High overhead exists when InterpreterEntryTrampoline building the interpreter frame
Reported by
manjian2...@gmail.com,
Jan 11 2018
|
|||
Issue descriptionSteps to reproduce the problem: 1. Follow the instructions in the "Chrome for Android" section: https://www.chromium.org/developers/creating-v8-profiling-timeline-plots 2. Open h5.m.taobao.com, or i.feng.com 3. Follow the instruction in the "Plotting" section 4. Builtins function InterpreterEntryTrampoline is always listed as the top builtins function. What is the expected behavior? Remove the overhead as every unoptimized function call will enter this builtin function. What went wrong? Push a undefined value to the stack to build the interpreter frame inside a loop one at a time. The optimized patch is in the attached file. Did this work before? N/A Chrome version: 63.0.3239.132 Channel: stable OS Version: 6.0 Flash Version:
,
Jan 12 2018
,
Jan 15 2018
Please close this issue. The NEON instructions most likely bring heavier overhead. The overall time consumed by JS increases.
,
Jan 18 2018
|
|||
►
Sign in to add a comment |
|||
Comment 1 by tkent@chromium.org
, Jan 11 2018