Re: [Qemu-devel] Useful Q&A

[Benjamin Brown <support@silentsoftware.co.uk>]

Johannes Schindelin wrote:

>Hi,
>
>On Wed, 16 Jun 2004, Benjamin Brown wrote:
>  
>
>>>>Will support for newer graphics cards be included any time soon?
>>>>        
>>>>
>>>No.
>>>      
>>>
>>Is this sort of thing difficult (I've not written an emulator before)?
>>    
>>
>
>Why on earth would you want to do that? It is an *emulated* graphics card.
>So any card which works should be fine. There is no point of maintaining
>two or more solutions to the same problem.
>
>  
>
Perhaps to give it the edge over commercial solutions? I notice somebody 
recently mentioned open 3dfx specs were available. Don't get me wrong 
I'm all for the project plan of get everything working, then tune 
performance but asking the developers to at least consider the pros and 
cons of adding the support in the later stages of qemu dev (i.e. qemu 
1.5-2?). I've some simple pros/cons here, pls add more.
pro:
entices ppl from commercial products
will allow greater compat with direct 3d games (and effects)
con:
requires time
pos. v difficult

>>To me the i386 support seems to be pointless - anybody who is seriously
>>going to be running QEMU will be using a minimum of a i586 (K6-2,
>>Pentium etc), granted QEMU it is not intended for just this processor
>>platform - what is the justification for a 386 compatible version....? :)
>>    
>>
>
>Well, there is more to it than just focusing on the newest, fastest, best
>processor there is.
>
>  
>
Please elaborate on this point? I was suggesting upping the bar to i586, 
hardly cutting edge :)

>First, why do you suppose at all, that the main use of QEMU is on INTEL
>platform? There are quite a few 386 compatible processors, and while not
>exactly being slow, they do not all support INTEL extensions.
>  
>
I didn't in my day to day use I use Sparcs and ARM machines. I was 
suggesting extra compile support for i586 - that is when the more 
important issues have been rectified (i.e. bugs)

>Second, you need time to write the special support for all those
>extensions. Don't underestimate the effort!
>  
>
I realise its a lot of work but consider it to be of benefit in the 
later stages of the project especially as the projects main competitors 
are x86 based and currently have the perforamnce edge.

>Third, any day of the week I would rather fix things that don't work yet
>than fine tune some performance issues.
>
>  
>
Agreed. Bugs always come first.

>Fourth, I strongly doubt that supporting those extensions does have a high
>impact on performance: They are too special. They are mainly for large
>vector an matrix calculations, and last time I checked, a word processor
>does not really use many of those.
>
>  
>
They would be useful in software directx, multimedia and game usage...! :)

>>Excellent software all round - wish I had the technical know how!
>>    
>>
>
>Use the source, Luke!
>
>  
>
I'm perusing it now when I need to see how certain things work, but as a 
beginner C programmer I'm fairly useless in this respect!

>to contribute back what
>you learnt as a means of saying "Thanks".
>
>  
>
I'm considering a writing launcher, and time permitting figuring out how 
to grab the qemu display and fitting it into the launcher window.

Benjamin