[Bf-committers] Back Buffer Issues

Ton Roosendaal ton at blender.org
Fri Dec 3 00:03:47 CET 2004


Hi,

Great, we really need help on that part of the site. I have to make an  
account in cms for it, and its a bit late, do that tomorrow and will  
send you info.

I also would be interested to find out what these "many other cards"  
is... probably more in the professional market? On all reports we had  
in until now, this effect was only a problem for Sun Solaris.

And the performance issue; how much slower are you talking about?

-Ton-

On 2 Dec, 2004, at 23:50, Matthew H. Plough wrote:

> Since this email pertains to the bf-committers discussion and also  
> pertains to documentation, I am sending it to both bf-committers and  
> bf-webcontent.
>
> After reading Daniel Dunbar's notes on the problems with front- and  
> back-buffer display schemes at  
> http://www.zuster.org/blender/drawingArchitecture.html , I understand  
> why no fix is possible.  However, no information is readily available  
> on the topic on any of the Blender web sites.  My card, and many other  
> cards on the market, do page flipping by default for performance  
> reasons.  Many of them can be switched to swap copy mode in the  
> advanced preferences.  Could we put this up on  
> http://www.blender3d.org/cms/Graphics_Cards.201.0.html , the page with  
> graphics card troubleshooting information?
> I suggest adding the following, or something similar:
>
> Common Issues
> o Some consumer...Display Properties.
> [add] o Blender's menus may be displayed incorrectly in certain buffer  
> swap modes.  Try changing the swap mode to Swap Copy in the  
> vendor-specific area of the <a href="#1118" target="_top">Advanced  
> Display Properties</a>.
>
> I would not mind becoming the maintainer of that page.
> Matt
>
> Ton Roosendaal wrote:
>
>> Hi,
>>
>> Blender likes the SwapCopy a lot. :)
>> This is becoming a standard in many (PC) cards now, and for some   
>> systems you cannot even tweak it... for example Macs only support  
>> this  method.
>>
>> The Swap-Exchange could work too, it is the old Blender default from   
>> the SGI period, but for it to work 100% I need someone with such a   
>> graphics card system who's devoted to get it to work. Not difficult   
>> coding, just tracking all little things in the code to keep it   
>> swap-exchange compatible.
>>
>> A year ago we discussed this here too, with Daniel Dunbar, going over  
>>  the best buffer strategy with Blender. The conclusion back then was  
>> to  go for 100% swap-copy, and code exceptions for the (older)  
>> systems that  don't support this.
>>
>> Now, what's there to fix?
>>
>> -Ton-
>>
>>
>>
>> On 2 Dec, 2004, at 19:25, Matthew H. Plough wrote:
>>
>>> Hi all!!
>>>
>>> I was playing around with my video card's OpenGL settings this   
>>> morning, and found an interesting clue on what's causing the odd   
>>> corruption shown in   
>>> http://www.princeton.edu/~mplough/tmp/buttons_error.png .
>>> First things first: I have a 3Dlabs Wildcat Realizm 200.
>>>
>>> Second things second ;) : Like most cards, my card has two options  
>>> for  dealing with double buffering -- page flipping (which only  
>>> exchanges  the front and back buffer pointers) and Force Swap Copy  
>>> (which  maintains a dedicated back buffer and a dedicated display  
>>> buffer).   When a buffer swap happens in forced mode, data from the  
>>> back buffer  is *copied* into the display buffer, but is not  
>>> cleared.  My best  interpretation of Blender's code leads me to  
>>> believe that it depends  on this preservation of back buffer data.
>>> Is this in fact the case?  If so, how can we fix this?
>>>
>>> Matt
>>
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at projects.blender.org
> http://projects.blender.org/mailman/listinfo/bf-committers
>
>
------------------------------------------------------------------------ 
--
Ton Roosendaal  Blender Foundation ton at blender.org  
http://www.blender.org



More information about the Bf-committers mailing list