<!DOCTYPE html><html><head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body>
<p><br>
</p>
<div class="moz-cite-prefix">On 7/8/24 9:04 AM, Charles Oliver
Nutter wrote:<br>
</div>
<blockquote type="cite" cite="mid:CAE-f1xRn_Lh0=zRp5=fO0ek=xoV2iqNzyagTAD9nG43JuXaz=Q@mail.gmail.com">
<div dir="ltr">The patched build does indeed work, and cuts
JRuby's base startup time by 50%! Bravo!
<div><br>
</div>
<div>$ time jruby -e 1<br>
jruby -e 1 3.53s user 0.14s system 276% cpu 1.326 total<br>
$ time jruby -J-XX:CacheDataStore=jruby.cds -e 1<br>
jruby -J-XX:CacheDataStore=jruby.cds -e 1 2.38s user 0.19s
system 385% cpu 0.667 total<br>
</div>
<div><br>
</div>
<div>This is a very promising start and the largest single
improvement we have seen. Comparison with our previous
champion, C1:</div>
<div><br>
</div>
<div>$ time jruby --dev -e 1<br>
jruby --dev -e 1 1.22s user 0.09s system 142% cpu 0.922 total<br>
</div>
<div><br>
</div>
<div>Interestingly, when I try to use the previously trained CDS
dump (via 100.times { org.jruby.Ruby.newInstance }) with a
command line that forces C1 (--dev flag passes
-XX:TieredStopAtLevel=1) I get a different crash: <a href="https://urldefense.com/v3/__https://gist.github.com/headius/d3138efa52eed9c678414c8cfe08af27__;!!ACWV5N9M2RV99hQ!O1UAeB7nD1zEm6bMPw7oDGNPEmvkkh6jK5YVmZJVKQP6Ugi5tSURjYzIpMvg0XmXhF6oaTom1wRsn5A$" moz-do-not-send="true">https://gist.github.com/headius/d3138efa52eed9c678414c8cfe08af27</a></div>
<div><br>
</div>
</div>
</blockquote>
<p>I filed <a class="moz-txt-link-freetext" href="https://bugs.openjdk.org/browse/JDK-8336033">https://bugs.openjdk.org/browse/JDK-8336033</a> . It looks
like a simple book keeping issue in the compiler code.<br>
</p>
<p><br>
</p>
<blockquote type="cite" cite="mid:CAE-f1xRn_Lh0=zRp5=fO0ek=xoV2iqNzyagTAD9nG43JuXaz=Q@mail.gmail.com">
<div dir="ltr">
<div>I'm wondering what else I can do to try to train the CDS
dump better, so that a larger command (jruby -S gem list) sees
more improvement (this loads a bunch more Ruby code into the
interpreter):</div>
<div><br>
</div>
<div>$ time jruby -S gem list > /dev/null<br>
jruby -S gem list > /dev/null 6.41s user 0.26s system 236%
cpu 2.821 total<br>
$ time jruby -J-XX:CacheDataStore=jruby.cds -S gem list >
/dev/null<br>
jruby -J-XX:CacheDataStore=jruby.cds -S gem list >
/dev/null 5.84s user 0.32s system 294% cpu 2.089 total<br>
</div>
<div><br>
</div>
</div>
</blockquote>
<br>
Does this load a lot of dynamically generated classes? The current
set of Leyden optimizations are mostly for classes loaded in the
built-in loaders (boot/platform/app).<br>
<p>I wonder how much of the time is spend in one-time activity. If
you run the above inside a loop (2x, 4x, 8x, 16x, etc), how does
it scale?</p>
<p>I think if we see a big drop between 1x vs 2x, then perhaps the
current set of Leyden optimizations can help (as long as most of
the time is spent in the built-in loaders).</p>
<p>In any case, dynamic language runtimes like jruby probably have
very different start-up characteristics than, say, your typical
"micro service" Java app. More analysis is needed.</p>
<p>Thanks</p>
<p>- Ioi<br>
