Did you know: All GDI apps render slower under Win7?


Recommended Posts

That would imply that no single driver release for the past year, either from NVIDIA or AMD, have proper ION or Radeon 4250M support for Windows 7....

Not implying anything of the sort. What I am implying is that sometimes drivers are buggy. It happens. This system that I'm currently sitting at is using an old ATI HD 3850 AGP, and last month's drivers caused a couple issues, rolling back to the version from the month before took care of it. Hence.. "driver issues".

ahhh this is why windows just didnt feel fluid enough. tried going back to winxp and i can confirm this result. IPS screens additionally produce some input lag combined with this it feels we went behind in terms of fluidity of gui experience over last 7-8 years.. not to flare things up but damn that was a total waste of time th last decade trying out different windows versions, upgrading to IPS screens etc when everything works great with old tech.!!!

quoting from the OP "On the other hand, not using DWM will free up a lot of CPU resources. But still not enough to make GDI render operations as fast and responsive as Windows XP, as XP has more CPU free to do other tasks."

ok got my answer :)

Lol what? That is from a Windows 7 WinHEC presentation which simply says Direct2D will replace GDI for new apps which it has already. Doesn't mean all GDI apps will transform themselves into Direct2D apps without developers rewriting them. GDI is deprecated as of Windows 7 (2009) because WPF was a managed code platform meaning all native code apps written from the beginning of Windows to 2009 and the dozens being written today still use GDI.

Sorry, try again.

Since GDI is an API, the back-end is not important to developers using it. If it was migrated to be a thin layer above DirectX, all existing GDI apps would "just work", just like all GDI apps written for XP "just work" when ran on Vista/7 even though the GDI isn't hardware-accelerated any more. Look at the patent I mentioned, it's pretty explicit : redirecting calls from one API to another one, in the context of graphics rendering.

Also, WPF has nothing to do with this. The successor of GDI is DirectX and its new set of APIs introduced in Win7 and backported to Vista such as Direct2D and DirectWrite, WPF is just a managed framework to develop hardware-accelerated apps.

The idea of abandonning GDI is not new ; in your beloved XP, Windows Explorer's right pane (the one with collapsible regions showing possible actions) is done using DirectUI, an undocumented proprietary framework from Microsoft (which uses XML files, making it a WPF predecessor for native code) that is hardware-accelerated and doesn't use the GDI at all.

i dont care about whats going on underneath. But i always thought this white black mess when we resize was present in every windows to date :p

i didnt realise it was not in XP.

Now i think about it. Was the design team sleeping? First those OLD icons... they bug me to death. And also this.

I never actually see this unless i drag my explorer window and there u can see a black and a white box next to the search bar. I mentioned it before somewhere. And its really annoying. I dont understand why is this not fixed. As it was there in Vista. And i havent tried to see if its there in Win8. Can anybody test?

Since GDI is an API, the back-end is not important to developers using it. If it was migrated to be a thin layer above DirectX, all existing GDI apps would "just work", just like all GDI apps written for XP "just work" when ran on Vista/7 even though the GDI isn't hardware-accelerated any more. Look at the patent I mentioned, it's pretty explicit : redirecting calls from one API to another one, in the context of graphics rendering.

Also, WPF has nothing to do with this. The successor of GDI is DirectX and its new set of APIs introduced in Win7 and backported to Vista such as Direct2D and DirectWrite, WPF is just a managed framework to develop hardware-accelerated apps.

The idea of abandonning GDI is not new ; in your beloved XP, Windows Explorer's right pane (the one with collapsible regions showing possible actions) is done using DirectUI, an undocumented proprietary framework from Microsoft (which uses XML files, making it a WPF predecessor for native code) that is hardware-accelerated and doesn't use the GDI at all.

Yeah but what's the use of DirectUI if its not publicly documented? There's also Milcore (the undocumented native code rendered for WPF). I mentioned WPF because Microsoft's original vision was to replace GDI with the hardware accelerated Avalon stack and it pretty much does everything Direct2D does and more but being managed code, it didn't see and will never see as much uptake by developers. So MS basically introduced a version of Windows after XP without a proper GDI replacement. It was only in Windows 7 that Direct2D was born and it's only been 2 years after that. Developers aren't going to start Direct2D apps overnight abandoning all their customers on XP. There are like 99% GDI apps and 1% Direct2D apps. GDI is going to be around forever for backward compatibility and there are always going to be more GDI apps than Direct2D apps for a long time which is why Microsoft should take the effort to accelerate all of the GDI functions in WDDM drivers. It will be a very long time before we see apps that are all using Direct2D. And Explorer's left pane uses DirectUI (the Task Pane), not the right pane.

