Recommended Posts

The latest nightly WebKit r31232 scores a 95/100 on Acid3, it is the same nightly number as the Windows version, so I would think the Windows version also scores a 95. The WebKit team made 3 or 4 blog posts in the last week, they were at 88/100 I think? and now they are almost to 100.

Somehow when I built webkitgtk-svn r31232 in Arch Linux and used it with Midori and Kazahakase, I got only 93/100 on Acid3. However when I built it on Windows and used it with Safari, it got 95/100.

So the Acid3 test result is not just dependent on the rendering/scripting engine, but also the platform and the browser shell? :shiftyninja:

The latest nightly WebKit r31232 scores a 95/100 on Acid3, it is the same nightly number as the Windows version, so I would think the Windows version also scores a 95. The WebKit team made 3 or 4 blog posts in the last week, they were at 88/100 I think? and now they are almost to 100.

Safari was also quite fast to pass the Acid 2 test, just 15 days after Acid2's release, so they are actually slower now that they have already spent more than 15 days and still not passed yet :p

On another note, I'm not sure whether Safari's focus on the Acid tests is a good thing or not. I mean, if they work specifically for it, it's actually quite easy to make a browser pass the Acid tests, but that doesn't necessarily mean better standards compliance. It's only meaningful to pass the Acid tests as a side product of better standards compliance, instead of passing the tests just for the sake of it.

The Acid3 test, like they describe it, is very hard to pass it, because it's like doing a whole exam for each test. There's a bunch of things to determine and in the end, it's only the final result that counts, as well as the time taken to do it. So it proves more that a browser complies to standards by passing the Acid 3 test. You couldn't say this about Acid 2 or even Acid 1.

And still, IMO, Safari is the browser that complies more with the standards. The only real problems I've encountered with it so far are :

1. Websites that say they are not compatible with Safari, because instead of excluding browsers, they made a selection. In these cases, IE and FF and sometimes Opera will work.

2. Websites who use ActiveX or some kind of weird coding.

3. Websites with obscure plugins.

I know it's not related to standards compliance, but I really don't think it's an issue for Safari. Obviously, there's always work to do for standards compliance because new tags appear every month or so at w3... but I bet they're better at that than FF, Opera and clearly, IE.

The Acid3 test, like they describe it, is very hard to pass it, because it's like doing a whole exam for each test. There's a bunch of things to determine and in the end, it's only the final result that counts, as well as the time taken to do it. So it proves more that a browser complies to standards by passing the Acid 3 test. You couldn't say this about Acid 2 or even Acid 1.

And still, IMO, Safari is the browser that complies more with the standards. The only real problems I've encountered with it so far are :

1. Websites that say they are not compatible with Safari, because instead of excluding browsers, they made a selection. In these cases, IE and FF and sometimes Opera will work.

2. Websites who use ActiveX or some kind of weird coding.

3. Websites with obscure plugins.

I know it's not related to standards compliance, but I really don't think it's an issue for Safari. Obviously, there's always work to do for standards compliance because new tags appear every month or so at w3... but I bet they're better at that than FF, Opera and clearly, IE.

Don't get me wrong, I LOVE KHTML/WebKit. IMO, for proprietary engine, Presto >>>> Trident, and for open source engine, KHTML/WebKit >>>> Gecko.

However the pace at which they pass the Acid tests is a bit... unnerving for me. I mean, Safari 3.04 beta = 40/100, Safari 3.1 = 75/100, and WebKit Nightly = 95/100. I wonder if those big score jumps are really due to works towards better web standards compliance, or due to some specific "optimizations" just for the test.

Albeit that's the common problem for sythetic tests, give me two weeks and I can write a Trident shell that passes the Acid 3 test, but not any more standards compliant than IE.

Albeit that's the common problem for sythetic tests, give me two weeks and I can write a Trident shell that passes the Acid 3 test, but not any more standards compliant than IE.

Go ahead :) Not that I doubt you, but...

Well, back on topic, I wonder why Webkit had three released builds just this day.

...

Albeit that's the common problem for sythetic tests, give me two weeks and I can write a Trident shell that passes the Acid 3 test, but not any more standards compliant than IE.

Time to put your money where your mouth is, In 2 weeks I want to see you get IE passing Acid3.

Go ahead :) Not that I doubt you, but...
Time to put your money where your mouth is, In 2 weeks I want to see you get IE passing Acid3.

