[IronPen]

Chrome often gets praise for using fewer resources than some other browsers, but I don't think many users are taking into account how much memory is being utilized by the browser. Besides the Chrome.app, you'll notice that there several other processes. Upon launch it runs one renderer process and two worker processes. Total RAM usage is about 150MB real RAM and double that in virtual RAM. That's not as efficient as a lot of other browsers.

[mduell]

1) That's one of the tradeoffs of their process architecture that enhances security/stability.

2) What's the private memory usage?

3) What else were you going to do with that $1.50 worth of RAM?

[Evoken]

That kind of memory usage is what makes Chrome so fast and stable. Yes, it can eat up a lot of ram with all the processes...but it puts it to good use.

[besson3c]

Where are you getting that 150 MB number anyway, just out of curiosity?

I ask because a lot of people don't really understanding how to translate these sorts of numbers into meaningful information.

[IronPen]

Just making a point that other browsers use a lot fewer resources. Safari has had a known memory leak since inception. Firefox immediately consumes mass amounts of RAM upon launch and refuses to release it as well. Each often gets criticized for this consumption, yet the amount they use isn't too far from what I see Chrome using at launch. As for the math, I simply opened Activity Monitor and added up the numbers.

[besson3c]

Originally Posted by IronPen 

Just making a point that other browsers use a lot fewer resources. Safari has had a known memory leak since inception. Firefox immediately consumes mass amounts of RAM upon launch and refuses to release it as well. Each often gets criticized for this consumption, yet the amount they use isn't too far from what I see Chrome using at launch. As for the math, I simply opened Activity Monitor and added up the numbers.

But, with all due respect, all of this is meaningless...

Consuming more RAM that is available is not bad, nor is it worse than other products that use less. RAM is a useful tool for caching things and speeding up requests and can be applied in an optional "use it if it is available" sort of fashion.

[pakirk]

I notice a gradual increase in the memory footprint of Chrome 10.0 due entirely to increasing memory taken up by the Flash plugin.

[besson3c]

Does it decrease when you block Flash?

[pakirk]

Seems to but I didn't test extensively since not having flash limits the usability of the browser some. What I see using the Activity Monitor is a steady increase in "Shockwave Flash (Chrome plugin host)", starting at 42 MB and going over 100MB over time. It never stopped increasing that I saw.

[IronPen]

Originally Posted by Evoken 

That kind of memory usage is what makes Chrome so fast and stable. Yes, it can eat up a lot of ram with all the processes...but it puts it to good use.

And where is your proof? I'm not mocking you, I just want to read the same material you read to come to this conclusion. More memory usage does not mean faster and more stable.

[IronPen]

Originally Posted by besson3c 

But, with all due respect, all of this is meaningless...

Consuming more RAM that is available is not bad, nor is it worse than other products that use less. RAM is a useful tool for caching things and speeding up requests and can be applied in an optional "use it if it is available" sort of fashion.

You made the same argument when I brought this up before. But his argument is the SAME as mine. Purely based on comparison, at launch, Chrome will consume more memory than some of the other browsers. Of all the browsers I have used, Camino is the only one that I have seen release memory as tabs or windows are closed. Just my observation.

As to the sarcastic posts I see on forums, and not just this one, that say, RAM is cheap, buy more. Well, the point is, I shouldn't need to keep adding RAM just to run a web browser. I can run other browsers that consume less memory at startup and take longer to get to the consumption that Chrome uses from launch. It's a good question, and one to which I have not heard a good answer.

How a computer manages RAM is a WHOLE other topic. Each OS has it's own kernel that does it's own magic to allocate RAM as it sees fit. Unused RAM is wasted RAM. There is file cachine at work and other management done by the kernel to increase performance of the system. I don't fully understand it and I doubt anyone here does either. However, that's not the question. The question is a good one and just asks, why does Chrome use more? Perhaps like the kernel, it is all to increase performance. I don't understand why there are two workers. What do they do? I have no clue. Answering these questions would be more helpful than giving an aswering in dismissal of the originators inquiry.

[besson3c]

I don't mean to sound dismissive or sarcastic or anything negative, but just to be purely practical about it there is no real way that anybody can answer that. Perhaps some browsers deal with plugins differently preloading vs. loading them on demand, perhaps some desire more for HTML5 storage, or for preloading HTML5 video/audio codecs, or their rendering engines claim different amounts of RAM, or...

