On Fri, Feb 20, 2009 at 3:18 PM, <ja...@njkfrudils.plus.com> wrote: > > On Friday 20 February 2009 16:42:19 Jason Martin wrote: >> > On Friday 20 February 2009 14:12:19 ja...@njkfrudils.plus.com wrote: >> > >> > What happened to core-2 mul_basecase and sqr_basecase ? , no-wonder >> > core-2 benchmarks are crap >> >> There aren't any :-) I was just using Gaudry's code for those routines. >> > > When you said the addmul is nearly optimal , I just assumed that the basecase > was as well.. > >> Should be able to use your amd64 code for those even on core2. The >> only big difference would be in inc/dec instructions which we can >> easily replace with "add" or even "lea" depending on if we need to >> preserve the carry bit for the loop. >> > > If that is the main/only change then I'll do it tomorrow and merge into > trunk , I dont want to spend much time on it now , I want to get on with the > new division code (get the low hanging fruit first)
Yes, that should be the only substantive change. The loops in basecase are short enough that the different pipelining for K10 vs. Core 2 shouldn't have a major impact. The inc and dec are what matter because the core 2 doesn't handle them well. Thanks for making the change! And, as has already been mentioned, YOUR CODE ROCKS!!! --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "mpir-devel" group. To post to this group, send email to mpir-devel@googlegroups.com To unsubscribe from this group, send email to mpir-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/mpir-devel?hl=en -~----------~----~----~----~------~----~------~--~---
