commit | 4a395f0e19b3e3f5790ef95fd7b72694cd4296b6 | [log] [tgz] |
---|---|---|
author | Shih-wei Liao <sliao@google.com> | Thu Jun 30 22:15:27 2011 -0700 |
committer | Shih-wei Liao <sliao@google.com> | Thu Jun 30 22:15:27 2011 -0700 |
tree | ffd733293209d887b6b6f70230db44fe2401757d | |
parent | aac57ca72129d8442d5a4c58e28d3eb56aed018b [diff] |
Enable the use of cache. With this check-in, end-to-end LLVM milestone achieved. For NDK samples, Plasma, running on Nexus-S, 800x480 screen: LLVM Time Run: 12.60ms per frame, Avg: 12.57 GCC Time Run: 50.89ms per frame, Avg: 52.34 LLVM can achieve 80 fps, while GCC yields 19 fps. GCC is visibly much slower on my Nexus-S. Nexst step: Refactor the code and clean up. Change-Id: I4a9d552a5482840b4270442702cb5734934d4658