Changeset: 71decab2094b
Author:jrose
Date: 2011-06-14 20:14 -0700
URL: http://hg.openjdk.java.net/mlvm/mlvm/hotspot/rev/71decab2094b
meth: adjust series for fixes pushed to hotspot-comp
! series
___
mlvm-dev mailing list
mlvm-dev@openjdk
Still crashing! See previous email for hs dumpfile, command-line, and
script, with link to data in the script.
- Charlie
On Tue, Jun 14, 2011 at 10:46 PM, Charles Oliver Nutter
wrote:
> On Tue, Jun 14, 2011 at 10:14 PM, John Rose wrote:
>> This looks like a failure fixed very recently (6/12) by
On Tue, Jun 14, 2011 at 10:14 PM, John Rose wrote:
> This looks like a failure fixed very recently (6/12) by the final version of
> meth-exc-7047697.patch.
>
> Please let me know ASAP if the latest version of that patch *fails* to fix
> your crash!
Doing an updated build from MLVM, where 'hotsp
On Tue, Jun 14, 2011 at 5:21 PM, John Rose wrote:
> Charlie, thanks for being our Chief Test Victim Subject on performance
> issues. I have a question for you (and for the list in general).
>
> Since 6/03 the following patch in the mlvm queue has set GWT to use ricochet
> frames by default:
>
This looks like a failure fixed very recently (6/12) by the final version of
meth-exc-7047697.patch.
Please let me know ASAP if the latest version of that patch *fails* to fix your
crash!
-- John
On Jun 14, 2011, at 8:00 PM, Charles Oliver Nutter wrote:
> I got this crash running a benchmark
Ok, numbers for fib with special paths for +1 -1 +2 and -2...this is
passing a primitive through invokedynamic, assumes we're calling
Fixnum always, and does not type-guard anything...so it's faster than
it ought to be, but it shows what difference the simpler overflow
guard has:
Normal overflow g
I got this crash running a benchmark with indy stuff enabled (only the
default stuff in JRuby master). Disabling use of invokedynamic fixes
it.
Build is based on MLVM a few days old. If it looks like something that
might be fixed, I can pull a more recent trunk OpenJDK.
~/projects/jruby ➔ jruby -
Changeset: 15a2431888d0
Author:jrose
Date: 2011-06-14 19:53 -0700
URL: http://hg.openjdk.java.net/mlvm/mlvm/jdk/rev/15a2431888d0
meth: tweak test
! meth-experiment.patch
___
mlvm-dev mailing list
[email protected]
http://mail.ope
Changeset: 7ad0b98b61b4
Author:jrose
Date: 2011-06-14 19:34 -0700
URL: http://hg.openjdk.java.net/mlvm/mlvm/jdk/rev/7ad0b98b61b4
meth: update for review
! meth-experiment.patch
! meth-rot-7052202.patch
! meth-unittests.patch
___
mlvm-dev
I did take your advice and add paths for +- 1 and 2, but did not see an
improvement in perf. That's not to say there wasn't one, but it may be lost in
the Fixnum object creation...
- Charlie (mobile)
On Jun 14, 2011, at 18:33, Rémi Forax wrote:
> On 06/14/2011 05:02 PM, Charles Oliver Nutter
Changeset: c04ab8f822b1
Author:jrose
Date: 2011-06-14 18:34 -0700
URL: http://hg.openjdk.java.net/mlvm/mlvm/hotspot/rev/c04ab8f822b1
meth: update to hotspot-comp
! meth-rot-7052219.patch < meth-rot-7052202.patch
! meth-sparc-7045514.patch
! series
_
On Jun 14, 2011, at 5:21 PM, Attila Szegedi wrote:
> Thanks for the clarification, John. I'll go with asFixedArity().
Great! BTW, this little method got into the API "by the skin of its teeth".
Here's an interesting related exercise: If I have a variable-arity method
handle (like #Arrays.asL
Thanks for the clarification, John. I'll go with asFixedArity().
Super excited as for the first time in history, Dynalink will get down to 0
failing testcases as OpenJDK bugs making them fail got squashed, yay!
Attila.
On Jun 13, 2011, at 12:51 AM, John Rose wrote:
> On Jun 11, 2011, at 2:18
Changeset: af9f1edb308f
Author:jrose
Date: 2011-06-14 17:04 -0700
URL: http://hg.openjdk.java.net/mlvm/mlvm/jdk/rev/af9f1edb308f
meth: split out 7054590, update various unit tests
! meth-argcount-6983728.patch
+ meth-int-7054590.patch
! meth-rot-7052202.patch
! meth-unittests.patch
On 06/14/2011 05:02 PM, Charles Oliver Nutter wrote:
> Disabled (for perf or incompleteness):
> * Math operator invocations with literal fixnum RHS (incomplete: no guards)
I'm working on an example for the cookbook that
allows integers to overflow to BigInteger and
has special paths when a constan
Charlie, thanks for being our Chief Test Victim Subject on performance issues.
I have a question for you (and for the list in general).
Since 6/03 the following patch in the mlvm queue has set GWT to use ricochet
frames by default:
http://hg.openjdk.java.net/mlvm/mlvm/jdk/file/tip/meth-experi
I need to start doing a TOC for my emails.
** 1. JRuby invokedynamic updates
** 2. Latest performance observations
** 3. Performance ideas going forward
Ok. Here we go.
** 1. JRuby invokedynamic updates
I've been landing a ton of additional call paths and logic surrounding
JRuby's various dispa
Hi,
In the latest push to the github repository I've added the smallest
compilation I can get to exhibit the problem - it contains only one
invokedynamic ( it compiles a method that does 1 + 1 ).
You can run it specifically (and only) using:
$JAVA_HOME/bin/java
-Xbatch -XX:+PrintCompilation -cp
Hi,
Your analysis of the behavior is correct. However, what makes me suspect
this is 292s fault is partly the -Xint part, partly the fact that this
didn't use to blow up - and I haven't changed the compilation strategy
for that specific test for quite some time. It also seems weird that it
works o
19 matches
Mail list logo
