What is the problem with Internet Explorer and SunSpider string bas64 test?

What is the problem with Internet Explorer and the
http://www2.webkit.org/perf/sunspide...sunspider.html
string base64 test.

I have four Pentium systems running Windows XP Professional
with Service Pack 3 and Internet Explorer V7. Various Pentium
processors are used.

I ran the SunSpider JavaScript test from www2.webkit.org
on all four machines using both IE7 and FireFox V3.5.

As expected, FireFox was faster.

However, the speed ratios for the various subtests
differed a great deal between the two browsers.

In some cases the ratios were in the 1-3 to 1 range.
For other subtests the ratios were in the 10-15 to 1 range.

However for some tests the ratios were greater than 100 to 1.
In particular, for the string-base64 test the
ratio was more than 1000 (one thousand) to one.

On one system the string-base64 test took 199278.4ms (3+minutes)
with IE7 and 41 milliseconds (1/20th of a second), for
a ratio of almost 4800 to 1.

I looked on Google and it seems like IE7 was having
problems with the string-base64 test as early as 2008 January,
but I couldn't find an explanation.

Could all of the people who had the problem have some setting
wrong?

Does anyone know if the performance problem with base64 is fixed
in Internet Explorer V8?

Thank you.


: 199278.4ms +/- 2.7%

Is SunJava (JRE) v1.6.0_14 installed? Are any earlier versions still
installed? What IE Add-ons are enabled and how many of them are present &
enabled in FF? Are all computers fully-patched at Windows Update?

Mark F wrote:
> What is the problem with Internet Explorer and the
> http://www2.webkit.org/perf/sunspide...sunspider.html
> string base64 test.
>
> I have four Pentium systems running Windows XP Professional
> with Service Pack 3 and Internet Explorer V7. Various Pentium
> processors are used.
>
> I ran the SunSpider JavaScript test from www2.webkit.org
> on all four machines using both IE7 and FireFox V3.5.
>
> As expected, FireFox was faster.
>
> However, the speed ratios for the various subtests
> differed a great deal between the two browsers.
>
> In some cases the ratios were in the 1-3 to 1 range.
> For other subtests the ratios were in the 10-15 to 1 range.
>
> However for some tests the ratios were greater than 100 to 1.
> In particular, for the string-base64 test the
> ratio was more than 1000 (one thousand) to one.
>
> On one system the string-base64 test took 199278.4ms (3+minutes)
> with IE7 and 41 milliseconds (1/20th of a second), for
> a ratio of almost 4800 to 1.
>
> I looked on Google and it seems like IE7 was having
> problems with the string-base64 test as early as 2008 January,
> but I couldn't find an explanation.
>
> Could all of the people who had the problem have some setting
> wrong?
>
> Does anyone know if the performance problem with base64 is fixed
> in Internet Explorer V8?
>
> Thank you.
>
>
>> 199278.4ms +/- 2.7%

On Thu, 2 Jul 2009 11:40:37 -0400, "PA Bear [MS MVP]"
<PABearMVP@gmail.com> wrote:

>
> Is SunJava (JRE) v1.6.0_14 installed? Are any earlier versions still
> installed? What IE Add-ons are enabled and how many of them are present &
> enabled in FF? Are all computers fully-patched at Windows Update?
I'm not sure how to get a list of all of the installed add-ons. It
turns out that various things that I don't think of as add-ons or
which came along with sites such as Google and Microsoft sticking in
crap that I didn't explicitly ask for means that I a lot more crap
than I expected, even on my newest machine, which I have only
installed about 5% of my software on.

On the my newest and fastest machine
Internet Explorer
> Tools > Manage Add-ons > Enable or Disable Add-ons...
shows about 40 "Add-ons that have been used by Internet Explorer"
and 9 "Downloaded ActiveX Controls (32-bit).

