38667 - Mozilla greatly slower on ibench benchmark vs. Netscape v4

Status: VERIFIED INVALID

(SeaMonkey :: General, defect, P3)

Description

[John.Zulauf]

Run the "Load pages" test of ZD Net's iBench benchmark.  Performance for 
Mozilla is 5-6 times slower and several MB larger vs. Netscape 4.61.

This occurs both in the nightly build and the Netscape v6 pre-release.

Comment 1

[Blake Ross]

mmmm....what kind of bug report is this??

Comment 2

[R.K.Aa.]

Performance issues are of course considered important, and will be the focus
target of M18. Untill then the focus is on getting Mozilla feature complete. 

Many performance bugs are already reported and known. I suggest you file
performance issue bugs in more detail, by component.

Please read the bug writing guidelines found at:
http://www.mozilla.org/quality/bug-writing-guidelines.html

When you search Bugzilla for bugs already filed on performance issues, these
will often be marked with "performance" or "perf" as keyword or in the
subject-line.

When filing a bug, also consider using the Bugzilla helper bug-form at
http://www.mozilla.org/quality/help/bug-form.html

Setting this bug invalid since it is too vague to be of practial value.

Comment 3

[John.Zulauf]

I'm not sure I understand your comments.  You seem to say that this isn't a bug 
because other performance issues are known and will be addresses for M18.  I 
can see setting a priority or milestone to somewhere in the future, I'm not 
sure I see "too vague to be of practical value" as sensical in context.  

While the bug doesn't address a specific crash or line of code it identifies a 
specific performance issue (5-6x worse) on a specific platform (Linux) vs. a 
specific benchmark (iBench) given a specific set of options ("Load pages" 
only).  It should be reproducible, and when run in the context of profiling 
software may yeild useful results. In any case it points out a benchmark 
against which Mozilla can be objectively measured vs. other browsers.

How else should this sort of measurement be contributed and posted?

Comment 4

[R.K.Aa.]

If these are pages with images, it might sort under the imagelib bugs.
Same goes for performance issues regarding animated images.
If the issues are with table rendering, it sorts under the performance bugs
regarding table rendering. If the issues are with rendering pages using
tree-widgets, that is also a known issues. Etc.

The term "page" isn't a separate component. When the various components that
make up a page are polished for performance, benhcmark tests like the one you
refer to will all in all be affected, but just stating something is generally
"slow" is still not so useful. So yes - i'm saying this isn't "a" bug: You seem
to be referring to a complex of bug, likely consisting of known issues. 

If you believe the issues are unknown and not yet filed, then please pinpoint
what you think is the issue, so the bug becomes possible to assign, and the
various module owners/teams get a chance to work on it.

You might also want to vote for bug 26502:
"Linux rendering performance is slower than windows."

Comment 5

[Asa Dotzler [:asa]]

Sorry for the spam.  New QA Contact for Browser General.  Thanks for your help
Joseph (good luck with the new job) and welcome aboard Doron Rosenberg

Comment 6

[Stephen Koren]

Verified invalid.