I'd love to, unfortunately I'm currently working my tail off building two large websites at once and my boss is constantly driving me to get them done by May, so I don't have the two weeks' time to work on this. Of course if you can really put in some $$$ to my money then maybe I can get some time on this otherwise pointless "project" ;)

btw I never said I can get IE pass Acid3, I'm not the IE chief in M$, I said I can make a Trident shell to pass Acid3. It's quite easy, you just retrieve the Acid3 html into a string, change parts of it that Trident can't handle into what Trident can run to produce the same results, and then feed it into the mshtml or just use the WebBrowser control. Basically an "emulator" of sorts that translates Acid3 codes into Trident codes, I've done similar applications before. Currently I don't have time to work full-time on it, but I may make a C# shell to pass over 20/100 in Acid3 with Trident this weekend, IF I can get some free time this weekend, just to prove I'm not bluffing.

Of course if I want to take the easy way out, I can always just retrieve the html, and checks if it's the Acid3, and then feed the WebBrowser control some pre-made html that will produce the reference result. That should take less than two hours, but I'll refrain from doing that :p

PS: now I've taken a better look at the Acid3 source, it seems I've overestimated its complexity. It won't require even one week to make a Trident shell that translates Acid3 specific codes into "M$-compliant" ones if I can work on it fulltime. But for now, back to ASP.NET and Flex coding :pinch:

...

btw I never said I can get IE pass Acid3, I'm not the IE chief in M$, I said I can make a Trident shell to pass Acid3. It's quite easy, you just retrieve the Acid3 html into a string, change parts of it that Trident can't handle into what Trident can run to produce the same results, and then feed it into the mshtml or just use the WebBrowser control. Basically an "emulator" of sorts that translates Acid3 codes into Trident codes

...

So you'd break the test until it's in a state where it doesn't actually test anything?

So you'd break the test until it's in a state where it doesn't actually test anything?

It's not exactly "breaking the test", just translating the acid3 codes into what the IE can do to produce the proper results. Unless you think emulators like qemu and ZSNES "break" the Linux applications and SNES games to make them run in a "shell" in windows, it's not "breaking the test". Theoretically, a full "W3C emulator" can be built as a shell for Trident so every HTML will show correctly in this Trident shell as what they should show in a fully W3C compliant render engine, but that'd be a lot more work. To pass Acid3 test, only those acid3 specific situations needs to be handled, that's exactly what I mean by the limitation of synthetic tests ;)

Man, those are some nicely organized bookmarks.

My bookmarks end up looking like my desk, scattered without any order.

:laugh: Thanks....I've always got a half dozen or so on the desktop too that I've dragged there temporarily.

Man, those are some nicely organized bookmarks.

My bookmarks end up looking like my desk, scattered without any order.

Yes. That happens to me aswell. I cannot get myself to organize the tons of bookmarks I have.

I wish there was a program who would pick up words from bookmarks and sort it according to that!

