Google releases Chrome speed test video, but it was faked

Today Google released a video demonstrated the “lightning speed” of Google Chrome.  The video compared the browser to three random scenarios, Chrome Vs. potato gun, Chrome Vs. sound waves and Chrome Vs. lightning.

Each scene was shot with a high speed camera, at 1920 x 1080 and up to 2700fps.  Google claims on their YouTube video description that this video was “filmed at actual web page rendering times,” including a “15Mbps Internet connection.“


As the video demonstrates, the browser is fast, very fast.  However, as one of Neowin's forum member, Singh400, spotted, the video was actually faked.  During the Chrome Vs. potato scene, the potato zooms past the browser loading a webpage, where you can clearly make out the file is actually loaded from the desktop of a computer, and not a real webpage.

chromefake1

The last two scenarios is difficult to say if Chrome is loading from a desktop file or is loading from an actual webpage.  Guessing from the first scenario and the speed loading time, it's likely that the last two scenarios were also loaded from the desktop.

Download Google Chrome

Thanks to Singh400 for the tip!

Report a problem with article
Previous Story

Adobe shows prototype Android tablet running Flash

Next Story

Medal of Honor ruined by annual releases, next due in October

106 Comments

Commenting is disabled on this article.

Not Impressed. I just tried this on my PC running Windows XP and Firefox. I have a 1.5mbps connection and used the URL for allrecipes.com. Rendered for me about as fast as that video. Type in the URL yourself. I've never been to that page before so it definitely wasn't in cache.

Cool advertising perhaps but that's it.

If loading from file:\\ why does it show the user typing in a google search to begin with? That makes this look fishy. Maybe it is only testing rendering speed but it is misleading.

It looks fake.
1) loading from file:\\ is tricky.
2) the refresh from up to down looks suspicious fake.
3) i don't see the vertical scan or any change in the lcd.
4) A ray is generated just a meter away from the screen and the lcd didn't show any interference, not even a blink.
5) I know that Chrome is not to fast, not at least in my rig.

I get why they used a saved copy of the recipes page. They want to take server response out of the "rendering speed" question

The last 2 look like URLs instead of file links. The Weather one infact is VERY short, so they were likely coming straight out of cache.

That said, what they have effectively proven here is that rendering time is the LEAST of your issues in web surfing, so who freaking cares if another browser is a half second longer in rendering if the real delay is server response and transmission which is all outside of the browser's control.

Mike Chaliy said,
Whooa! Adress bar actually suggests that Google uses Win7 or Vista .

In the screenshot, it looks like Chrome's non-Aero skin (most likely XP).

Edited by Denis W., May 6 2010, 6:12pm :

rm20010 said,

In the screenshot, it looks like Chrome's non-Aero skin (most likely XP).

In XP, a user's files were placed under C:\Documents and Settings\. In the image, the file was under C:\Users, the location used by default by Vista and 7. And it is possible to turn off Aero in Vista and 7.

They are not testing the speed of their net-connection, or speed of the websites server, they are testing how fast Chrome renders web-pages...

kraized said,
They are not testing the speed of their net-connection, or speed of the websites server, they are testing how fast Chrome renders web-pages...
Yet they mention the connection speed, talk about actual web page rendering, and pretend to type in a URL.

kraized said,
They are not testing the speed of their net-connection, or speed of the websites server, they are testing how fast Chrome renders web-pages...

Like pointed out a dozen of times already Google mentioned using a "15Mbps Internet connection", which is completely irrelevant and misleading if the pages are loaded of the desktop...

I don't really care if it is fake or not.

The actual loading times compared with IE/Opera/Firefox don't really make any difference to me as they will only go as quick as my connection.

Mr Spoon said,
I don't really care if it is fake or not.

The actual loading times compared with IE/Opera/Firefox don't really make any difference to me as they will only go as quick as my connection.

I notice a HUGE difference between Chrome's overall speed and IE's.

Faster than sound waves and lightning, yeah right. This is a marketing video, it has nothing to do with reality so it doesn't really matter if the page is loaded from desktop or not, it's still impossible.

