commit | c853eed83eb9670ba671b6f271f612e958f64d34 | [log] [tgz] |
---|---|---|
author | David Srbecky <dsrbecky@google.com> | Thu Sep 13 14:33:22 2018 +0100 |
committer | David Srbecky <dsrbecky@google.com> | Wed Oct 03 14:26:06 2018 +0100 |
tree | 7f1c0b8bd8328194ead74002f5883e54295f44cd | |
parent | ededf08e2f4a5df3401a5d4badb98ff3c8cb8fb9 [diff] |
Use the small thread-local cache for mterp invokes. This speeds up non-quickened interpreter by 2% (measured on golem). Test: ./art/test.py -b -r --interpreter --host Change-Id: I6b00db1b2da7fda4cb0a34beb62d3857ae3d72df