Performance Difference between machines

[nafej]

I was running some performance tests for the web application I work
on, both on my desktop and laptop. I found that javascript processes
MUCH slower on my desktop, even though it's a more powerful machine.
The stats are as follows:

Laptop: Pentium M, 1.86gHz, 1gb RAM, IE7
Desktop: Pentium 4, 2.80gHz, 1gb RAM, IE6

I have confirmed that co-workers with similar setups show the same
results. Also, when I run javascript tests on both boxes (from
http://andrew.hedges.name/experiments/speed_test/index.html and
others), my laptop returns a time of approximately 50% of my desktop.

Also, if I test performance on a static HTML page, my desktop actually
loads the page faster. So, the page is being returned/rendered faster
on my desktop, but it just seems to be that client side code is
running much slower there.

I have checked and the browser or it's settings don't seem to be the
issue (I tried my desktop with IE7, and I also confirmed that the
settings appear to match)

Also, my laptop has many apps running during these tests while my
desktop doesn't, making me think that the desktop would be faster if
anything.

Does anyone know what other settings, configurations, etc could be
causing this?? Any input would be GREATLY appreciated!

Thanks, Jeff

 

[Erwin Moller]

I was running some performance tests for the web application I work
on, both on my desktop and laptop. I found that javascript processes
MUCH slower on my desktop, even though it's a more powerful machine.
The stats are as follows:

Laptop: Pentium M, 1.86gHz, 1gb RAM, IE7
Desktop: Pentium 4, 2.80gHz, 1gb RAM, IE6

I have confirmed that co-workers with similar setups show the same
results. Also, when I run javascript tests on both boxes (from
http://andrew.hedges.name/experiments/speed_test/index.html and
others), my laptop returns a time of approximately 50% of my desktop.

Also, if I test performance on a static HTML page, my desktop actually
loads the page faster. So, the page is being returned/rendered faster
on my desktop, but it just seems to be that client side code is
running much slower there.

I have checked and the browser or it's settings don't seem to be the
issue (I tried my desktop with IE7, and I also confirmed that the
settings appear to match)

Also, my laptop has many apps running during these tests while my
desktop doesn't, making me think that the desktop would be faster if
anything.

Does anyone know what other settings, configurations, etc could be
causing this?? Any input would be GREATLY appreciated!

Thanks, Jeff

Hi Jeff,

Benchmarking is easy, designing a good balanced computer is difficult.

If you only look at CPU speed and RAM, you miss the rest, like motherboard,
videoadapter, etc.
It is hard to design a computer from scratch, and decide which components
will perform best in combination with other components.

An example: Remember to old winmodems? These modems used the CPU to do their
job, so they didn't need that hardware themselfs, so they were cheaper. But
your computer runs a lot slower when online.

Also: drivers. A great videoadapter with stinking drivers will not be
performing at peak efficiency.

etc. etc.

Regards,
Erwin Moller

 

[Tom Cole]

Hi Jeff,

Benchmarking is easy, designing a good balanced computer is difficult.

If you only look at CPU speed and RAM, you miss the rest, like motherboard,
videoadapter, etc.
It is hard to design a computer from scratch, and decide which components
will perform best in combination with other components.

An example: Remember to old winmodems? These modems used the CPU to do their
job, so they didn't need that hardware themselfs, so they were cheaper. But
your computer runs a lot slower when online.

Also: drivers. A great videoadapter with stinking drivers will not be
performing at peak efficiency.

etc. etc.

Regards,
Erwin Moller- Hide quoted text -

- Show quoted text -

I have no tests to support it, but you also have to take into account
IE7 vs. IE6. It is the browser that parses and executes the script so
that is where I would start.

A better comparison would be to run IE7 or IE6 on both machines and
then see what you get...

 

[Julian Turner]

I was running some performance tests for the web application I work
on, both on my desktop and laptop. I found that javascript processes
MUCH slower on my desktop, even though it's a more powerful machine.
The stats are as follows:

Laptop: Pentium M, 1.86gHz, 1gb RAM, IE7
Desktop: Pentium 4, 2.80gHz, 1gb RAM, IE6

I have confirmed that co-workers with similar setups show the same
results. Also, when I run javascript tests on both boxes (fromhttp://andrew.hedges.name/experiments/speed_test/index.htmland
others), my laptop returns a time of approximately 50% of my desktop.

Also, if I test performance on a static HTML page, my desktop actually
loads the page faster. So, the page is being returned/rendered faster
on my desktop, but it just seems to be that client side code is
running much slower there.

I have checked and the browser or it's settings don't seem to be the
issue (I tried my desktop with IE7, and I also confirmed that the
settings appear to match)

Also, my laptop has many apps running during these tests while my
desktop doesn't, making me think that the desktop would be faster if
anything.

Does anyone know what other settings, configurations, etc could be
causing this?? Any input would be GREATLY appreciated!

Thanks, Jeff

Hi

I have noted suggestions on some blogs that Microsoft made some
improvements to the JavaScript interpreter in IE7, which may account
for this.

Julian

 

[nafej]

Thanks guys.

This isn't a browser issue though. 5 others on my team have run the
same performance tests and they always run faster on their laptops
then their desktops. But, most of them have the opposite setup as me
(IE6 on their laptop, IE7 on their desktop). So, it doesn't appear to
have anything to do with the browsers Javascript interpreter (though
I'm going to research this angle more).

I'm starting to think it's due to something motherboard related or
something like that that's out of my control?? Very curious though.

If anyone wants to run this test of curiousity (http://
andrew.hedges.name/experiments/speed_test/index.html), our laptops
were usually returning Time Differences of approx. 18 seconds, and our
desktops return in approximately 35 seconds. For test 2 on that page,
we get 33 laptop and 65 desktop. Interesting...

 

[Richard Cornford]

I was running some performance tests for the web application
I work on, both on my desktop and laptop. I found that
javascript processes MUCH slower on my desktop, even though
it's a more powerful machine. The stats are as follows:

Laptop: Pentium M, 1.86gHz, 1gb RAM, IE7
Desktop: Pentium 4, 2.80gHz, 1gb RAM, IE6

<snip>

Drawing useful conclusions from any comparison tests requires that the
number of variable factors be reduced, so comparing IE 7 with IE 6 on
different machines with different processors of different speeds is not
going to do more than highlight that there are differences. Attributing
those differences will not necessarily be possible.

It has been observed in the past that a Pentium 3 will do direct string
comparison faster than it will do regular expression test, while on a
Pentium 4 the reverse is true (and unexpectedly so). The proposed
explanation for this is that the Pentium 4 has a much longer pipeline and
a longer pipeline provides an advantage for the type of processing
required by regular expressions, while hindering the execution of simple
code. As I recall the Pentium M has a longer pipeline than the P3 and a
shorter pipeline than the P4 (and there are other physical differences
that should be expected to be significant), but any differences observed
may depend a great deal on the nature of the code being tested.

Richard.