Google Chrome is still blazing fast...as i know... i try all the browsers out there and nothing compares to Google Chrome.

Back in the real world, the insignificance in rendering time compared to the time required to load the page over a connection makes most speed tests irrelevant.

billyea said,
Back in the real world, the insignificance in rendering time compared to the time required to load the page over a connection makes most speed tests irrelevant.

Yep. Some of us are still using slow internet connections. And my ISP upgraded the connection for new buyers, but the old had to call them (if they ever found out in the first).

I don't care whether it was faked or not, its still a clever bit of marketing, and an excellently produced video. But I still use Firefox!

Mr Dickie said,
I don't care whether it was faked or not, its still a clever bit of marketing, and an excellently produced video. But I still use Firefox!

I would use Firefox if only its memory usage was lower. As it is, I have to stay with Chrome.

mysticalone said,
Singh400 fails at understanding what render means.
Oh please, spare us your liberal ass-kissing interpretation.

mysticalone said,
Singh400 fails at understanding what render means.

Why did they mention "15Mbps Internet connection", then?

How often do you load pages from your desktop? Who cares how fast it is then. I want a fast browser when I'm browsing.

I'm sorry ..... I do not see what the hell Google is going on about. Speed is relative.... relative to what in this case ??? It "SHOULD" be relative to other browsers not this crap. All this is is advertising for Chrome as it doesn't prove a bloody thing as far as speed goes.

Caveman-ugh said,
I'm sorry ..... I do not see what the hell Google is going on about. Speed is relative.... relative to what in this case ??? It "SHOULD" be relative to other browsers not this crap. All this is is advertising for Chrome as it doesn't prove a bloody thing as far as speed goes.

Maybe it isn't (or shouldn't be) about "benchmarks" in the usual sense of the word. Kind of a marketing joke.

Any browser can load a local file fast enough to be insignificant to the user. The test is disingenuous and irrelevant, it's just an advertisement to make people BELIEVE that it is fast. A worthy test would be to compare its page load times with other browsers on the top X sites of the web. Another factor that is more difficult to quantify is how quickly it shows users part of the page while its waiting to load the rest on low bandwidth connections. From what I've noticed Chrome tends to wait for the majority of the page instead of showing the user what it has and then having to reflow when it gets more.

Andrew Lyle said,
Just like the opera mini test, comparing safari and itself on the EDGE network (iPhone)

No, not like the Opera Mini test at all. In fact, the complete opposite.

The whole point of Opera Mini is that it uses a server to crunch the page before sending it to the phone. That's what Opera always say when they talk about Opera Mini. So not only are they not misleading people, but they are going out of their way to brag about their awesome server compression technology.

It's funny to see those claiming it was a "render test". Given this is a marketing video, not a technical video, it's obvious they're trying to sell you on how fast it renders pages... on the Internet. The mention of available bandwidth and the scene where AllRecipes is typed in clearly point this direction.

URL or file on disk, the test still seems odd as they use the Forward button -- not Go/Refresh.

Rafael said,
It's funny to see those claiming it was a "render test". Given this is a marketing video, not a technical video, it's obvious they're trying to sell you on how fast it renders pages... on the Internet. The mention of available bandwidth and the scene where AllRecipes is typed in clearly point this direction.

URL or file on disk, the test still seems odd as they use the Forward button -- not Go/Refresh.

Exactly.

Rafael said,
It's funny to see those claiming it was a "render test". Given this is a marketing video, not a technical video, it's obvious they're trying to sell you on how fast it renders pages... on the Internet. The mention of available bandwidth and the scene where AllRecipes is typed in clearly point this direction.

URL or file on disk, the test still seems odd as they use the Forward button -- not Go/Refresh.


Except they mention a "15Mbps Internet connection".

This falsely gives the viewer the impression that the connection is relevant (leading the viewer to believe that the page is being loaded over said connection).

Rafael said,
It's funny to see those claiming it was a "render test". Given this is a marketing video, not a technical video, it's obvious ...
Yes. Marketing.

Burger King burgers are shown customized with sesame seeds glued on perfectly. Microsoft commercials show drawings of orchestras appearing around a young kid.