</p>
<p><br>
</p>
<blockquote type="cite" cite="mid:CAE-f1xRn_Lh0=zRp5=fO0ek=xoV2iqNzyagTAD9nG43JuXaz=Q@mail.gmail.com">
<div dir="ltr">
<div>- Charlie</div>
</div>
<br>
<div class="gmail_quote">
<div dir="ltr" class="gmail_attr">On Thu, Jul 4, 2024 at
11:50 PM <<a href="mailto:ioi.lam@oracle.com" moz-do-not-send="true" class="moz-txt-link-freetext">ioi.lam@oracle.com</a>>
wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<div> <font face="monospace">Hi Charlie,<br>
<br>
Thanks for the bug report.<br>
<br>
It turns out some Leyden optimizations can't handle the
module options specified by jruby, such as these:<br>
<br>
--add-opens=java.base/java.nio.channels=org.jruby.dist<br>
--module-path=/xxxx/jruby-9.4.8.0/lib/jruby.jar<br>
<br>
I filed <a href="https://bugs.openjdk.org/browse/JDK-8335735" target="_blank" moz-do-not-send="true" class="moz-txt-link-freetext">https://bugs.openjdk.org/browse/JDK-8335735</a><br>
<br>
I'll work on a proper fix. Meanwhile, I have a work-around
in this branch<br>
<br>
<a href="https://urldefense.com/v3/__https://github.com/iklam/jdk/tree/work-around-8335735-jruby-crash-due-to-lack-of-module-support__;!!ACWV5N9M2RV99hQ!O1UAeB7nD1zEm6bMPw7oDGNPEmvkkh6jK5YVmZJVKQP6Ugi5tSURjYzIpMvg0XmXhF6oaTomY2MJITU$" target="_blank" moz-do-not-send="true">https://github.com/iklam/jdk/tree/work-around-8335735-jruby-crash-due-to-lack-of-module-support</a><br>
<br>
If you build a JDK from this source branch, you can see
the following improvements:<br>
<br>
# [1] No optimizations<br>
$ time env JAVA_HOME=$MYJAVA
~/Downloads/ruby/jruby-9.4.8.0/bin/jruby \<br>
-e '10.times { org.jruby.Ruby.newInstance }'<br>
real 0m3.215s<br>
user 0m14.868s<br>
sys 0m0.373s<br>
<br>
<br>
# [2] Leyden<br>
$ time env JAVA_HOME=$TBP0
~/Downloads/ruby/jruby-9.4.8.0/bin/jruby \<br>
-J-XX:CacheDataStore=jruby.cds \<br>
-e '10.times { org.jruby.Ruby.newInstance }'<br>
real 0m1.880s<br>
user 0m9.001s<br>
sys 0m0.268s<br>
<br>
<br>
</font><font face="monospace"># [3] Compare with CDS in the
JDK mainline<br>
$ env JAVA_HOME=$MYJAVA
~/Downloads/ruby/jruby-9.4.8.0/bin/jruby \<br>
-J-XX:DumpLoadedClassList=jruby.classlist \<br>
-e '10.times { org.jruby.Ruby.newInstance }'<br>
$ $MYJAVA/bin/java -Xshare:dump \<br>
--module-path=/home/iklam/Downloads/ruby/jruby-9.4.8.0/lib/jruby.jar
\<br>
-XX:SharedClassListFile=jruby.classlist \<br>
-XX:SharedArchiveFile=jruby.jsa -Xlog:cds<br>
$ time env JAVA_HOME=$MYJAVA
~/Downloads/ruby/jruby-9.4.8.0/bin/jruby \<br>
-J-XX:SharedArchiveFile=jruby.jsa \<br>
-e '10.times { org.jruby.Ruby.newInstance }'<br>
real 0m2.628s<br>
user 0m13.007s<br>
sys 0m0.428s<br>
<br>
Please try it out and let us know if you run into other
problems.<br>
<br>
Thanks<br>
<br>
- Ioi</font><font face="monospace"><span style="color:rgb(23,43,77);font-size:14px;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-align:left;text-indent:0px;text-transform:none;word-spacing:0px;white-space:normal;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"><br>