There are a gazillion variables. At the end of the day, the comparisons are probably not terribly useful. What is useful is comparing consumption and performance having a large number of the same pages open so that you can compare how the browsers scale. That would be a fairly interesting comparison, and in my opinion, this is where Chrome absolutely shines.

[IronPen]

@besson3c: Those are valid arguments. Take Flash, for example. Camino doesn't run it in a separate process, like Safari. So the Safari app itself won't look like it is being a CPU hog while running the video, but the subprocess that Flash runs in, will. I'm not sure how Chrome does it, but I know that Flash comes packaged with the browser. Plugins and such are a mixed bag. I would consider that a separate issue. But I understand your point, wanting to compare Safari with extensions to Chrome with extensions. One thing we need to be mindful of is that although Chrome uses Webkit as its renderer, a lot of work has been done with Chrome that makes it very unique from Safari or other Webkit-based browsers. That's not necessarily a bad thing but I have seen arguments saying, "it's just Safari with tabs on top" or "it's just Safari with..." and that's not a fair assessment.

I think this is a good discussion and would like to see more technical discussion on the matter of Chrome's RAM consumption, the purpose of the renderer vs. workers and helpers and what they each really do. And I'd like a clearer explanation of Chrome's data tracking habits. I keep hearing that the ID can be tracked back to the user personally. Firefox and other browsers track browsing habits, but they do not link that activity directly back to the user. It's clearly stated in their privacy documentation. I haven't seen similar documentation from Google, which prevents me from making Chrome my default browser. I have nothing to hide, but I don't want to support that sort of behavior.

[mduell]

Originally Posted by IronPen 

And where is your proof? I'm not mocking you, I just want to read the same material you read to come to this conclusion. More memory usage does not mean faster and more stable.

You've inverted cause and effect. Because Chrome separates each tab into it's own process, sandboxes plugins, etc it uses more memory as a result.

See the security and stability sections of the wikipedia article.

[besson3c]

Originally Posted by IronPen 

@besson3c: Those are valid arguments. Take Flash, for example. Camino doesn't run it in a separate process, like Safari. So the Safari app itself won't look like it is being a CPU hog while running the video, but the subprocess that Flash runs in, will. I'm not sure how Chrome does it, but I know that Flash comes packaged with the browser. Plugins and such are a mixed bag. I would consider that a separate issue. But I understand your point, wanting to compare Safari with extensions to Chrome with extensions.

I'm not sure where I made that point?

Safari runs browser plugins as processes so that they are sandboxed and do not affect the entire browser when they crash. Chrome takes this a step further and runs each tab as a process which is a model that has worked so well that Apple has copied that with Webkit2. The way that resources are managed at the kernel level is very complex, which is why it is nearly impossible to make direct comparisons from browser to browser simply based on their designs (as opposed to end results).

My point was that really all we can do is measure end-result performance particularly with a large number of tabs open simultaneously.

I think this is a good discussion and would like to see more technical discussion on the matter of Chrome's RAM consumption, the purpose of the renderer vs. workers and helpers and what they each really do. And I'd like a clearer explanation of Chrome's data tracking habits. I keep hearing that the ID can be tracked back to the user personally. Firefox and other browsers track browsing habits, but they do not link that activity directly back to the user. It's clearly stated in their privacy documentation. I haven't seen similar documentation from Google, which prevents me from making Chrome my default browser. I have nothing to hide, but I don't want to support that sort of behavior.

I think this is a myth.

It is too expensive to be phoning home constantly, Chrome wouldn't perform as well as it did if it did this, and Google has other analytics and means to determine how often people in general visit certain sites. User agents strings are a part of how each browser identifies itself, so Google could in theory track that its own browser was used for accessing certain sites accessed through their search engine, and I suppose it is possible for some sort of browser download ID to also be attached to the user agent string so that Google is aware that this is you, but this is pure speculation.

