'Real' JavaScript benchmark topped by...Microsoft
Douglas Crockford - the man who "discovered" JSON and a senior JavaScript architect at Yahoo! – has released a new benchmark designed to test the "actual" performance of the major web browsers on "real" JavaScript applications. And according to the test, the browser with the speediest JavaScript engine...is not Google Chrome. In …
This topic is closed for new posts.
You are correct about one thing
This article does lack enough information to accurately decide if he is a "stooge" or not. On the other hand, the internet is a vast, but indexed place.
The code he based his benchmark on is publicly available, and quite heavily littered with comments (to Mr. Crockford's credit). If you care to peruse and/or analyze it:
https://github.com/douglascrockford/JSLint
A simple LinkedIn login does give an impressive list of recommendations and work history for him. From the companies that come up, I suspect your suspicion is wide of the mark. Merely a cursory look for information (not that it should be necessary. His assertions should be assessed on their own merit) show Arctic Fox to be likely correct, and this analysis was likely presented in good faith.
As I stated above, I disagree with his methodologies and conclusions, but have no reason to doubt his earnestness. In the future, rather then make suppositions about someones motives based on your admitted ignorance (you keep bringing up the lack of information on the background of Mr Crackford in the article), maybe you could ask google? It would certainly beat libeling someone.
@Oninoshiko
Thanks. My problem was not in fact with criticism of the guy's work here, it was the ad hominem attacks that got up my nose! As it so happens (in as much as my opinion is worth anything on a topic that is not exactly within my field) I agree with the criticisms you posted above. The benchmark does appear to be highly artificial - a failing with many such perhaps?
:)
AF
@arctic fox (nasally challenged)
"ad hominem attacks that got up my nose"
Dude, this is the El Reg commentards section..
You're gonna need nostrils like zeppelin hangers if you're gonna hang out here.
Re: Even if you disagree with the guy's methodology some of the usual........
Erm, how do his disparaging comments about Google somehow mean that he's not paid my MS? He faintly praises them then gives a guess as to why Chrome fails his specially designed test. Nothing there to suggest he's not paid by MS, especially given the suspicious use of a certain company's pre-beta release vs. production builds of every other browser. That seems to me to be a test specifically designed to get that company some good press on an upcoming product...
Disparaging?
I will quote him again:
"Crockford admits that he expected Chrome to top the list. "My guess is that they overspecialized for specific styles of programming, and that Chrome was tripped up by a real program. There are some very smart people at Google, and I would expect them to rectify this.""
In what way can his remarks be construed as "disparaging"?
dis·par·age (d-sprj)
tr.v. dis·par·aged, dis·par·ag·ing, dis·par·ag·es
1. To speak of in a slighting or disrespectful way; belittle.
He referred to Google's programmers in a perfectly respectful and friendly fashion making it clear that he did not expect the result he got and that he respected their abilities. *That* was what I was referring to. Those who attack the messenger get no respect from me. Constructive and properly argued criticism (e.g Oninoshiko's first post on this thread) however is another matter.
Not how I read it
Didn't the article say that he read a puff piece about IE10, and decided to test it to see if it is true? In that case, if he wanted to reproduce their work, he would test IE10 against current browsers, not as-yet unreleased ones?
Seems rather biased
I'm guessing the author of the report isn't a multi billionaire and needs money to put food on the plate?
So is possible in the slightest that he has been given some money to come up with a test to show off how great IE10 is going to be?
Why show a preview browser against real browsers? it's pretty obvious it is an article that is trying to drum up support for the next version of IE.
Performance is a pretty pointless way of selling a browser to someone. Security and reliability should be the top considerations.
regardless of merit
The quality of IE is just not relevant.
You have to purchase it regardless of what you think.
So why test IE at all? You still have to buy it.
Why even talk about it? You still have to buy it anyway.
Microsoft is a rapist that thinks that rape is okay if they can argue the sex was good. That is all there is to their PR about IE. Entirely. People are supposed to feel good about being raped. They no choice in the matter.
