Large increase in startup time from JDK8 to JDK9

[Claes Redestad]

Hi,

jigsaw does come with an increase in the number of bytecodes executed during startup, since module descriptors for all system modules are created eagerly. This also means there is a significant difference between a JDK and a JRE image (fewer modules). 

There is also a huge difference between the exploded jdk and images: if you build a JDK yourself make sure you do startup investigation on one of the images under build/.../images/jdk|jre and not the exploded jdk under build/.../jdk - jlink does some magic already to speed up startup!

Last but not least: Does your hello world load code from a jar file? In that case things could be made worse by a recent regression:

http://mail.openjdk.java.net/pipermail/core-libs-dev/2016-July/042538.html 

Thanks!

/Claes

  

Andrew Haley <aph at redhat.com> skrev: (25 juli 2016 12:32:11 CEST)
>I've been noticing a significant increase in startup time, and decided
>to investigate.
>
>With JDK8 -XX:+TraceBytecodes on HelloWorld I see:
>
>760620 bytecodes executed in 13.5s (0.056MHz)
>[BytecodeCounter::counter_value = 760620]
>
>And with JDK9 I see:
>
>4916036 bytecodes executed in 120.0s (0.041MHz)
>[BytecodeCounter::counter_value = 4916036]
>
>I'm guessing this is because of Jigsaw processing, but I haven't
>investigated any further.  I would have thought that such a severe
>degradation was a release blocker.
>
>Does anyone here know about this?  Has anyone investigated?
>
>Andrew.

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.