I'm running Windows 2000 Pro with SP4 on my System. I have still not decided to install DirectX 9 because i don't know wether it works with the Original Drivers (latest officals) from 3DFX...I think they are at least DirectX 8 certified.
Does anybody know how my Voodoo5 would react on this Update?

Voodoo5 will work with DirectX9, but it is not DirectX9 compliant. This means that you mostly likely will not be able to play newer games or apps designed for DirectX9. Some of them might work, but with certain problems, and bugs.

Most of the time you wouldn't need DX9, and the use of DX9 with DX8 or 7 apps will bring some slowdown.

The only reasons to install DX9 are to see if you can play any newer game, or you want to benchmark V5 in 3DMark2003.

If none of the above applies, remain with DirectX8 or 8.1 installed.

Personally, I had Win98SE run DirectX9 and a Voodoo5 back in time : No Problems observed :)

Still, there are repeated reports of some 3dfx Users, which machines basically stopped working correctly after they upgraded DirectX. I do believe though, that this is mostly configuration related (this happened to some when upgrading from DX6>7 and DX7>8 already)..

Directx9 is no big deal actually. I saw one game (GunMetal i think), and by the way it looked, it's even lower in terms of functions and power used, than a DirectX6 compliant game. So bad jokes happend, and fortunately for the sellers, on big money...

Right. But if you wanna play a new game which doesn't uses the newest shaders (that means it would be shown correctly on the screen without picture faults on the Voodoo5) you have to install DirectX 9, no matter wether it uses DirectX 9 Shaders. There goes the dilemma.
Second question: how large is the slowdown in DX8/7 games when using DX9?

The slowdown, around 7,8% when using D3d games that were not designed for DX9. But ironically, if they were designed for DX9, they would go even slower. Why? The use of larger libraries, more CPU jumps, means less CPU performance...

Hmm ....  ::)

There is no such thing as "DirectX n.n is x.x% slower than the previous version".

In most cases, DX9 is just as fast (or slow) as DX8.
I made the experience that it helped a few synthetic Benchmarks a bit (M$ tweaked a few miror things), but there's no drastic Performance difference involved.

For me, it just enabled the use some DirectX 9 ATI Demos at the release time; otherwise I wouldn't have even noted the difference.

If it was a fixed 7.8% (?) slower, MicroSoft would have never released it (they would have fired some Coders instead)  ::)

I wouldn't be so sure. I tested Dx8 on a Trio 3D, as oposed to Dx7 inside various games. When using Dx8 (even if the game wasn't), a certain drop in speed was noticed, and also graphics corruption. I wouldn't believe it is elseway with DX9, since when I installed DX9 on a banshee, I saw a visible decrease in performance when using Dx8 games (including a benchmark in 3DMark2001 SE).

Also, from tests, even the tinyest change in a library inside DX can turn something from fast to crawl. So...

Ehm....

A Trio 3D ???

Sorry, but testing like 8 year old Cards with modern DirectX releases is sort of pointless, isn't it (?)

Heck, I could test a Matrox M3D with DirectX 9.0b, but well... *scratches head*

Quote:

The slowdown, around 7,8% when using D3d games that were not designed for DX9. But ironically, if they were designed for DX9, they would go even slower. Why? The use of larger libraries, more CPU jumps, means less CPU performance...

Ha, going from DX8 --> DX9 actually uped my 3dmark2000 score (from 3550 to almost 3600)


Quote:

I tested Dx8 on a Trio 3D...

:o wow  :o

I see...results are different from user to user. so everybody sticks again to his own ones..
:P