It is illegal to commingle IE code with the OS. And Microsoft continues to do that. So where is the decision by any consumer? Does it matter is IE is good or not? Ask someone who has been raped. Ask someone who has a copy of IE.
Either you have the choice as a consumer or you do not. With Microsoft you absolutely do not have the choice. You MUST BUY IE. You must pay for the R&D. You must pay for the marketing. You MUST pay for the PR crap. You must pay.
So why are you even bothering to evaluate it? You have to buy it regardless. And you can not remove it. You are raped. Only Microsoft employees claim to think otherwise. Or, consultants paid by Microsoft.
Let it go Lewis...
Whether or not one pays for IE is not relevant to the discussion or the article, so change the record because it is really wearing thin.
Would a MS stooge...
really create a bench mark that placed IE9, MS's current "flagship" browser, in second to last place? If he's fixed to to place IE10 in first place, I'm sure he could have fixed it so that IE9 placed higher.
@AC RE: Would a MS stooge....
A very good and logical point - just do not expect the "cognoscenti" to appreciate it!
:)
AF
do we care?
All computers are fast enough, its only a browser, and given how many people they employ, it would be remarkable if M$ did not turn out some good code now and again.
My complaints against M$ have rarely had much to do with the code, particularly their compilers. It's the politics and marketing that gt up my nose.
I don;t see that a benchmark like this is anything but flamebait.
I agree
And IMHO the problem with browsers isn't slow Javascript, it's crap user interfaces, eg: stop buttons that have no effect or more or less crash the browser (Mozilla Firefox: Not responding). Pages that carry on downloading anyway or go blank if you press the stop button (Chrome). Is it really rocket science to kill every activity connected with a page in, say 200mS, given a multi GHz processor? Why worry about the difficult stuff, which is mostly pretty good already, when they can't get what should be the easy stuff right?
What I utterly fail to understand is ...
why is everyone so obsessed with Javascript performance? When do you ever sit there in front of your browser thinking "hmmm, this script is really taking a long time to add a table row and apply a CSS style" ?
You don't. No one does. Chasing JS performance is a waste of time, for the vast majority of users it is completely irrelevant.
Re: what I utterly fail to understand is...
I'd go further. If your script execution time is even measurable against the network time on a real internet connection then you should be taken outside and hanged by your CAT6 network cable until dead. That's true even if you've designed the pages so that http's caching model works as nature intended, rather than spewing "custom" pages for every request.
DOM performance is probably the only interesting figure of merit for browsers, since that's all that the browser *ought* to be doing.
JS Guru? Yes. Right about what to target? No.
"Asked if benchmarks should measure crap JavaScript code as well as well-written applications, Crockford says no. "There is indeed a lot of crap JavaScript out there, and most of it does not benefit from the faster JavaScript engines because performance of those applications is limited by the DOM (Document Object Model, the browser's crap API)," he tells us."
I would argue that targeting the performance of bad code is more important as that's what users are more likely to be faced with. Browser creators are far more likely to be concerned with real-World here-and-now applications than getting great performance from well written code they are unlikely to hit upon. Besides, if it is well written it won't need as much of a boost as the shite stuff.
ie 10
microsoft maybe making ie 10 the fastest but it still doesnt deter that its ie and ie is always buggy somewhere inside that browser il never leave chrome. google will keep updating chrome making it faster and they will beat microsoft to it
Not good
I lost interest when he talked about 'Real Word' then went on about a piece of specialist code.
I'm no expert by any means, but I can think of a 'Real World' (tm) test that will be far more useful.
Simply select a number of sites with the usual average of 3-4 internal bits of JS and another 5-6 'foreign' imports.
From a clean start run all of these one after the other and time them.
Where do I pick up my fee?
"Real"
It's a test of pure JavaScript. Rather than application performance I'd consider it more as a test of overhead.
I haven't done heavy JS since 2002/2003. Just like many other applications, the performance bottleneck was in updating the screen (via the DOM), although IE also had painfully slow access to "object"s in additional windows since that went through a painfully slow RPC mechanism, which necessitated putting functions in the "dumb" child window to do any DOM work.
