Subzero: Fix a performance regression in the register allocator.

Subzero: Fix a performance regression in the register allocator.

The register allocator does not need to be considering rematerializable variables at all.  When it does, there tends to be a big performance cost because the live range of a rematerializable variable tends to be large, leading to lots of expensive overlap computations against register allocation candidates.

BUG= none
R=jpp@chromium.org, sehr@chromium.org

Committed: https://gerrit.chromium.org/gerrit/gitweb?p=native_client/pnacl-subzero.git;a=commit;h=4c5c57159b68c909bd4edef27c7ac09795a7f33b

[Jim Stichnoth, 2015-11-17 01:05:35]

Description was changed from

==========
Subzero: Fix a performance regression in the register allocator.

BUG= none
==========

to

==========
Subzero: Fix a performance regression in the register allocator.

The register allocator does not need to be considering rematerializable
variables at all.  When it does, there tends to be a big performance cost
because 

BUG= none
==========

[Jim Stichnoth, 2015-11-17 01:07:56]

Description was changed from

==========
Subzero: Fix a performance regression in the register allocator.

The register allocator does not need to be considering rematerializable
variables at all.  When it does, there tends to be a big performance cost
because 

BUG= none
==========

to

==========
Subzero: Fix a performance regression in the register allocator.

The register allocator does not need to be considering rematerializable
variables at all.  When it does, there tends to be a big performance cost
because the live range of a rematerializable variable tends to be large, leading
to lots of expensive overlap computations against register allocation
candidates.

BUG= none
==========

[Jim Stichnoth, 2015-11-17 01:07:56]

stichnot@chromium.org changed reviewers:
+ jpp@chromium.org, kschimpf@google.com, sehr@chromium.org

[Jim Stichnoth, 2015-11-17 01:09:26]

This brings the MINIMAL=1 translation time for pnacl-llc.pexe from 6s back down
to <4s.

[John, 2015-11-17 01:10:45]

lgtm

[sehr, 2015-11-17 01:11:53]

lgtm

[Jim Stichnoth, 2015-11-17 01:17:51]

Committed patchset #1 (id:1) manually as
4c5c57159b68c909bd4edef27c7ac09795a7f33b (tree was closed).