The most important operations are GPU accelerated, but if performance is more important, then they could use Direct2D/Direct3D on Vista/7, and Direct3D on old OSs like XP or 2K (e.g. Firefox renders all web page content on the CPU, and composites it on the GPU on 2K and XP)

GDI is a deprecated technology, so MS isn't going to spend time improving it for old applications.

Yeah but what's the use of DirectUI if its not publicly documented? There's also Milcore (the undocumented native code rendered for WPF). I mentioned WPF because Microsoft's original vision was to replace GDI with the hardware accelerated Avalon stack and it pretty much does everything Direct2D does and more but being managed code, it didn't see and will never see as much uptake by developers. So MS basically introduced a version of Windows after XP without a proper GDI replacement. It was only in Windows 7 that Direct2D was born and it's only been 2 years after that. Developers aren't going to start Direct2D apps overnight abandoning all their customers on XP. There are like 99% GDI apps and 1% Direct2D apps. GDI is going to be around forever for backward compatibility and there are always going to be more GDI apps than Direct2D apps for a long time which is why Microsoft should take the effort to accelerate all of the GDI functions in WDDM drivers. It will be a very long time before we see apps that are all using Direct2D. And Explorer's left pane uses DirectUI (the Task Pane), not the right pane.

For the last time: GDI can and will be rewritten as a layer above DirectX, thus hardware-accelerating all existing GDI apps without the need to create a new version of WDDM whose point would be backward-compatibility.

Why can't you understand that? You're asking Microsoft to do something they already plan to do, in a way that would be much worse (adding backwards compatibility to WDDM...).

Microsoft's plans for the future include moving to a fully managed OS (Midori), and using virtualization to run older apps. Which implies having the least possible amount of code to virtualize, which implies rewriting old stuff as thin, lightweight layers above the new stuff instead of keeping and maintaining old code.

(PS: Yes, I mean the left pane, not the right one, sorry)

For the last time: GDI can and will be rewritten as a layer above DirectX, thus hardware-accelerating all existing GDI apps without the need to create a new version of WDDM whose point would be backward-compatibility.

Why can't you understand that? You're asking Microsoft to do something they already plan to do, in a way that would be much worse (adding backwards compatibility to WDDM...).

Microsoft's plans for the future include moving to a fully managed OS (Midori), and using virtualization to run older apps. Which implies having the least possible amount of code to virtualize, which implies rewriting old stuff as thin, lightweight layers above the new stuff instead of keeping and maintaining old code.

(PS: Yes, I mean the left pane, not the right one, sorry)

please drop me a line when its done, right now going back to xp feels like a breath of fresh air. seriously drop me a pm when this gets done win7 is great too but ill be on xp for the while. cheers.

For the last time: GDI can and will be rewritten as a layer above DirectX, thus hardware-accelerating all existing GDI apps without the need to create a new version of WDDM whose point would be backward-compatibility.

Why can't you understand that? You're asking Microsoft to do something they already plan to do, in a way that would be much worse (adding backwards compatibility to WDDM...).

Microsoft's plans for the future include moving to a fully managed OS (Midori), and using virtualization to run older apps. Which implies having the least possible amount of code to virtualize, which implies rewriting old stuff as thin, lightweight layers above the new stuff instead of keeping and maintaining old code.

(PS: Yes, I mean the left pane, not the right one, sorry)

Question to ask is why hasn't this been done in 7?

Second question is will this be in Windows 8?

Assuming good design, testing, and optimization, this should speed up my netbook by a large constant factor.

For the last time: GDI can and will be rewritten as a layer above DirectX, thus hardware-accelerating all existing GDI apps without the need to create a new version of WDDM whose point would be backward-compatibility.

Why can't you understand that? You're asking Microsoft to do something they already plan to do, in a way that would be much worse (adding backwards compatibility to WDDM...).