These things don't REALLY happen. These aren't the actual products we consume. They are nothing more but marketing imagery.

One really should not be expecting technical benchmark data from commercials.

The difference Mark is no one goes to McDonald's and demands a hamburger that looks as picture perfect as the one in the commercial. McDonald's is not purposely misleading the customer with inaccurate information.

Google, well, they just keep running father from their "do no evil" slogan.

PreKe said,

Except they mention a "15Mbps Internet connection".

This falsely gives the viewer the impression that the connection is relevant (leading the viewer to believe that the page is being loaded over said connection).

I agree... maybe you misread my comment?

random_n said,
That video looks like it would have been a ton of fun to make

I watched the making of video, they seemed to enjoy it. Where else do you get to do this kinda stuff?

Benjamin Rubenstein said,
It's still fast. I also doubt that other browsers would render it as fast, even locally.

http://ie.microsoft.com/testdrive/

Not only the IE9 preview, but many OTHER browsers run many of the speed tests considerably faster. Compiling and 'initially' rendering a page is far different from actually keeping the performance acceptable on dynamic content.

They should specify this, and since most people browses the internet using the internet! The test should use the internet. I don't care the speed of a browser in a high speed connection but in a slow connection and see how they manage to improve the experience

basques said,
Well the test is about the rendering speed of the HTML not about the bandwidth or internet speed.

But they do mention the internet connection speed, misleading the viewer into assuming that it's relevant.

Even if they faked it, it was cool.

Why is there a Google Chrome download link at the bottom if they are saying that they are liars? You don't call someone out on something and then also put up their download link to get the browser.

Tech Star said,
Even if they faked it, it was cool.

Why is there a Google Chrome download link at the bottom if they are saying that they are liars? You don't call someone out on something and then also put up their download link to get the browser.

Maybe to test it yourself.

Tech Star said,
Why is there a Google Chrome download link at the bottom if they are saying that they are liars? You don't call someone out on something and then also put up their download link to get the browser.
Because they're not vindictive, and it doesn't mean Chrome is a bad browser?

Udedenkz said,
I don't think anyone thought it was real.

Yeah, I did not think it was real either.
The whole thing was a joke.

A person should be really stupid to think that
developers use a elaborate setup like firing a potato, splash paint etc. to measure or tune their application performance.

Thumbs up for the Cool Video Google.
People who thought it was real

Edited by figgy, May 6 2010, 2:11am :

Tekkerson said,
Read the first comment.

Not a real world test, a test that is supposed to look like a realworld test is simply dishonest.

Also, this implies to average user that their internet speed will be uber fast with Chrome. Try that on 256kbit connection!

Opera would probably dominate.

Edited by Udedenkz, May 6 2010, 3:21am : Opera

Udedenkz said,

Not a real world test, a test that is supposed to look like a realworld test is simply dishonest.

Also, this implies to average user that their internet speed will be uber fast with Chrome. Try that on 256kbit connection!

Opera would probably dominate.

256kbit connections still exist?

Hell-In-A-Handbasket said,
Makes sense, as browser test would be rendering speed, not Internet speed I think

Exactly - Internet speed would be constant between all browsers, so the rendering time is pretty much the main variable when it comes to speed.

+1. The rendering wasn't faked, I think having an internet connection is irrelevant since the test is about rendering. I think it is assumed that speed varies within different internet connections. Shame on you Neowin for being sensationalist in news.

Edited by Tekkerson, May 6 2010, 1:16am :

Just because one is loaded from the desktop doesn't prove the rest of the tests are loaded from desktop as well. The other tests could be legitimate real sites. It would be wrong to assume that they weren't using some tests with their internet connection when we don't really know.

Even then, I believe the tests are focused on rendering times rather than connection speed. So it makes perfect sense that they would launch them from desktop.

Either way, xTdub mentioned something interesting. The last test seems to have a short url, perhaps they were having mixed tests.

Andrew Lyle said,
I understand that, but what's with the Internet connection if they're all loaded from the desktop?

By loading from desktop, it makes net speed irrelevant.

I would edit this, since imo it seems a little misleading. I know Neowin wants ton of views on their news posts but come on!