I disabled everything except for
1. these ActiveX Controls (32-bit)
Java Plug-in 1.4.2_03 file jp2iexp.dll
Java Plug-in 1.6.0_13 file jp2iexp.dll
Java Plug-in 1.6.0_13 file jp2iexp.dll
^-***Not sure why the above shows up 2 times
both from Publisher
"(Not verified) Sun Microsystems, Inc."
Java Plug-in 1.6.0_13 file npjpi160_13.dll

2. Add-ons that have been used by Internet Explorer
Java Plug-in 1.6.0_13 file jp2iexp.dll
Java Plug-in 1.6.0_13 file jp2iexp.dll <-***again repeated
Java(TM) Platform SE binary file jp2iexpl.dll
Java(tm) Plug-In 2 SSV Helper file jp2ssv.dll
JQSIEStartDetectorImpl Class file jqs_plugin.dll

System is has updates from Microsoft Until about 2009 June 5,
so there is another set to go

I ran the SunSpider benchmark test again and everything has
the same time except for string-base64 which was 35109.4ms
and now is 15931.0ms (compared to 28.6ms in FireFox V3.5).
Still taking about 100 times too long compared to the other
tests.

How do I remove all of the old Java Plug-in stuff and just
put in the latest version? A search at www.sun.com for
1.6.0.13 gets hits, but 1.6.0.14 doesn't get a hit,
and I'm not sure how to remove the old stuff.
>
> Mark F wrote:
> > What is the problem with Internet Explorer and the
> > http://www2.webkit.org/perf/sunspide...sunspider.html
> > string base64 test.
> >
> > I have four Pentium systems running Windows XP Professional
> > with Service Pack 3 and Internet Explorer V7. Various Pentium
> > processors are used.
> >
> > I ran the SunSpider JavaScript test from www2.webkit.org
> > on all four machines using both IE7 and FireFox V3.5.
> >
> > As expected, FireFox was faster.
> >
> > However, the speed ratios for the various subtests
> > differed a great deal between the two browsers.
> >
> > In some cases the ratios were in the 1-3 to 1 range.
> > For other subtests the ratios were in the 10-15 to 1 range.
> >
> > However for some tests the ratios were greater than 100 to 1.
> > In particular, for the string-base64 test the
> > ratio was more than 1000 (one thousand) to one.
> >
> > On one system the string-base64 test took 199278.4ms (3+minutes)
> > with IE7 and 41 milliseconds (1/20th of a second), for
> > a ratio of almost 4800 to 1.
> >
> > I looked on Google and it seems like IE7 was having
> > problems with the string-base64 test as early as 2008 January,
> > but I couldn't find an explanation.
> >
> > Could all of the people who had the problem have some setting
> > wrong?
> >
> > Does anyone know if the performance problem with base64 is fixed
> > in Internet Explorer V8?
> >
> > Thank you.

> System is has updates from Microsoft Until about 2009 June 5,
> so there is another set to go

Then you computer is NOT fully patched! Please address this ASAP.

> How do I remove all of the old Java Plug-in stuff and just
> put in the latest version?

See http://aumha.net/viewtopic.php?f=26&t=40890

My point was that there can be a lot of stuff runing in the IE background
(often by default) that may not be found in and impeding FF.

Every single one of your enabled IE Add-ons must load everytime you open a
new window or tab.
--
~Robear Dyer (PA Bear)
MS MVP-IE, Mail, Security, Windows Client - since 2002


