Hi all, we have backported a set of AArch64 costs model related revisions in the linaro/gcc-4_9-branch at r211843. The backported revisions are:
210493 : [AArch64 costs 1/18] Refactor aarch64_address_costs. 210494 : [AArch64 costs 2/18] Add cost tables for Cortex-A57 210495 : [AArch64 costs 3/18] Wrap aarch64_rtx_costs to dump verbose output 210496 : [AArch64 costs 4/18] Better estimate cost of building a constant 210497 : [AArch64 costs 5/18] Factor out common MULT cases 210498 : [AArch64 costs 6/18] Set default costs and handle vector modes. 210499 : [AArch64 costs 7/18] Improve SET cost. 210500 : [AArch64 costs 8/18] Cost memory accesses using address costs 210501 : [AArch64 costs 9/18] Better cost logical operations 210502 : [AArch64 costs 10/18] Improve costs for sign/zero extend operations 210503 : [AArch64 costs 11/18] Improve costs for rotate and shift operations. 210504 : [AArch64 costs 12/18] Improve costs for sign/zero extracts 210505 : [AArch64 costs 13/18] Improve costs for div/mod 210506 : [AArch64 costs 14/18] Cost comparisons, flag setting operators and IF_THEN_ELSE 210507 : [AArch64 costs 15/18] Cost more Floating point RTX. 210508 : [AArch64 costs 16/18] Cost TRUNCATE 210509 : [AArch64 costs 17/18] Cost for SYMBOL_REF, HIGH and LO_SUM 210510 : [AArch64 costs 18/18] Dump a message if we are unable to cost an insn. 210512 : [AArch64 costs] Fixup to costing of FNMUL 211205 : aarch64_if_then_else_costs refactor 211206 : aarch64_if_then_else_costs Thanks, Yvan
