MIPS: [turbofan] Introduce optional Float64Min and Float64Max machine operators.
authorbalazs.kilvady <balazs.kilvady@imgtec.com>
Thu, 12 Mar 2015 16:43:39 +0000 (09:43 -0700)
committerCommit bot <commit-bot@chromium.org>
Thu, 12 Mar 2015 16:43:51 +0000 (16:43 +0000)
commit323ec9e8bda05bdc0a0f7eb55e910f7c28b2e1bc
treeeb6bb55cb5bd1d532af76809cbf9f041e833619b
parent36df5747bf12d82f206bcc1ca7d743843d8869fc
MIPS: [turbofan] Introduce optional Float64Min and Float64Max machine operators.

Port 99f8d57f3c5d967bc3b10356b22d7ec7be55e89f

Original commit message:
Basically recognize certain x < y ? x : y constructs and turn that into
Float64Min/Float64Max operations, if the target machine supports that.
On x86 we lower to (v)minsd/(v)maxsd.

BUG=

Review URL: https://codereview.chromium.org/1006513002

Cr-Commit-Position: refs/heads/master@{#27168}
src/compiler/mips/instruction-selector-mips.cc
src/compiler/mips64/instruction-selector-mips64.cc