Main Area and Open Discussion > General Software Discussion

Maxthon (web browser) resource utilization...

<< < (2/6) > >>

Lashiec:
Remember that memory usage also depends of what pages you have opened. If the 15 tabs are pages with 300 photos each, there's no much you can do to lower memory usage using another browser. What it's important is memory usage in the long-term, that is, at the end of the day, that's when a browser shows it's a memory pig.

Curt:
I too tried to open a fav' folder of 21 links. Firefox was using between 30 and 60 % CPU with a constant total CPU usage of 100%, while Maxthon 1.6 was using between 10 and 50 % with a total of 80 - 90 %. Both had some peaks around 90%.

With all 21 pages open, Firefox was taking up 252 MB Virtual Size, 94 MB Working Set, and 115 MB Private Bytes. In the same discipline Maxthon 1.6 was taking up 280 MB Virtual Size, 25 MB Working Set, and 78 MB Private Bytes. So they are using the memory differently.

A strange part was when I closed 21 pages at the same time while watching Process Explorer: Maxthon 1.6 would still use 240 MB Virtual Size, 10 MB Working Set, and 56 MB Private Bytes - meaning the hard part on memory is the browser, not the homepages.


Sorry for the layout, I will not win an pedagogical prize for this post, I know.

Darwin:
Sorry for the layout, I will not win an pedagogical prize for this post, I know.
-Curt (February 17, 2008, 03:56 PM)
--- End quote ---

 :-* You're still a star in my eyes! Great post!

nontroppo:
A few points:

a) It is not only what you use, but how you use it that counts ;-) IE (which Maxthon is a shell for), is pretty conservative in memory allocation. But its navigation performance is horrid. Opera uses memory for fast history navigation more efficiently than any other browser on the planet. Using 3 tabs of google images, I can instantly navigate through 600 images worth of data in Opera where IE fails after a few[1]. The raw memory numbers are not too different, but functionality certainly is!

b) Using memory can result in much better performance. Though sadly not available anymore, Mark Russinovitch's great article about hoax memory optimizers was scathing on the idea of free RAM. Modern OS's work better when memory that you paid for is actually used. Worrying about memory allocation is largely futile, as performance is much better overall. I know at least Opera has done extensive work on balancing this. If you have lots of free RAM they will use it because it results in better overall performance. Opera's core works on mobile phones, so it can render in extremely limited conditions. Every byte is tweaked and optimized. On desktop, they simply extend cacheing of data to drive up speed.

c) The OS and the machine, not only the app, is responsible for memory allocation. At least for Opera, it uses less memory when there is less available (I've done multi-tab browsing of modern web pages using a P166 w/ 128MB RAM)...

d) I'm not really sure what measuring VM is telling you? A combination of working set and private bytes are much more informative IINM (process explorer terminology).

The benefits from moving away from IE and its shells from a technical perspective are numerous. IE/Maxthons engine is ancient and showing its age. It renders slowly, has poor support for modern technology and innumerable bugs that are constantly worked around. Lots of modern pages add in additional hacks to get IE & shells working, which will add to its overhead. Performance of IE and Maxthon are poorer in many areas than modern engines (though tables in IE are fast (popular 5 years ago for web layouts), CSS is poor (popular for newer pages).

So I *do* think there are clear advantages with regards to "effective" resource utilization to move away from IE and its shells, and other advantages too.


----
[1] I've done tests on history memory - using 10 pages of navigation through 3 google image searches in different pages for "picasso", "magritte" and "dali" (20 images per page = 600 images in 30 pages total). IE fails on page switching after a few back navigations, forcing a redraw and uses more memory than opera. Firefox has a 5 page limit (thus fails half of the instant-back operations) and uses the most memory. Opera allows instant-back for ALL pages using ~25MB less private bytes than IE and ~35MB less private bytes than Firefox. I find that pretty impressive...
--- End quote ---
http://my.opera.com/community/forums/findpost.pl?id=1789849

Darwin:
I'm not really that concerned about RAM usage - it's the hit to my CPU that is frustrating me. As an aside - the VM that I am quoting *seems* to be roughly analogous to Process Explorer's Private Bytes, which is why I quoted it. Doing so is force of habit, more than anything else because, as just mentioned, it's the CPU utilization that is causing me to break out into hives.

My initial post in this thread was seeking exactly this kind of comment, nontroppo, so thank you! I am very tied to IE/Maxthon because of some plugins that I use. These have analogues in Opera and FF so I'm just being lazy... (or at least stuck in my ways)! As noted above (ie in an earlier posting) Opera has already impressed me for its speed.

Navigation

[0] Message Index

[#] Next page

[*] Previous page