Mark F wrote:
> I'm not sure how to get a list of all of the installed add-ons. It
> turns out that various things that I don't think of as add-ons or
> which came along with sites such as Google and Microsoft sticking in
> crap that I didn't explicitly ask for means that I a lot more crap
> than I expected, even on my newest machine, which I have only
> installed about 5% of my software on.
>
> On the my newest and fastest machine
> Internet Explorer Tools > Manage Add-ons > Enable or Disable Add-ons...
> shows about 40 "Add-ons that have been used by Internet Explorer"
> and 9 "Downloaded ActiveX Controls (32-bit).
>
> I disabled everything except for
> 1. these ActiveX Controls (32-bit)
> Java Plug-in 1.4.2_03 file jp2iexp.dll
> Java Plug-in 1.6.0_13 file jp2iexp.dll
> Java Plug-in 1.6.0_13 file jp2iexp.dll
> ^-***Not sure why the above shows up 2 times
> both from Publisher
> "(Not verified) Sun Microsystems, Inc."
> Java Plug-in 1.6.0_13 file npjpi160_13.dll
>
> 2. Add-ons that have been used by Internet Explorer
> Java Plug-in 1.6.0_13 file jp2iexp.dll
> Java Plug-in 1.6.0_13 file jp2iexp.dll <-***again repeated
> Java(TM) Platform SE binary file jp2iexpl.dll
> Java(tm) Plug-In 2 SSV Helper file jp2ssv.dll
> JQSIEStartDetectorImpl Class file jqs_plugin.dll
>
> System is has updates from Microsoft Until about 2009 June 5,
> so there is another set to go
>
> I ran the SunSpider benchmark test again and everything has
> the same time except for string-base64 which was 35109.4ms
> and now is 15931.0ms (compared to 28.6ms in FireFox V3.5).
> Still taking about 100 times too long compared to the other
> tests.
>
> How do I remove all of the old Java Plug-in stuff and just
> put in the latest version? A search at www.sun.com for
> 1.6.0.13 gets hits, but 1.6.0.14 doesn't get a hit,
> and I'm not sure how to remove the old stuff.
>
>> PA Bear wrote:
>> Is SunJava (JRE) v1.6.0_14 installed? Are any earlier versions still
>> installed? What IE Add-ons are enabled and how many of them are present
>> &
>> enabled in FF? Are all computers fully-patched at Windows Update?
>>
>> Mark F wrote:
>>> What is the problem with Internet Explorer and the
>>> http://www2.webkit.org/perf/sunspide...sunspider.html
>>> string base64 test.
>>>
>>> I have four Pentium systems running Windows XP Professional
>>> with Service Pack 3 and Internet Explorer V7. Various Pentium
>>> processors are used.
>>>
>>> I ran the SunSpider JavaScript test from www2.webkit.org
>>> on all four machines using both IE7 and FireFox V3.5.
>>>
>>> As expected, FireFox was faster.
>>>
>>> However, the speed ratios for the various subtests
>>> differed a great deal between the two browsers.
>>>
>>> In some cases the ratios were in the 1-3 to 1 range.
>>> For other subtests the ratios were in the 10-15 to 1 range.
>>>
>>> However for some tests the ratios were greater than 100 to 1.
>>> In particular, for the string-base64 test the
>>> ratio was more than 1000 (one thousand) to one.
>>>
>>> On one system the string-base64 test took 199278.4ms (3+minutes)
>>> with IE7 and 41 milliseconds (1/20th of a second), for
>>> a ratio of almost 4800 to 1.
>>>
>>> I looked on Google and it seems like IE7 was having
>>> problems with the string-base64 test as early as 2008 January,
>>> but I couldn't find an explanation.
>>>
>>> Could all of the people who had the problem have some setting
>>> wrong?
>>>
>>> Does anyone know if the performance problem with base64 is fixed
>>> in Internet Explorer V8?
>>>
>>> Thank you.

On Fri, 3 Jul 2009 00:51:33 -0400, "PA Bear [MS MVP]"
<PABearMVP@gmail.com> wrote:

>
> > System is has updates from Microsoft Until about 2009 June 5,
> > so there is another set to go
>
> Then you computer is NOT fully patched! Please address this ASAP.
Can't do on any of my computers now; perhaps next month.

I did try IE8 on a VMware Workstation V6 virtual machine and
there was no problem with string-base64.
>
> > How do I remove all of the old Java Plug-in stuff and just
> > put in the latest version?
>
> See http://aumha.net/viewtopic.php?f=26&t=40890
>
> My point was that there can be a lot of stuff runing in the IE background
> (often by default) that may not be found in and impeding FF.
>
> Every single one of your enabled IE Add-ons must load everytime you open a
> new window or tab.
Do they have to load? In any event I can see that there is no I/O
activity when the string-base64 test is running and many of the tests
take reasonable times to run. Note that all of the other sub-tests
run in the same environment and many of them run almost a fast
as they do under FireFox.

The problem has to be that the IE7 behaves in a pathological manner
for string-base64.

Can anyone run the
http://www2.webkit.org/perf/sunspide...sunspider.html
test with IE7 and see what happens for the string-base64 test? If he
doesn't have a problem I could try to see what the difference is
in excess IE7 crap on my systems.

Does anyone have a way to get a cut and paste list of add-ons from
IE7?