Can they really claim this?? Faster than Chrome?
http://www.apple.com/safari
Printable View
Can they really claim this?? Faster than Chrome?
http://www.apple.com/safari
Chrome and Safari are somewhat built around the same core, so it's not unlikely.
is Chrome really that fast?
Chrome is very fast, yes.
oh lol. i havent really tried it. I'm using Firefox.
Chrome = Firefox innards + Safari rendering engine + custom JS engine (V8)
Google basically said
"We rely on web apps, this is the sort of browser speed & functionality we're looking for, for the next-gen web apps"
The whole idea was that existing established browsers would take what Google had done, and reimplement it
Don't know too much about Safari
But see Firefox 3.1 alphas (minefield) for Mozilla's blazing fast JS engine & multi-process UI based on Google's work
Yes, it's possible Safari is faster than Chrome
that was the whole idea of Chrome
To jumpstart browser features
Niuce, can't wait to finish setting up my brand new macbook so that I can install it.
I looks just like Chrome, just a little better, however, after using for an hour we back to Chrome, doesn't appear faster on my PC...
Safari has never been terribly fast on PC. At work I use it occasionally, it crawls, it's slower than Firefox, but on my Mac it's screaming fast. Same framekwork and webkit, I don't get it. :?
Is Chrome considered stable now? I used it for a few weeks when it first launched but all I kept hearing was not to use it yet as it's too exploitable and unsafe.
I feel bad for anyone who has a compute ancient enough where browsers actually feel different performance-wise.
Though on small devices it could be of some use, but for desktops/laptops... come on.
blade5x, no matter how fantastic your computer is, different software will always perform differently. It's also higly relevant whether the browser uses a lot of power. Browser developers could easily make the fastest browser ever. Problem is, it requires 1GB free RAM, a 3GHz dual-core processor and a CUDA compatible GPU and operating system.
What browser developers aim for, is good performance with less hardware usage. I.E. FireFox has always been wellknown for stupid memory leaks, where its RAM usage would just slowly rise higher and higher. That's why browser performance is relevant, even if you have a good computer.
the main issue today is Javascript performance
Javascript is (inherently) single threaded
so to get any sort of performance out of JS heavy web-apps, it's now up to the browser to act as a little mini-OS and manage workloads between JS function calls
Wasn't the bug only happening in FireFox? I mean sure, maybe it wasn't doing that on Linux, but if it's doing it on Windows and no other browsers are, then it's a memory leak.
The problem was Windows's memory management
the issue specifically is memory fragmentation
When you load a webpage, it typically loads lots of images as well
these images are compressed (Jpeg, Gif, Png, etc.)
In order to use a webpage properly, these images have to be uncompressed, and sit in memory for the duration of the pages life
You have to store them uncompressed, else when you scroll a page (or any action that changes a page) the images would flicker as they are redrawn from their compressed states
A webpage can potentially take up many megs in memory due to large quantities of images being stored as uncompressed bitmaps
So, you have an initial "high memory usage" issue to start with
As you are navigating around a website, some images drop off (because they're only on one page) while some stick around (as they are present on all pages)
the various smilies and user avatars & sig pics are present on some pages, but not all
whereas the top banner on this site, and various style images are present on all pages
So, when you first view a page on a website, the memory is filled up with stuff
a bit solid block of used memory
then you move to a different page, and some memory is freed up as images drop off, while some cannot be as they're still needed
moving around a website will split up the memory block as new images may be too big to fit in the freed-up holes or the OS memory manager just chooses not to reuse previously freed memory
20 pages into the site, the memory block looks like a swiss cheese
there's a whole heap of "free'd" memory, but it's in small clumps dotted all over the place
An OS will report an application is "using" memory if it's been previously allocated but not taken back
Windows was reporting that Firefox was using more and more memory, purely because the Windows memory manager was not smart enough to deal with all these small scattered blocks of free memory
The OS did not reclaim the memory, so Firefox is still "using" it (even though it's been free'd)
All browsers on windows (as far as I know) suffered from this,
but Firefox seemed to be most affected due to the way it implimented it's page caching
I honestly don't know
(does anyone outside MS?)
Many *nix OS's use SLAB allocation, developed by Sun in 1995
http://en.wikipedia.org/wiki/Slab_allocation
SLAB combats memory fragmentation in a number of ways
Linux used an implimentation of Sun's SLAB allocator from 1999 - 2007
I know about 2 years ago, most Linux distros migrated over from the SLAB allocator to the newer SLUB allocator
This reduced memory fragmentation issues even further
http://lwn.net/Articles/229984/
Windows. No clue
but in any case, it's an OS issue not an application issue
I just installed it on my Macbook, it really is faster than chrome in all of my benchtest.
It does look suspiciously like Chrome though. I guess Apple decided that they've been ripped off so many times, it was time to do it to someone else :P
Apple has been ripping people off. With Ipods anyway, i dont know about their other stuff.
The benchmarks Apple showed are fake.
Safari 4 actually runs just a bit slower than Opera 9.6x on Windows, and just a bit faster than Firefox. Chrome is still the fastest browser.
comparing app speeds on windows is like comparing car top speeds on a road peppered with pot-holes and debris
and that's leaving aside the scientifically rigorous sample count of one...
hardly trolling, but whatever....