Tekkerson said,
Just because one is loaded from the desktop doesn't prove the rest of the tests are loaded from desktop as well. The other tests could be legitimate real sites. It would be wrong to assume that they weren't using some tests with their internet connection when we don't really know.

Even then, I believe the tests are focused on rendering times rather than connection speed. So it makes perfect sense that they would launch them from desktop.

Either way, xTdub mentioned something interesting. The last test seems to have a short url, perhaps they were having mixed tests.

Given the length of the address in the bar for the other 2 tests I would be able to safely assume they were also loaded from the desktop.

Hell-In-A-Handbasket said,
Makes sense, as browser test would be rendering speed, not Internet speed I think

How a browser handles internet traffic and lag are just as important. It is the EXACT reason IE4 won over Netscape4 in the 90s because it handled slow and dropped network connections far faster and better.

thenetavenger said,

How a browser handles internet traffic and lag are just as important. It is the EXACT reason IE4 won over Netscape4 in the 90s because it handled slow and dropped network connections far faster and better.

+1.

Hell-In-A-Handbasket said,
Makes sense, as browser test would be rendering speed, not Internet speed I think

Loading a 'site' from the Desktop means no Name resolution (DNS) is happening. Fetching a site from the Internet could involve more such factors. Also images, from the Internet the 'pieces' of the image are downloaded, and assembled together in the browser window. From the Desktop, these steps don't need to be done.

Loading a 'site' from the Desktop means no Name resolution (DNS) is happening. Fetching a site from the Internet could involve more such factors. Also images, from the Internet the 'pieces' of the image are downloaded, and assembled together in the browser window. From the Desktop, these steps don't need to be done.

Andrew Lyle said,
I understand that, but what's with the Internet connection if they're all loaded from the desktop?

Huh? Who cares about that?


This is only faked if Google somewhere claimed that the browser performance is also depending on the Internet connection speed. It seems natural to take that out of the equation, and note how they also aren't comparing it to other browsers that do load pages from the web.


If they're aiming for perfect results without external factors messing up the shooting, like a DNS server suddenly giving a delayed response of 500 ms, this makes perfect sense to me. They need the raw performance sine the kind of film is time critical and anything out of the equation of Chrome browser performance will mess it up completely.

Edited by Northgrove, May 6 2010, 7:47am :

Tekkerson said,
I would edit this, since imo it seems a little misleading. I know Neowin wants ton of views on their news posts but come on!

Yes, I came here thinking the actual filming was faking it to make it look like Chrome performed better.

Tekkerson said,
I think having an internet connection is irrelevant since the test is about rendering.

Then why did they mention their internet connection speed? That gives the false impression that it's relevant, and leads the viewer to believe that it's loading things over that connection.

but when they typed in allrecipes in the url box, they were implying there was internet connection, the ad was intentional misleading imo

Andrew Lyle said,
I understand that, but what's with the Internet connection if they're all loaded from the desktop?

Facepalm, Neowin...

Dead'Soul said,
Facepalm, Neowin...
What part of this is hard to understand?
Google claims on their YouTube video description that this video was “filmed at actual web page rendering times,” including a “15Mbps Internet connection.“

Tekkerson said,
I would edit this, since imo it seems a little misleading. I know Neowin wants ton of views on their news posts but come on!

+1

Kirkburn said,
What part of this is hard to understand?

+1. Some people just want to "be different" and quickly make contradictory comments.

ttsky said,
but when they typed in allrecipes in the url box, they were implying there was internet connection, the ad was intentional misleading imo

When they typed allrecipes into the box is was to show that they were loading a website. Do you think the average user who can barely grasp the concept of the browser itself would be able to make sense of a file location being inputed into the URL bar?

geoken said,
When they typed allrecipes into the box is was to show that they were loading a website. Do you think the average user who can barely grasp the concept of the browser itself would be able to make sense of a file location being inputed into the URL bar?

Look, not only did they type in the URL, but the info box contained the connection speed. Everything in this video misleads the viewer into thinking that they are referring to actually connecting to a web server, downloading the content, and then finally doing the rendering.