Microsoft's plans for the future include moving to a fully managed OS (Midori), and using virtualization to run older apps. Which implies having the least possible amount of code to virtualize, which implies rewriting old stuff as thin, lightweight layers above the new stuff instead of keeping and maintaining old code.

(PS: Yes, I mean the left pane, not the right one, sorry)

There is nothing that says in the patent or PPT slide that clearly says it applies to versions of Windows later than Windows 7, in fact the patent and the PPT are from 2009. Neither has MS officially written anything on this, MSDN or elsewhere. Why can't you understand that unless you work on the graphics team in Windows and speak for Microsoft, saying "it will return" is only wishful thinking. I will believe it when I see it working in a future release. Information released at //Build/ about WDDM 1.2 doesn't mention it either though I would be glad to have been proven wrong if MS surprises by announcing full GDI acceleration is returning in Windows 8. Future/planned concepts does not always end up in shipping products.

I remember in the windows 7 beta, they talked a lot (internally) about GDI calls being redirected in DWM as an accelerated function of Direct2D... not everything was transitioned to this but a lot of it was, we had performance tests and everything to test out to make sure it was working at full speed

Making GDI run on top of Direct2D probably wouldn't work that well, it's an really old API and probably has a bunch of corner cases were the behavior differs from Direct2D, and implementing code to special case those issues would slow any re-implementation down.

But again, GDI is old, why would MS spend time fixing and enhancing it, when they've replaced with with Direct2D?

Making GDI run on top of Direct2D probably wouldn't work that well, it's an really old API and probably has a bunch of corner cases were the behavior differs from Direct2D, and implementing code to special case those issues would slow any re-implementation down.

But again, GDI is old, why would MS spend time fixing and enhancing it, when they've replaced with with Direct2D?

because even to this day people are still writing code in GDI and GDI+, until I can go into .NET and say make a new brush and a rectangle with a texture fill brush in Driect2D without having ti reference Direct2D people will use GDI+ you have to write wrappers that force all those calls through Direct2D almost every windows control is written in GDI or GDI+ very few are written in Direct2D even to this day

Question to ask is why hasn't this been done in 7?

Second question is will this be in Windows 8?

Assuming good design, testing, and optimization, this should speed up my netbook by a large constant factor.

That's a good question. I don't know the answer, of course, but I think we can assume that all of these "why don't they optimize feature X" questions have the same answer - they had more important stuff to do.

Seriously, when was the last time you noticed a big slowdown because of GDI being software-powered?

@neufuse: Are you talking about WPF or WinForms? WinForms is a layer above GDI, but you should be able to do this kind of stuff in WPF.

because even to this day people are still writing code in GDI and GDI+, until I can go into .NET and say make a new brush and a rectangle with a texture fill brush in Driect2D without having ti reference Direct2D people will use GDI+ you have to write wrappers that force all those calls through Direct2D almost every windows control is written in GDI or GDI+ very few are written in Direct2D even to this day

Or, you could just use WPF with NET, seeing as it's visual stack uses DirectX, and all DIrect2D does is map itself on to DirectX / Direct3D anyway :p Unfortunately, it shows that it's building on DirectX - the amount of code you need just to get a Direct2D Windows up and running is nearly the same as a DirectX / Direct3D Window - though Microsoft do seem to be pushing .NET more anyway, and that's the life of trying to code native Windows applications. Hassle.

Or, you could just use WPF with NET, seeing as it's visual stack uses DirectX, and all DIrect2D does is map itself on to DirectX / Direct3D anyway :p Unfortunately, it shows that it's building on DirectX - the amount of code you need just to get a Direct2D Windows up and running is nearly the same as a DirectX / Direct3D Window - though Microsoft do seem to be pushing .NET more anyway, and that's the life of trying to code native Windows applications. Hassle.

you COULD use WPF, but that still leaves all the MFC coders, the Win32 coders out in land of generally using GDI... and .NET coders generally use the forms based development approach not WPF, until MS says forms is dead and WPF is the standard this isn't going to change... we still use forms where I work, because its what we've done for ever, and when you work in the IDE the forms approach just seems easier, WPF has a lot of stuff that needs worked on still in the IDE to make it perfect for visual development

@neufuse: Are you talking about WPF or WinForms? WinForms is a layer above GDI, but you should be able to do this kind of stuff in WPF.

