8343597: C2 SuperWord: RelaxedMath for faster float reductions #21895
eme64
wants to merge
7 commits into
openjdk:master
from
eme64:JDK-8343597-RelaxedMath-reductions
JDK-8343597
c12bad8e
openjdk
changed the title
first part of intrinsification
dd34d40c
intrinsify with regular F/D add/mul
234b81a0
add RelaxedMathAddF - but does not yet vectorize...
defb7775
vectorizes, but still with strict reduction
f9e42afa
no_strict_order generated, moved out of loop, but turned into strict …
f066bcfa
also non-strict when moved out of loop
deaadf88
Reviewers
No reviews
Assignees
No one assigned
Labels
core-libs
Milestone
No milestone
Login to write a write a comment.
Login via GitHub