Lewis!
Wow, what a blast from the past! I used to watch you as old "Cut 'n Paste Mettler, Esq." back on ZDNet in the 90s before they banned you for being so annoying! Man, those were good times. Just click any article on MS and look for your slightly modified posts pasted just like all your other ones, changed just enough for a good chuckle.
Anyway, good to know you're still around spouting the same bile. Cheers!
Not representative
This is not representative.
First, JSLint will be doing a few specific types of manipulations. It's like benchmarking a system by running, well, Lint, or perhaps GCC. A system could be great at running that, but horrible at running mathematical computations, for instance. I suspect JSLint will be heavy on string manipulations, which from my recollection of REAL benchmark results, was one thing that IE was faster at than other browsers, while being slower at others. This is why a real benchmark does not just run a single application.
Second, what's the matter with making a JIT that works well with crap code? If crap code is heavily used in real-life Javascript (and let's just say, i think it is...), it should be included in a benchmark. If DOM is slow, then the DOM component on the benchmark will be low. This does not mean that DOM should be excluded so that IE can have fast results.
Optimizing for benchmarks is certainly a concern -- there've been cases in the past where companies have been caught going so far as SPECIFICALLY targetting benchmarks getting abnormally high results even when compared to the exact same application outside the benchmark. But, with Opera, Firefox, Chrome, IE, all targetting different benchmarks and the resultant "pissing contests" involving numerous different benchmarks, I don't know if this is a big problem.
Failure to understand
Re: "why is everyone so obsessed with Javascript performance? When do you ever sit there in front of your browser thinking "hmmm, this script is really taking a long time to add a table row and apply a CSS style" ?"
I have had significant speedups on my 1.3ghz Atom, going from Firefox 3.0 to 3.6, due to 3.0 slowing to a dead crawl on some of the bloated Javascript some pages had. It significantly improved my battery life too.
"You don't. No one does. Chasing JS performance is a waste of time, for the vast majority of users it is completely irrelevant."
This is the kind of attitude that ends up with the bloatware that plagues the lives of Windows users (both the Microsoft-included ones, and the horribly written apps that can be added later.) Some have the attitude that, if their application runs adequately as the only thing running on a single-user system, they are done. Traditional UNIX tihnking has ALWAYS accomodated a large range of system speeds, and accomodated the idea that there may well be multiple users on the system. Your app runs well enough on a XYZmhz system? Good, but see if you can make it faster, then the system can accomodate more simultaneous users. In addition, on a battery-powered device, improvements in performance SAVE BATTERY POWER.
@Lewis Mettler
Quick suggestion: Next time you're going to post something, give it a quick proof read over first, and ask yourself, "Do I sound like a raving maniac for writing this?". This will help you to avoid doing silly things like calling Microsoft a rapist in a public forum and stuff.
This Means Nothing To ME...
as IE will not run on my computer... nor will any other crap from Microsoft...
penguins for the win...
what about the obvious possibility?
that they TUNED IE to work well on that benchmark .. at the expense of all the benchmarks.
Billy
Sometimes in life, we may want to heap all of our woes on the back of a beast, and send it forth into the wilderness, hoping neither beast nor trouble will again find us fumbling with our noodles. While the goat known as "IE" (henceforth referred to as "Billy") may have been painted by its master with every wicked stripe, it has returned cleansed, after a time, and deserves at least a salt lick, if not a pat on the head.
The wickedness ascribed to this cloven-hooved mossback (ie: rape), was never its aim. One can never fortell what a semi-domesticated scapegoat will get into, but unrequited this and that could not have been its original goal. It must have been the heat.
Oh yeah, it also has little to do with Javascript benchmarks.
Bug has been fixed in Chrome
http://code.google.com/p/v8/issues/detail?id=1345
Waste of time
By the time IE10 comes out all the other browsers would have moved on and probably by several versions so the article is a complete waste of time in my books.
I to am suspicious of him testing IE10 and not the latest test builds of the other browsers.
This topic is closed for new posts.