</span></font><font face="monospace"><span style="color:rgb(23,43,77);font-size:14px;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-align:left;text-indent:0px;text-transform:none;word-spacing:0px;white-space:normal;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline"></span></font><br>
<font face="monospace"><br>
</font>
<div><font face="monospace">On 7/4/24 1:06 PM, Charles
Oliver Nutter wrote:<br>
</font></div>
<blockquote type="cite">
<div dir="ltr"><font face="monospace">Hello friends! Long
time lurker, first time poster.</font>
<div><font face="monospace"><br>
</font></div>
<div><font face="monospace">Like others I was excited to
hear that the first EA of Leyden had dropped. Sadly,
I have another crash to report.</font></div>
<div><font face="monospace"><br>
</font></div>
<div><font face="monospace">I see there are two other
crashes reported, but mine appears different
(ClassPrelinker::is_indy_resolution_deterministic):</font></div>
<div><font face="monospace"><br>
</font></div>
<div><font face="monospace"><a href="https://urldefense.com/v3/__https://gist.github.com/headius/79c6460ed55c1d80c82e1c0209897e24__;!!ACWV5N9M2RV99hQ!O1UAeB7nD1zEm6bMPw7oDGNPEmvkkh6jK5YVmZJVKQP6Ugi5tSURjYzIpMvg0XmXhF6oaTomO4igb9c$" target="_blank" moz-do-not-send="true">https://gist.github.com/headius/79c6460ed55c1d80c82e1c0209897e24</a><br>
</font></div>
<div><font face="monospace"><br>
</font></div>
<div><font face="monospace">Perhaps unsurprisingly, the
additional flags provided by Vladimir Kozlov (<span style="color:rgb(0,0,0)">-XX:+UnlockDiagnosticVMOptions
-XX:-ReduceAllocationMerges) did not appear to
change the result.</span></font></div>
<div><font face="monospace"><span style="color:rgb(0,0,0)"><br>
</span></font></div>
<div><font face="monospace"><span style="color:rgb(0,0,0)">This is with JRuby's
minimally invokedynamic-based mode, which has
given us the shortest startup time in the past
(second only to disabling tiers 2-4 and staying in
C1).</span></font></div>
<div><font face="monospace"><span style="color:rgb(0,0,0)"><br>
</span></font></div>
<div><font face="monospace" color="#000000">I am on
MacOS AArch64 Sonoma 14.5, testing against JRuby
master, but reproduction should be as easy as
downloading the JRuby binary tarball, unpacking, and
running bin/jruby with the command line above.</font></div>
<div><font face="monospace" color="#000000"><br>
</font></div>
<div><font face="monospace"><a href="https://urldefense.com/v3/__https://www.jruby.org/download__;!!ACWV5N9M2RV99hQ!O1UAeB7nD1zEm6bMPw7oDGNPEmvkkh6jK5YVmZJVKQP6Ugi5tSURjYzIpMvg0XmXhF6oaTom4BSxN20$" target="_blank" moz-do-not-send="true">https://www.jruby.org/download</a><font color="#000000"><br>
</font></font></div>
<div><font face="monospace"><span style="color:rgb(0,0,0)"><br>
</span></font></div>
<div><font face="monospace"><span style="color:rgb(0,0,0)">I am eager to work with
Leyden folks to investigate issues, and I am
planning to be at JVMLS this year to discuss
collaborating more!</span></font></div>
<div><font face="monospace"><span style="color:rgb(0,0,0)"><br>
</span></font></div>
<div><font face="monospace"><span style="color:rgb(0,0,0)">- Charlie</span></font></div>
</div>
</blockquote>
</div>
</blockquote>
</div>
</blockquote>
</body>
</html>