jdk
8343597: C2 SuperWord: RelaxedMath for faster float reductions
#21895
Open

Commits
  • JDK-8343597
    eme64 committed 48 days ago
  • first part of intrinsification
    eme64 committed 48 days ago
  • intrinsify with regular F/D add/mul
    eme64 committed 40 days ago
  • add RelaxedMathAddF - but does not yet vectorize...
    eme64 committed 40 days ago
  • vectorizes, but still with strict reduction
    eme64 committed 40 days ago
  • no_strict_order generated, moved out of loop, but turned into strict version
    eme64 committed 39 days ago
  • also non-strict when moved out of loop
    eme64 committed 39 days ago
Loading