google Groups "mpir-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to mpir-devel+unsubscr...@googlegroups.com. To post to this group, send email to mpir-devel@googlegroups.com. Visit this group at http://groups.google.com/group/mpir-devel. For more options, visit https://groups.google.com/groups/opt_out.
