Account Options

  1. Sign in
The old Google Groups will be going away soon, but your browser is incompatible with the new version.
Google Groups Home
« Groups Home
v8-dev
Home
+ new post
About this group
Join this group
Issue 2349 in v8: eliminate HTransitionElementsKind dominated by HCheckMaps with a different map set
Options
There are currently too many topics in this group that display first. To make this topic appear first, remove this option from another topic.
There was an error processing your request. Please try again.
flag
   3 messages - Collapse all  -  Translate all to Translated (View all originals)
The group you are posting to is a Usenet group. Messages posted to this group will make your email address visible to anyone on the Internet.
Your reply message has not been sent.
Your post was successful
 
From:
To:
Cc:
Followup To:
Add Cc | Add Followup-to | Edit Subject
Subject:
Validation:
For verification purposes please type the characters you see in the picture below or the numbers you hear by clicking the accessibility icon. Listen and type the numbers you hear
 
codesite-nore...@google.com  
View profile  
 More options Sep 29 2012, 6:36 am
From: codesite-nore...@google.com
Date: Sat, 29 Sep 2012 10:36:09 +0000
Local: Sat, Sep 29 2012 6:36 am
Subject: Issue 2349 in v8: eliminate HTransitionElementsKind dominated by HCheckMaps with a different map set
Print | Individual message | Show original | Report this message | Find messages by this author
Status: Accepted
Owner: ----
CC: da...@chromium.org,  verwa...@chromium.org
Labels: Type-Bug Priority-Medium

New issue 2349 by vego...@chromium.org: eliminate HTransitionElementsKind  
dominated by HCheckMaps with a different map set
http://code.google.com/p/v8/issues/detail?id=2349

I was looking at register allocation in kraken-audio-beat-detection and  
noticed that in the innermost loop there is a HTransitionElementsKind that  
is dominated by a map check that actually allows to prove that that  
transition can never happen.

I tried writing a simple pass to eliminate transition and it seems to  
improve perf by ~15%.

Attachments:
        transitions-eliminator.patch  3.6 KB


 
You must Sign in before you can post messages.
To post a message you must first join this group.
Please update your nickname on the subscription settings page before posting.
You do not have the permission required to post.
codesite-nore...@google.com  
View profile  
 More options Oct 1 2012, 10:14 am
From: codesite-nore...@google.com
Date: Mon, 01 Oct 2012 14:14:17 +0000
Local: Mon, Oct 1 2012 10:14 am
Subject: Re: Issue 2349 in v8: eliminate HTransitionElementsKind dominated by HCheckMaps with a different map set
Print | Individual message | Show original | Report this message | Find messages by this author
Updates:
        Owner: verwa...@chromium.org

Comment #1 on issue 2349 by da...@chromium.org: eliminate  
HTransitionElementsKind dominated by HCheckMaps with a different map set
http://code.google.com/p/v8/issues/detail?id=2349

(No comment was entered for this change.)


 
You must Sign in before you can post messages.
To post a message you must first join this group.
Please update your nickname on the subscription settings page before posting.
You do not have the permission required to post.
codesite-nore...@google.com  
View profile  
 More options Oct 1 2012, 12:26 pm
From: codesite-nore...@google.com
Date: Mon, 01 Oct 2012 16:26:48 +0000
Local: Mon, Oct 1 2012 12:26 pm
Subject: Re: Issue 2349 in v8: eliminate HTransitionElementsKind dominated by HCheckMaps with a different map set
Print | Individual message | Show original | Report this message | Find messages by this author

Comment #2 on issue 2349 by verwa...@chromium.org: eliminate  
HTransitionElementsKind dominated by HCheckMaps with a different map set
http://code.google.com/p/v8/issues/detail?id=2349

Unfortunately with my elements transition CL in, it's reduced to ~1%; which  
is overshadowed by the 4% regression on JsonParseFinancial:

http://70.32.156.88:8080/Comparison#targetA%3Dv8%3BmachineTypeA%3Dlin...


 
You must Sign in before you can post messages.
To post a message you must first join this group.
Please update your nickname on the subscription settings page before posting.
You do not have the permission required to post.
End of messages
« Back to Discussions « Newer topic     Older topic »

Create a group - Google Groups - Google Home - Terms of Service - Privacy Policy
©2013 Google