DirectX 10 = DirectX 9 for Vista

Even if we removed the API and OS, if you suddenly had a way of saying "nothing less than a G80 or R600 plays here" then you could do some amazing things as they are just criminally powerful toys ;)
If you aimed at 720p/30FPS on a 8800GTS, with no support for any lower-end GPU, I'm very confident you could do some ridiculously amazing things indeed. Console GPUs are pretty much toys compared to that...

Now, if someone actually did that. Hmm! ;)
 
Hmmmm. Time to load up Steam and add to the Vista + 64bit + DX10 crowd.
 
Well thats the real world
I repair a fair few pc's for people and i very rarely see a pc with more than 512mb
and even more rarely with a decent gfx card
 
Hmmmm. Time to load up Steam and add to the Vista + 64bit + DX10 crowd.

That's the weakness of even such a large sample as the Steam survey.
How many Website reviewers and magazine reviewers are there all over the world? How many who are in QA at various companies? How many are in the industry and shop up just to keep themselves up to date, just as I read up on the latest happenings in my field?

Just how many of those 2500 people are NOT connected to the industry in one way or the other?

When we are talking about these very small populations (similar for SLI, not to mention Crossfire!), the people who end up on the survey for work related reasons likely skew the data significantly.
 
IMHO the question should not “Can D3D10 render better visual effects?â€￾. We properly all know that beside of some corner case (were DX9 API hacks could help) D3D9 is able to produce all this “D3D10 effectsâ€￾ they show us. This may require some help from the CPU (Geometry shader as example). The real question is more like “Does D3D10 allows us to get the same visual effect with less overhead?â€￾

Most developers are already able to produce much better visual effects but at the cost of lower FPS. When D3D10 can do the same at higher FPS it does anything we currently need most.
 
Comparing with DirectX9,DirectX10's disvantage is far greater than its advantage
relying on this point that DirectX10 is only for Vista.There is no problem for developing better visual effect based on DirectX9 than DirectX10.But there are not many cases around there.
http://www.ogre3d.org/phpBB2/viewtopic.php?t=21587&highlight=
http://forum.beyond3d.com/showthread.php?t=40805

I'm not sure what you're trying to suggest; that first link has nothing to do with DX9 or DX10, the second link is about how the improved functionality and performance within DX10 is making a very realistic shadowing method possible on today's hardware.

If you're trying to say that the first link is somehow showing the same features as the second link, then you're mistaken.

Next, I still don't see how D3D10 is at a "disadvantage" at all. Compared to D3D9, D3D10 has nothing but advantages. If you want to say that being tied to Vista is a disadvantage, it's my opinion that you're wrong: the new OS is a requirement for the performance increases available in D3D10.

If you spent the time reading the feedback right here in this thread from people who are actually developers and not just forum junkies with non-graphics-related jobs, you'd find that nearly all of them agree that the performance set couldn't happen in XP.

The features can certainly be "back ported", but D3D10's main claim to fame is far-improved performance for tasks that previously were too compute-intensive. That performance is obtained through the new driver and kernel model, and is something that realistlcally couldn't be expected of XP.
 
I don't think opinions here will help to corret wrong thing.It's easier said than done!
"developing better visual effect based on DirectX9 than DirectX10."Do you know what does it mean?It means that all general ways you can think about are not enough,also needs extra thoughts which make use of unique solutions and algorithms etc(unconcerned with DirectX9 and DirectX10) that can be applied by DirectX9 and DirectX10 to produce much better visual effect without lower FPS.

For example,Real-Time animating Fluid Simulation
1


http://img204.imageshack.us/img204/3599/devr2007060611595443fv4.jpg

2
http://img528.imageshack.us/img528/5067/devr2007060611595703nx0.jpg

3
http://img389.imageshack.us/img389/1095/devr2007060612014804ke0.jpg

4
http://img79.imageshack.us/img79/7549/devr2007060612014954ur2.jpg

P.S This idea came into my head on september 2005.I brought it up at website on July 2006.
http://www.devmaster.net/forums/showthread.php?t=6507

5 gow_head_pop.gif


http://img91.imageshack.us/img91/8600/gowheadpopkw9.gif
It released similar effect on september 2006.

I have researched many solutions and refined algorithms and they can produce similar effect as Vray on realtime 3d based on DirectX9.More important thing is that they can continually improve to Maxwell level.In my own case,DirectX9 is only option for being less expense and better effect after considering large amout of customers' benefit!
 
Alright this discussion is getting silly... of course you can implement everything in DX9 that you can in DX10 with no restrictions... I can implement everything on my Pentium 200MHz with no video card; it'll be pretty slow though.

There is no argument here: DX10 is simply more powerful and more efficient than DX9. It has a superset of the functionality - which allows better and more efficient implementations of algorithms - and less overhead. i.e. everything that can be done in DX9 can be done equally or better in DX10.

I have researched many solutions and refined algorithms and they can produce similar effect as Vray on realtime 3d based on DirectX9.More important thing is that they can continually improve to Maxwell level.In my own case,DirectX9 is only option for being less expense and better effect after considering large amout of customers' benefit!
Okay no one is arguing that we suddenly don't need new graphics algorithms - far from it! What DX10 does is further expand the scope of algorithms that can be expressed efficiently and executed on graphics hardware. A more apt comparison would be the DX9 version of the shadows demo that I posted earlier and the DX10 one that you linked: you may notice that the latter runs faster and provides many more features than the former.

Anyways I'm gonna stop talking. DX10 is better than DX9, period. Everything will eventually switch over to DX10 once people have suitable hardware, because there's no reason not to.
 
What would the chances be of devs implementing DX10 features and have an alternate OpenGL renderer that does the same for DX10 hardware on DX9 OSes?
 
Well ATI doesn’t include the necessary extensions at the moment and maybe they never would. Therefore developers would not be very willing to invest time on such a port. The situation may change with Mount Evans but we are far away from this.

Even without knowing the exact details of Mount Evans I believe a DX10 core renderer can port in around 1 – 2 month to this new OpenGL API. But there is a much bigger problem. All the shaders written in HLSL need to be converted too.
 
But there is a much bigger problem. All the shaders written in HLSL need to be converted too.
I think that could be automated. You might not get the best bits that way, but it's much better than having to rewrite them by hand. Debug and profile.
 
Back
Top