Forms, yes.. the most commonly used development method, WPF is slowly growing but its no where near forms, and doesn't have as many 3rd party controls premade for it either... although Devexpress is making some inroads on that though

you COULD use WPF, but that still leaves all the MFC coders, the Win32 coders out in land of generally using GDI... and .NET coders generally use the forms based development approach not WPF, until MS says forms is dead and WPF is the standard this isn't going to change... we still use forms where I work, because its what we've done for ever, and when you work in the IDE the forms approach just seems easier, WPF has a lot of stuff that needs worked on still in the IDE to make it perfect for visual development

Using Visual Studio to create WPF interfaces is a pain for sure... using Microsoft Expression Blend though and it's much closer to a joy :p Shame they don't appear to be doing much too update WPF though. With all these massive performance enchancements they do to browser engines these days, you'd think they'd go in and try to do something similar with the .NET framework. Aww well.

I dunno, I've had a pretty easy time with it in C++.

That's because it's written in C++ lol. Why do you think I said C, and not C/C++. Using any of the Direct* interfaces in C is horrible because they all use COM.

Very well documented on MSDN. Plus you can mix and match GDI with Direct2D, also well documented on MSDN. Of course there's a learning curve, always is with the new and improved API's.

I'm sorry but nothing will ever get me to use COM or the Direct* interfaces. At least the GDI and Win32 interfaces are C friendly, COM and Direct2D by relation most certainly aren't.

because even to this day people are still writing code in GDI and GDI+, until I can go into .NET and say make a new brush and a rectangle with a texture fill brush in Driect2D without having ti reference Direct2D people will use GDI+ you have to write wrappers that force all those calls through Direct2D almost every windows control is written in GDI or GDI+ very few are written in Direct2D even to this day

GDI+ isn't a layer on top of GDI or such, it's a different engine entirely (GDI+ runs entirely in software, which is why you can use it on 9x and such). But even then you still have to reference GDI+ to use it in .NET apps (Through System.Drawing)

And while the normal Windows controls use GDI, that doesn't mean programs have to (Look at Firefox, the entire UI is drawn through Direct3D), and eople who care about performance aren't going to be animating buttons and text boxes (and those who do care about performance and are still using GDI for some reason, would most likely be using Direct3D or such to composite it to save re-rendering)

That's because it's written in C++ lol. Why do you think I said C, and not C/C++. Using any of the Direct* interfaces in C is horrible because they all use COM. I'm sorry but nothing will ever get me to use COM or the Direct* interfaces. At least the GDI and Win32 interfaces are C friendly, COM and Direct2D by relation most certainly aren't.

So you're refusing to use the official API's provided to access these systems, never mind ignoring the tools that the platform was written in to begin with, then going to cry about how hard it is to use? Yea, that makes sense. :argh:

So you're refusing to use the official API's provided to access these systems, never mind ignoring the tools that the platform was written in to begin with, then going to cry about how hard it is to use? Yea, that makes sense. :argh:

What?? GDI and the Win32 API's are official. The Direct*D API's are not a requirement to build Windows GUI's. You do know that right?