This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
  • Posts

    • My issue is I can't access the forum on mobile if the site is set to Desktop mode on Vivaldi because it can't complete the Cloud flare am I a bot check! I know this is a Vivaldi issues as it has started happening on all cloud flare check sites, it's so annoying, I've reported it but no fix yet.
    • Are you going to do performance benchmarks comparing all states? I'd be interested in seeing that in the next "part".
    • My father still uses a programme written in dbase3. Still manages to work with a little help from dosbox. 
    • Microsoft hides these secret Windows 11 performance boost settings available on every PC by Sayan Sen Windows enthusiasts often look for ways to extract as much performance out of their systems as possible, and it's often the case that they try and do so while trying to minimize the heat and power consumption. This is especially relevant in the case of mobile Windows PCs since laptops and notebooks tend to get hot and management of that heat and power is harder in such a form factor. As such users often turn to techniques like under-volting which can be used to squeeze out the maximum capabilities of a chip while also maintaining lowered power levels. There are official apps from AMD and Intel with the likes of Ryzen Master and XTU (Extreme Tuning Utility). While these are quite handy, most enthusiasts probably prefer to dig into the BIOS and play around with settings there like Curve Optimizer on Ryzen, which lets users set various frequency-voltage scaling values. These are essentially called P-States. If you are not familiar with them, Processor Power Management is done through Advanced Configuration and Power Interface (ACPI) P-states and C-states. While P-states or performance pwoer states handle CPU voltage-frequency scaling, C-states deal with CPU sleep states so that some of the CPU functions, which are not necessary at that moment, can be disabled. The P-states and C-states work together to make the processor run more efficiently. It helps the OS and apps determine which cores can be parked and which should be boosted. Of course not every user is an enthusiast or knows the technicalities and integrities of how things like overclocking or undervolting work. Thankfully for them Windows itself offers something pretty cool, though it is hidden by default on all systems. By default, Windows only has two P-States, "Minimum Processor State" and "Maximum Processor State." However, this can be changed with a Registry trick to expand the options under a secret "Processor performance boost mode" dropdown. This essentially enables the HWP or hardware P-States available on a device, and these are not controlled just by the OS itself as the underlying hardware gets involved too. In total there are five Processor Performance Boost Mode profiles that control how Windows requests and allows CPU turbo/boost behavior under the different power policies. They are: Disabled: In this mode, processor boosting is effectively turned off. The CPU will avoid entering turbo or boost frequencies and instead operate closer to its base frequency ceiling. This can significantly reduce power consumption and heat output, but at the cost of reduced burst performance and responsiveness in short workloads. Enabled: This is the standard behavior where boost functionality is allowed under normal conditions. The processor can opportunistically increase frequency when workload demands it, balancing performance gains with power and thermal constraints as managed by the system. Aggressive: Aggressive mode favors performance more heavily, allowing the CPU to enter higher boost states more readily and sustain them longer. This should in theory improve responsiveness under bursty or heavy workloads but increases power draw and thermal output compared to the default enabled behavior. Efficient Enabled: This mode still allows boosting, but with a stronger bias toward energy efficiency. The system attempts to use boost more selectively, avoiding unnecessary frequency spikes when the performance gain is marginal. Efficient Aggressive: This is a hybrid approach where boost is still performance-responsive, but the system continuously weighs efficiency more heavily than in Aggressive mode. It aims to deliver noticeable performance improvements while reducing wasted power in less demanding scenarios. Here's how to enable the Processor performance boost mode: Open Registry Editor: Press Win+R, type regedit, and click OK. Go to: HKLM\SYSTEM\CurrentControlSet\Control\Power\PowerSettings\54533251-82be-4824-96c1-47b60b740d00\be337238-0d82-4146-a960-4f3749d470c7 (where HKLM stands for HKEY_LOCAL_MACHINE_) Modify the value of Attributes from 1 to 2 (you can find modify option by right-clicking) After that, exit Registry, you should now be able to see the new "Processor performance boost mode" dropdown menu: As you can see there are now five new P-States or CPPC states or power profile available that help define the boost mode processor setting on your PC. Wrapping it up here's a quick run-down of the settings as defined by Microsoft itself. Setting Description Disabled The corresponding P-state-based behaviour is disabled. Collaborative Processor Performance Control (CPPC) behaviour is disabled. Enabled The corresponding P-state-based behaviour is enabled. CPPC behaviour is Efficient Enabled. Aggressive The corresponding P-state-based behaviour is enabled. CPPC behaviour is Aggressive. Efficient Enabled The corresponding P-state-based behaviour is Efficient. CPPC behaviour is Efficient Enabled. Efficient Aggressive The corresponding P-state-based behaviour is Efficient. CPPC behaviour is Aggressive. Aggressive At Guaranteed Windows calculates the desired extra performance above the guaranteed performance level, and asks the processor to deliver that specific performance level. Efficient Aggressive At Guaranteed Windows always asks the processor to deliver the highest possible performance above the guaranteed performance level. In the next part we shall be comparing these settings to explore how much of a benefit or regression they can provide in terms of performance and power efficiency. If you decide to change the values on your system and are experiencing problems like crashes or an overheating PC, make sure to revert the steps back to the original state.
  • Recent Achievements

    • Week One Done
      rolfus earned a badge
      Week One Done
    • One Month Later
      Leroy Jethro Gibbs earned a badge
      One Month Later
    • Conversation Starter
      flexorcist earned a badge
      Conversation Starter
    • One Month Later
      AndreaB earned a badge
      One Month Later
    • One Month Later
      agatameier earned a badge
      One Month Later
  • Popular Contributors

    1. 1
      +primortal
      518
    2. 2
      +Edouard
      198
    3. 3
      PsYcHoKiLLa
      147
    4. 4
      ATLien_0
      93
    5. 5
      Steven P.
      78
  • Tell a friend

    Love Neowin? Tell a friend!