Closed Bug 427481 Opened 18 years ago Closed 16 years ago

Sunspider test suite 6% slower in Camino compared to Minefield builds

Categories

(Camino Graveyard :: General, defect)

x86
macOS
defect
Not set
normal

Tracking

(Not tracked)

RESOLVED INCOMPLETE

People

(Reporter: phiw2, Unassigned)

References

()

Details

(Keywords: perf)

Attachments

(3 files, 1 obsolete file)

The sunspider javascript benchmark suite is consistently slower when using Camino. Results for one run (numbers are in the same range over various runs, various builds): Minefield Camino 3500.6ms 3720.0ms Tests with a default profile for both browsers, on a consumer iMac (core2 duo, 2Ghz, 2Gb ram). Filed as UNCO, not sure if this is really a Camino bug or if there is an underlying Core bug.
Camino build: 2.0a1pre (1.9pre 2008040622)
build: Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9pre) Gecko/2008040623 Minefield/3.0pre
I didn't test Minefield, but for reference. Tests with my own build, on a MacBookPro ( Core 2 Duo, 2.16 GHz, 2 GB RAM). Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.4; en; rv:1.9pre) Gecko/2008040717 Camino/2.0a1pre (like Firefox/3.0pre) Total: 3485.4ms +/- 0.3%
Attached file Sunspider Results (obsolete) —
Attached file Sunspider Results
Attachment #314051 - Attachment is obsolete: true
Attachment #314052 - Attachment mime type: application/octet-stream → text/plain
Test with Official Nigtly Build (2008040700) ================================ RESULTS (means and 95% confidence intervals) -------------------------------------------- Total: 3513.8ms +/- 0.6% --------------------------------------------
Are you testing Minefield nightlies? They're built with PGO (http://developer.mozilla.org/en/docs/Building_with_Profile-Guided_Optimization), and that resulted in about a 10% gain in SunSpider numbers, if I recall correctly - your Camino build isn't, presumably.
(In reply to comment #7) > Are you testing Minefield nightlies? They're built with PGO On the Mac? My understanding is that PGO was not being used on the Mac because there were too many obstacles for this stage in the game (bug 417045), and that wiki page you cited only talks about Windows.
Oh, I thought Camino was a Windows browser that used native win32 widgets. (You're right of course, not sure what I was thinking.)
Shark information from a complete run in each browser would be helpful, so we can see if the extra time is anywhere obvious.
This is obviously happening, so I'm going to confirm it, and if it turns out it's a Core bug (which we'll probably know once we get Shark info), we'll have plenty of information to give them when we kick it over there.
Status: UNCONFIRMED → NEW
Ever confirmed: true
Reports are the 1.9.2 experimental builds are faster than Firefox 3.6 at sunspider, and since we've never gotten Shark profiles here, I think we should probably just WONTFIX this one.
Status: NEW → RESOLVED
Closed: 16 years ago
Resolution: --- → INCOMPLETE
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: