win95/nt image display performance regressed from 4.x display performance

VERIFIED WORKSFORME

Status

P1
critical
VERIFIED WORKSFORME
20 years ago
10 years ago

People

(Reporter: mle, Assigned: pnunn)

Tracking

Trunk
x86
Windows 95

Firefox Tracking Flags

(Not tracked)

Details

(Whiteboard: [Perf], URL)

(Reporter)

Description

20 years ago
I have run tests on the file at the URL provided - it has just a few image files
in the 10-70 KB range in JPEG, GIF, and PNG formats.  The tests were run
on the same computer (133MHz, 64M) under both '95 and NT.  Here is approximate
wall clock times to load the file:

NT   Build from December    6 seconds
95   Build from December   70 seconds
     Build 02/08 (today)   55 seconds

Navigator 4.0 under '95 on this computer displays the file instantly
and reports all the graphics files loaded in one or two seconds.

'95 is obviously very sick but the poor performance on NT
compared to Navigator 4.0 seems quite notable to me as well.

I would appreciate knowing what you think about this problem
if you could spare a moment to send me a note.

Michael Leventhal
CITEC
mle@citec.fi

Updated

20 years ago
QA Contact: 1698
(Assignee)

Updated

20 years ago
Status: NEW → ASSIGNED
(Assignee)

Comment 1

20 years ago
This sounds like it is related to the cache
problem in the current builds. People are
currently working on the cache problem. I will
keep this bug open to be sure I check it after
the next iteration of netlib fixes.
-pn
(Assignee)

Comment 2

20 years ago
Just a quick note. I see a huge difference in performance between
viewer and appviewer on linux. I don't have a current, unmodified
build on wintel for testing. I am assuming the test was run on
apprunner.   Let me know if I'm wrong..
-thanks,
pn
(Assignee)

Updated

20 years ago
Summary: Win '95 Image Load 12X slower than NT → win95/nt image display performance regressed from 4.x display performance
(Reporter)

Comment 3

20 years ago
My tests were run using the viewer application.
(Reporter)

Comment 4

20 years ago
I have important new information about this bug.  The performance problem only
occurs when the html and graphic files are read off disk - read off the
internet performance is comparable to Navigator 4.x.
I repeated the test using the 2/10 build today with the same result.  Under '95
approximately 55 seconds to read the file off local disk, 11 over the 'net.
Under NT from local disk it is about 6 seconds.  Same 133mHz, 64M machine.
(Assignee)

Comment 5

20 years ago
Thanks, Michael. This info narrows down the
problem considerably.
-pn

Updated

20 years ago
Whiteboard: [Perf]

Comment 6

20 years ago
Putting on [Perf] radar.
(Assignee)

Updated

20 years ago
Target Milestone: M5
(Assignee)

Comment 7

20 years ago
eli:
Could you run a quick test. I think the issues
for this bug have been addressed and this is now
closable.
-pn
(Assignee)

Updated

20 years ago
Target Milestone: M5 → M6

Comment 8

20 years ago
Sure. Will do on Tuesday.

Comment 9

20 years ago
Okay, Wednesday.

The short answer is that loading the driving.jpg image from this page after
saving it locally (4.27.99 build, Win95) simply results in the chrome being
replaced with random bit garbage, and doesn't actually load the image.

I'll spend more time investigating this tomorrow and have useful comments back.
Thanks.

Comment 10

20 years ago
So, now that we actually a *working* File/Open command, I'm no longer seeing this
problem on the 4.3.99 builds (using the driving.jpg image from the user manual.)

Specifically, I do note that a network image load is taking twice as long as a
local load, but I can attribute this time to the fact that it redraws the chrome
& sidebar upon reload. (Will write up a separate bug report for that.)

mle@citec.fi, are you still seeing a performance degradation on the current
builds? If so, could you possibly share more of your wisdom on this issue?

Thank you!
(Assignee)

Updated

20 years ago
Status: ASSIGNED → RESOLVED
Last Resolved: 20 years ago
Resolution: --- → WORKSFORME
(Assignee)

Comment 11

20 years ago
Closing.

Updated

20 years ago
Status: RESOLVED → VERIFIED

Comment 12

20 years ago
Since no feedback received, yeah, verifying as WORKSFORME.

Comment 13

20 years ago
Since no feedback received, verifying as WORKSFORME.
Product: Core → Core Graveyard
You need to log in before you can comment on or make changes to this bug.