This topic is now closed to further replies.
  • Posts

    • Dude, im talking about simply disable it from settings app. Because of the eu regulation, you could disable it here for years.
    • One big question about Mars was answered thanks to Einstein's 100 year old theory by Sayan Sen Image via DepositPhotos Scientists at the U.S. National Institute of Standards and Technology (NIST) have calculated how time passes on Mars compared with Earth, adding detail to how timekeeping would need to work beyond Earth’s orbit. The study, published in The Astronomical Journal, found that clocks on Mars run an average of 477 microseconds, or millionths of a second, faster per day than clocks on Earth. A microsecond is one millionth of a second, a very small unit used in precise scientific timing systems such as atomic clocks, which measure time using consistent atomic behavior. This difference is not constant. Because Mars moves around the Sun in a non-circular path (an eccentric orbit, meaning its distance from the Sun changes over time instead of staying fixed) and is affected by gravity from other bodies, the daily difference can vary by as much as 226 microseconds over a Martian year. The study also identifies smaller repeating changes of about 40 microseconds per day linked to synodic cycles (repeating periods that describe how planets line up with each other as they orbit the Sun from different positions). These longer patterns affect how time differences slowly rise and fall. To make these estimates, researchers compared Mars with Earth and the Moon. The work looks at relativistic proper time (the time actually measured by a clock depending on its speed and the strength of gravity where it is located, as described in Einstein’s relativity). This shows that each world has its own slightly different “rate” of time. This becomes more important as space missions expand into cislunar space (the region between Earth and the Moon) and toward Mars. On Earth, time systems rely on atomic clocks and satellites, which stay closely synchronized for navigation and communication. The study is based on Albert Einstein’s theory of relativity, which shows that time is affected by gravity and motion. Stronger gravity makes clocks run slower, while weaker gravity makes them run faster. “The time is just right for the Moon and Mars,” said NIST physicist Bijunath Patla. “This is the closest we have been to realizing the science fiction vision of expanding across the solar system.” A day on Mars is about 40 minutes longer than on Earth, and a Martian year lasts 687 Earth days. But the main question is not just about days and years, but how fast time itself passes. An atomic clock placed on Mars would function normally, but compared with one on Earth, the two would slowly drift apart due to differences in gravity and motion. This requires careful calculation of what is similar to a time-zone difference across planets. Researchers modeled Mars using a reference surface and included gravitational effects from the Sun, Earth, the Moon, and other planets. This includes a multi-body gravitational system (often described as a three-body or four-body problem, where predicting motion becomes difficult because multiple large objects all pull on each other at the same time through gravity). Mars also follows a Keplerian orbit (an idealized elliptical orbit based on simple gravitational laws that assume smooth motion, before adding real-world disturbances from other bodies). In addition, the researchers accounted for solar tides (small changes in gravitational force caused by the Sun that slightly distort planetary motion and timing, especially in systems involving Earth and the Moon). These combined effects are described as relativistic proper-time offsets (small but measurable differences in elapsed time between locations caused by gravity and motion), which must be included when comparing clocks across planets. “But for Mars, that’s not the case. Its distance from the Sun and its eccentric orbit make the variations in time larger. A three-body problem is extremely complicated. Now we’re dealing with four: the Sun, Earth, the Moon and Mars,” Patla explained. “The heavy lifting was more challenging than I initially thought.” Although the differences are extremely small, they matter for navigation and communication systems that depend on precise timing. Even modern networks on Earth, such as mobile systems, rely on timing accuracy at very small fractions of a second. Communication between Earth and Mars currently takes about four to 24 minutes or more depending on planetary positions, meaning signals are not real-time. A shared and accurate time system could help future missions reduce confusion in navigation and data exchange. “If you get synchronization, it will be almost like real-time communication without any loss of information. You don’t have to wait to see what happens,” Patla said. Researchers note that fully developed interplanetary communication networks are still far in the future. However, understanding how time behaves across planets helps prepare for those systems. “It may be decades before the surface of Mars is covered by the tracks of wandering rovers, but it is useful now to study the issues involved in establishing navigation systems on other planets and moons,” said Neil Ashby. “Like current global navigation systems like GPS, these systems will depend on accurate clocks, and the effects on clock rates can be analyzed with the help of Einstein’s general theory of relativity.” Patla added that the results also help improve understanding of time itself under relativity. “It's good to know for the first time what is happening on Mars timewise. Nobody knew that before. It improves our knowledge of the theory itself, the theory of how clocks tick and relativity,” he said. Source: NIST, IOPscience This article was generated with some help from AI and reviewed by an editor. Under Section 107 of the Copyright Act 1976, this material is used for the purpose of news reporting. Fair use is a use permitted by copyright statute that might otherwise be infringing.
