GDI repainting speed slower in newer versions of Windows 10

[MS Bob 11]

 

[spaceelf]

I too am horribly concerned about 40 year old display interfaces and 18 year old framework performance.

[+virtorio MVC]

I'd like to see a similar test, but using a pure Win32 app (not a WinForms app).

 

 

[MS Bob 11]

3 hours ago, LostCat said:

I too am horribly concerned about 30 year old display interfaces and 18 year old framework performance.

The 30 year old display interface and 18 year old framework is used in a huge, huge number of apps - Win32 native code C++ apps as well as WinForms apps. Technically, some parts of Windows, some technologies are even older than 30 years. Do you refuse to use them and use "only" the newer bits e.g. exclusively apps or OS UI bits which use Direct2D and DirectWrite? 😆

[spaceelf]

5 hours ago, MS Bob 10 said:

The 30 year old display interface and 18 year old framework is used in a huge, huge number of apps - Win32 native code C++ apps as well as WinForms apps. Technically, some parts of Windows, some technologies are even older than 30 years. Do you refuse to use them and use "only" the newer bits e.g. exclusively apps or OS UI bits which use Direct2D and DirectWrite? 😆

Do I refuse to use them?  No.  Do I use much using them?  No.  Do I care about how fast they draw virtually nothing?  No.

[neufuse Veteran]

so me a test on real hardware not in a VM, yeah VM testing is good but it puts another layer of complexity there, maybe the VM graphics implementation is causing issues? who knows

[adrynalyne]

23 minutes ago, neufuse said:

so me a test on real hardware not in a VM, yeah VM testing is good but it puts another layer of complexity there, maybe the VM graphics implementation is causing issues? who knows

Right. I’ve never seen 1:1 graphics performance in a VM vs bare metal either. 

Edited January 27, 2021 by adrynalyne

[MS Bob 11]

Yeah that's true. Slow performance in a VM is always a possibility. However that Twitter user says he tried it on physical hardware as well.

Update:

Second test:

Third test:

 

Finally he posted the test itself for you to test it on your hardware if you have different versions installed in multi-boot configuration on the same hardware:

https://github.com/ADeltaX/REGDI/

[neufuse Veteran]

I can't reproduce this on pure hardware using, using the provided code there is barely any difference between the original windows 10 version and the current 20h2 build that is publicly released on the same hardware and using latest video card drivers difference is within 1%

[adrynalyne]

7 minutes ago, neufuse said:

I can't reproduce this on pure hardware using, using the provided code there is barely any difference between the original windows 10 version and the current 20h2 build that is publicly released on the same hardware and using latest video card drivers difference is within 1%

So we might be seeing a video driver issue too in what they found.