Mailing-List: contact reviews-help@impala.incubator.apache.org; run by ezmlm
Precedence: bulk
Message-Id: <201706080016.v580GDLW001002@ip-10-146-233-104.ec2.internal>
Date: Thu, 8 Jun 2017 00:16:13 +0000
From: "Impala Public Jenkins (Code Review)" <gerrit@cloudera.org>
To: Michael Ho <kwho@cloudera.com>, impala-cr@cloudera.com,
        reviews@impala.incubator.apache.org
Subject: =?UTF-8?Q?=5BImpala-ASF-CR=5D_IMPALA-4164=3A_Avoid_overly_aggressive_inlining_in_LLVM_IR=0A?=
In-Reply-To: <gerrit.1495309518000.I2d87ae8d222b415587e7320cb9072e4a8d6615ce@gerrit.cloudera.org>
References: <gerrit.1495309518000.I2d87ae8d222b415587e7320cb9072e4a8d6615ce@gerrit.cloudera.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Content-Disposition: inline
User-Agent: Gerrit/2.12.7
archived-at: Thu, 08 Jun 2017 00:16:17 -0000

Impala Public Jenkins has submitted this change and it was merged.

Change subject: IMPALA-4164: Avoid overly aggressive inlining in LLVM IR
......................................................................


IMPALA-4164: Avoid overly aggressive inlining in LLVM IR

When generating IR functions during codegen, we used to
always tag the functions with the "AlwaysInline" attribute.
That potentially leads to excessive inlining, causing very
long optimization / compilation time with marginal performance
benefit at runtime. One of the reasons for doing it was that
the "target-cpu" and "target-features" attributes were
missing in the generated IR functions so the LLVM inliner
considers them incompatible with the cross-compiled functions.
As a result, the inliner will not inline the generated IR
functions into cross-compiled functions and vice versa unless
the "AlwaysInline" attributes exist.

This change fixes the problem above by setting the "target-cpu"
and "target-features" attributes of all IR functions to match
that of of the host's CPUs so both generated IR functions and
cross-compiled functions will have the same values for those
attributes. With these attributes set, we now rely on the
inliner of LLVM to determine whether a function is worth being
inlined. With this change, the codegen time of a query with very
long predicate went from 15s to 4s and the overall runtime went
from 19s to 8s.

Change-Id: I2d87ae8d222b415587e7320cb9072e4a8d6615ce
Reviewed-on: http://gerrit.cloudera.org:8080/6941
Reviewed-by: Michael Ho <kwho@cloudera.com>
Tested-by: Impala Public Jenkins
---
M be/src/codegen/llvm-codegen.cc
M be/src/codegen/llvm-codegen.h
A testdata/workloads/targeted-perf/queries/primitive_long_predicate.test
3 files changed, 214 insertions(+), 12 deletions(-)

Approvals:
  Impala Public Jenkins: Verified
  Michael Ho: Looks good to me, approved


-- 
To view, visit http://gerrit.cloudera.org:8080/6941
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-MessageType: merged
Gerrit-Change-Id: I2d87ae8d222b415587e7320cb9072e4a8d6615ce
Gerrit-PatchSet: 11
Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-Owner: Michael Ho <kwho@cloudera.com>
Gerrit-Reviewer: Dan Hecht <dhecht@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins
Gerrit-Reviewer: Michael Ho <kwho@cloudera.com>
Gerrit-Reviewer: Tim Armstrong <tarmstrong@cloudera.com>