    • TeraCopy 4.0 Build 26 by Razvan Serea TeraCopy is a compact program designed to copy and move files at the maximum possible speed, also providing you with a lot of features. Copy files faster. TeraCopy uses dynamically adjusted buffers to reduce seek times. Asynchronous copy speeds up file transfer between two physical hard drives. Pause and resume transfers. Pause copy process at any time to free up system resources and continue with a single click. Error recovery. In case of copy error, TeraCopy will try several times and in the worse case just skips the file, not terminating the entire transfer. Interactive file list. TeraCopy shows failed file transfers and lets you fix the problem and recopy only problem files. Shell integration. TeraCopy can completely replace Explorer copy and move functions, allowing you work with files as usual. TeraCopy is free for non-commercial use only. For commercial use you need to buy a license. The paid version of the program includes the following features: Copy/move to your favorite folders. Save reports as HTML and CSV files. Select files with the same extension/folder. Remove the selected files from the copy queue. TeraCopy 4.0 Build 26 changelog: Added support for receiving files via the LocalSend protocol. Improved exception handling and automated bug report upload. Fixed several minor bugs and small memory leaks. Build 26 (June 24) Fixed a rare exception when a transfer completed. Features added since version 3.17: Enhanced speed graph. New multi-threaded copy engine. Support for copying to multiple targets. Queue system for managing multiple copy operations. Support for receiving files via the LocalSend protocol. TeraCopy entry in the modern Windows Explorer context menu. Integrated toolbar in the title bar. Why receive LocalSend transfers with TeraCopy? Handle file conflicts: Skip, overwrite, or rename files when a file with the same name already exists. LocalSend always creates another copy, which can waste time and disk space, especially when resuming an interrupted transfer. Filter unwanted files: Apply ignore lists or remove files manually before accepting a transfer, so unnecessary files are not downloaded. Better performance on fast networks: In tests over a 10 Gbps connection, TeraCopy received files several times faster than the standard LocalSend app on Windows. Download: TeraCopy 4.0 Build 26 | 14.5 MB (Freeware, paid upgrade available) View: TeraCopy Website | Screenshot Get alerted to all of our Software updates on Twitter at @NeowinSoftware
    • Briefly used Turbo Pascal (and Turbo C++) in 97 and soon after that I bought PC magazine that included a full version of Delphi 2. I still use Delphi today, some 29 years later.
    • Age of Empires Mobile comes to PC, here's how to carry over progress from your phone by Ivan Jenic Image: YouTube/Microsoft Microsoft just released Age of Empires Mobile for PC. The game, officially called Age of Empires Mobile: PC Edition, is available for free on Steam and Microsoft Store, almost two years after its initial release for handheld devices. Age of Empires is one of those franchises that entire generations grew up with. The original came out in 1997, and immediately got people hooked to building civilizations and crushing their enemies on the battlefield. However, the franchise today is a far cry from its roots, as Age of Empires Mobile is, well, a game optimized for handheld devices, and not a classic RTS title we’ve all loved for years. And, of course, it includes in-game purchases. The PC version is still a mobile game at its core, but it’s been optimized for desktop play. There’s mouse control, full keyboard compatibility, and a refined UI. Microsoft also refreshed the visuals with some 4k textures, so the game should look better on larger screens. The game supports Crossplay, so you can switch between your phone, tablet, and PC without losing anything. But linked progress doesn’t come out of the box, as you have to enable it first. Here’s how to link your progress: On your mobile device, open Age of Empires Mobile. Go to Settings (Gear icon) > Account. Select Bind Account and choose a sign-in option. Once you enable account binding, sign in on PC using the same method, and your progress will be accessible across all your devices. Xbox Game Pass subscribers also get a bonus reward pack on PC, which includes: 1 Monthly Pass Token 1 Custom Resource Chest 10 Universal 60-Minute Speed-Ups 1,000 Empire Coins Exclusive Player Portrait Frame You can find more info about Age of Empires Mobile: PC Edition, as well as download links, on the Age of Empires official website.
  • Recent Achievements

    • One Year In
      Philsl earned a badge
      One Year In
    • Dedicated
      Scoobystu earned a badge
      Dedicated
    • First Post
      Tom Schmidt earned a badge
      First Post
    • One Month Later
      D0nn13 earned a badge
      One Month Later
    • Rookie
      +ChiefOfNeo went up a rank
      Rookie
  • Popular Contributors

    1. 1
      +primortal
      458
    2. 2
      +Edouard
      177
    3. 3
      PsYcHoKiLLa
      124
    4. 4
      Michael Scrip
      79
    5. 5
      Xenon
      76
  • Tell a friend

    Love Neowin? Tell a friend!