I personally don't think that knowing your browsing habits on a personal level is really of interest to Google though. Google is after trends. If you access MacNN frequently Google doesn't need to know that you are IronPen, have a wife and 2 kids, and live in Wisconsin (stuff I just made up), because what is more interesting to them is that you are probably a male under 40 years of age that is interested in technology, probably live in a city, etc. This demographic information can be determined simply based on what we know about the demographic of MacNN based on its content. Google can manage to escape the stigma of infringing on privacy by accomplishing their same goals this way, because there is nothing wrong with simply studying a website and its content and making assumptions about its demographic.

What is going to be interesting is when the world switches to IPv6, and when we can track visits back to your unique static IP address. This is going to introduce all sorts of privacy considerations, especially at the ISP level.

[IronPen]

Originally Posted by besson3c 

I'm not sure where I made that point?

Safari runs browser plugins as processes so that they are sandboxed and do not affect the entire browser when they crash. Chrome takes this a step further and runs each tab as a process which is a model that has worked so well that Apple has copied that with Webkit2. The way that resources are managed at the kernel level is very complex, which is why it is nearly impossible to make direct comparisons from browser to browser simply based on their designs (as opposed to end results).

My point was that really all we can do is measure end-result performance particularly with a large number of tabs open simultaneously.

Very true. I like that Chrome has separated each tab into its own process. And it would seem to address issues, to some degree, that many browsers have with caching and increased RAM consumption over time. Since the resources of that tab can be given back when it is closed, you don't end up with all the RAM being consumed by the single browser process.

The one problem I have seen with some browsers, Camino and Safari, because they are my #1 and #2 browsers, is that after they consume lots of memory, due to Flash or whatever, they are more crash prone.

Originally Posted by besson3c 

I think this is a myth.

It is too expensive to be phoning home constantly, Chrome wouldn't perform as well as it did if it did this, and Google has other analytics and means to determine how often people in general visit certain sites. User agents strings are a part of how each browser identifies itself, so Google could in theory track that its own browser was used for accessing certain sites accessed through their search engine, and I suppose it is possible for some sort of browser download ID to also be attached to the user agent string so that Google is aware that this is you, but this is pure speculation.

I personally don't think that knowing your browsing habits on a personal level is really of interest to Google though. Google is after trends. If you access MacNN frequently Google doesn't need to know that you are IronPen, have a wife and 2 kids, and live in Wisconsin (stuff I just made up), because what is more interesting to them is that you are probably a male under 40 years of age that is interested in technology, probably live in a city, etc. This demographic information can be determined simply based on what we know about the demographic of MacNN based on its content. Google can manage to escape the stigma of infringing on privacy by accomplishing their same goals this way, because there is nothing wrong with simply studying a website and its content and making assumptions about its demographic.

What is going to be interesting is when the world switches to IPv6, and when we can track visits back to your unique static IP address. This is going to introduce all sorts of privacy considerations, especially at the ISP level.

I don't know that it's a myth. I just need to spend more time researching it myself, but I have seen numerous posts, which doesn't necessarily mean it's true, stating that there is an ID that CAN be traced back to you. Whereas Firefox tracks browsing habits, but data cannot be traced back to you personally. Google isn't obtaining personal information about me, just browsing habits. They use this data to determine what ads to show you, and who knows how else. There is enough concern to cause me to hesitate to use Chrome on a regular basis. Obviously, I bear the responsibility of investigating this. But I don't believe Google makes it very clear. Flock, the browser based on Chromium, doesn't make this clear in their documentation either. Firefox clearly does. Again, I'm not a proponent of Firefox and I'm not necessarily an opponent of Chrome. I'm just stating what I know. Once I have time to read more on this topic for myself I will come back to this thread.

[IronPen]

So I read several pages of content from Google regarding their privacy policies and data collected by Chrome. The ID that I had heard about which can be tied back to the user was not mentioned anywhere. So either it is rumor, or it is is something they don't disclose, or I just missed it. But I'm going to side with besson3c at this point and say, it's rumor. When you think about it, your IP address, which is easily obtainable, can be traced back to you. As I said, I'm not trying to hide anything, but if I have a preference to use a browser that clearly states within their privacy policy that it won't identify you personally, versus one that isn't so clear...I know which browser or software I will choose. Good discussion. If someone else has more to add, please contribute.

[besson3c]

I think the best argument against the rumor for whomever believes that it is true is simply to analyze the costs of that data tracking vs. the gains, especially when similar information is already at Google's disposal for the duration of your IP lease.