On Wed, Dec 14, 2016 at 2:48 PM, mmarco <mma...@unizar.es> wrote: > If you provide several binaries, I think it would be wise to advertise the > lowest common denominator as the "standard" one, so it works for every > user.
Indeed--it would be the default, most prominently advertised one. More specialized builds, if they exist at all, would be listed as well but in terms such that if you don't know what it is you probably don't need/want it. That said, I'd be curious what some common benchmarks might be to test performance of something like, say, NTL, with our without extended SIMD support. > El miércoles, 14 de diciembre de 2016, 12:59:13 (UTC+1), Erik Bray escribió: >> >> On Wed, Dec 14, 2016 at 12:47 PM, Erik Bray <erik....@gmail.com> wrote: >> > On Wed, Dec 14, 2016 at 12:42 PM, Dima Pasechnik <dim...@gmail.com> >> > wrote: >> >> Hi Erik, >> >> >> >> there are bugs in 7.4 (which is the version your Windows version is >> >> based >> >> on, right?) >> >> that break SAGE_FAT_BINARY. >> >> See https://trac.sagemath.org/ticket/21850 and >> >> https://trac.sagemath.org/ticket/21849 >> >> and https://trac.sagemath.org/ticket/21772 >> >> >> >> Did you backport these? >> > >> > No, I did not. Thanks for pointing these out--at least one of these >> > is the likely culprit then. For example I note in the .debug_info >> > section that givaro was built with AVX support, which Peter's CPU does >> > not have. I'll backport these fixes and make a new build. >> >> Perhaps a thing in the future to do will be to provide a "fat" >> installer, with different builds of some packages (and only those >> packages where it applies) for different CPU architectures. The >> easiest thing to do would be to just provide entire separate >> installers for different CPUs, and we might start with that. But >> really the installer should automatically select the correct binaries >> for the target system and users shouldn't have to think about it. >> >> But first it would be best to see what the demand is like before >> putting a lot of effort into it. >> >> >> On Wednesday, December 14, 2016 at 11:04:01 AM UTC, Erik Bray wrote: >> >>> >> >>> On Wed, Dec 14, 2016 at 11:33 AM, Peter Luschny <peter....@gmail.com> >> >>> wrote: >> >>> >> Could you tell me what kind of CPU you have? And remind me, was >> >>> >> this >> >>> >> running in a VM (it shouldn't matter but I'm curious)? >> >>> > >> >>> > >> >>> > No, no, it ran on bare iron. >> >>> >> >>> Okay, so there are at least a few instructions my CPU has that yours >> >>> doesn't, such as the SSSE3 instruction set (as opposed to SSE3). >> >>> Still hard to narrow it down much though. >> >> >> >> -- >> >> You received this message because you are subscribed to the Google >> >> Groups >> >> "sage-devel" group. >> >> To unsubscribe from this group and stop receiving emails from it, send >> >> an >> >> email to sage-devel+...@googlegroups.com. >> >> To post to this group, send email to sage-...@googlegroups.com. >> >> Visit this group at https://groups.google.com/group/sage-devel. >> >> For more options, visit https://groups.google.com/d/optout. > > -- > You received this message because you are subscribed to the Google Groups > "sage-devel" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to sage-devel+unsubscr...@googlegroups.com. > To post to this group, send email to sage-devel@googlegroups.com. > Visit this group at https://groups.google.com/group/sage-devel. > For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at https://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
