Hi there, I'm just starting to look at the mpn division functions , for the straight forward linear functions like mpn_divrem_1 mpn_divexact_1 etc , there is a lot of wasted "bandwidth" on the K8(other cpu's would be similar). For example the function mpn_divexact_1 has a 10c/l thruput which is restricted by the limb/carry dependances. If we didn't have to worry about that we could get it to about 5 or 6 , or we could do some other operation . For example a mpn_divexactadd_1 where we get the add for free , but no one wants a divexactadd.
Any suggestions? these could be usefull 1)two independant divisions 2)a bidirectional divexact Jason --~--~---------~--~----~------------~-------~--~----~ 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 -~----------~----~----~----~------~----~------~--~---
