SFFT alpha drivers - V3/4/5 - Join to Testing

[Rolo01]

Okay,
I found some time for testing Alpha 50, here are the results :

A major improvement : Battle for Middle Earth is finally playable !

Here are some ingame screens :
Image Insert:

78.95 KB
Mines of Moria, some white textures, but playable !

Image Insert:

106.56 KB
Battle scene, lots of texture problems.

Image Insert:

93.89 KB
Nice ingame screen...

In the opening landscape, there are texture errors in Alpha 49 and 50 :
Image Insert:

121.15 KB
In Alpha 48 there were no errors on the landscape, but the game would crash to desktop when entering a level, so Alpha 49/50 is the way to go...

Sid Meiers Pirates! is working :
Image Insert:

83.71 KB
I installed Patch 1.02 and used 3D-Analyze 2.36b to emulate T&L,Dx8.1 Caps,bump maps, sim.textures and cube maps.

GTA San Andreas is working :
Image Insert:

94.26 KB
I used 3D-Analyze 2.16a to emulate T&L,Dx8.1 Caps,bump maps, sim.textures and cube maps.

So far, very impressive and good work, SFFT !
As usual, testing was done on Voodoo4 AGP on Pentium4-3400 with WinXP SP1.

[Rolo01]

Some more results :
Ankh 2 exits to desktop with Alpha 49/50, it worked with some white textures in Alpha 48.

XIII does not start anymore, even when using 3D-Analyze.

I got a BSOD when messing around in Pirates! and GTA:San Andreas WITHOUT 3D-Analyze, with 3D-Analyze it worked fine.
Don´t know, maybe the info is usefull :
STOP: 0x0000008E (0xC0000005,0xBF9FFDB0,0xF21CEAF4,0x00000000)
***    3dfxvs.dll - Address BF9FFDB0 base at BF9C3000

3D objects in Need for Speed:Underground 2 are still black !

[Glide]

Very well, Rolo01.

Thank you of course for your report; I've just a small question for you.
That's, do you prefer this current alpha 50 release or previous alpha 48?
Tnx in advance and good testing,

Bye bye.

[Rolo01]

As I had no major issues with Alpha 50 and was very excited when I saw Battle for Middle Earth finally work on a Voodoo, then I would say I prefer Alpha 50.

[m14radu]

Rush for Berlin - unexpected error.



Conflict Global Storm - the game exits after a very short time.



Ice Age2 - playble.

Br.
Radu.

[Kokohemmo]

Hi

Tested SFFT 50 in F1 2002 and BF 1942. Both works fine now, but when enabling 4,8xFSAA, it didn't work. 3dfx tools enabling FSAA, but no effect.

Unreal and Unreal Tournament no picture att all.

Radu, did you try enabling FSAA too ?

Regards
Kokohemmo

[m14radu]

no, just "Fastest Performance".

[osckhar]

Hi,
FSAA does not work via V.Control. You must to do it manually though.

Oscar

[Rolo01]

Hmm, thats odd. I enabled 2xFSAA when testing GTA:San Andreas and it worked with Alpha 50.

[m14radu]

I've tried today "The hell in Vietnam" game.
The game starts well, but the ingame texture are corrupt.

No luck with " Tomb rider - The Legend". The game should work, the minimal requirements is a geforce3 card.

Unfortunately i don't find anymore the C&C Tiberum War DVD - last time when tried the game it don't work.
Can soembody give a try to this game ?

Br.
Radu.

[ps47]

..geforce3 usually means basic pixel shaders=no good.

[m14radu]

hmm, i tried also using the 3DAnalyze, (with the option emulate pixel shader enabled ), with the same result.
I have my doubt that this option in 3DAnalyze don't work.
Until now there was no game where i enable the option ( using 3DAnalyze ), and the game starts. None.
Star Wars - Republic Comando is a good example !

[mhanor]

sfft51 is an interesting release, for me at lest
I've noticed that glide2x.dll and glide3x.dll were recompiled
there are important improvements with sfft51 regarding the problems I've posted with quake2... now I can join a server (tested r1q2 with r1gl and default opengl) and the driver doesn't hang right before entering the game (after trying to connect to a server)
it still hangs under certain circumstances with 0xEA (I'll try and find out more)

a new issue... playing r1q2 with r1gl @640 (I don't know if it happens only with these paramenters) then typing quit at the console, after exiting the game, the desktop reapers with the icons rearranged... this doesn't happen every time... I can't figure it out (my desktop is @1024x768x16bpp) it has something to do with switching from r1gl to software (see below)

there's still that question from r1q2 to create a crash report when switching from software to r1gl/default opengl

also, in software mode, when selecting r1gl @1024x768, the result is r1gl @640x480 (probably it has something to do with those errors when creating the opengl context); after that I can switch to 1024

when switching back from r1gl to software... the game enters a software windowed mode... even if I select fullscreen... after entering that software mode, the ingame fullscreen option is disabled; if I select it to Yes the second time (being in software), the game refuses to do fullscreen... I don't know if its related to the video driver or to r1q2

another thing... I've tested the behaviour of ioquake3 on alt-tab... with sfft50 and below it always crashed in glide3x!grDrawTriangle+0x2d
the same thing happens with sfft51... I've tried sfft50 with GlideXP (http://www.users.on.net/triforce/glidexp/) (I'm sure I've replaced glide2x.dll and glide3x.dll, I don't remember about the OGL and ICD dlls) and the game didn't crashed, it handled very well the change on alt-tab and I could reenter the game with no problems... unfortunately, GlideXP has other problems (like not-so-good image quality)
r1q2 with r1gl also crashes at glide3x!grDrawTriangle+0x2d

I've uploaded 2 minidumps (and some basing info provided by windbg) for sfft51's 0xea hang/bsod (this time the driver verifier was able to restrain the driver... no more 0xea bsod... it only hangs and the OS terminates it)

https://www.3dfxzone.it/public/uploaded/mhanor/2007117202526_sfft51_minidumps.zip
unfortunately, I've uploaded it twice... I forgot how the 3dfxzone.it forum works with attached files

because the opengl ICD and OGL files have not been changed, I've decided to post the new findings here and not in MesaFX's thread

[roflkopp]

A few reasons why releasenotes are NOT that bad:


- people get an idea what they should test
- people get a better reason (than a higher versionnumber) to install the new driver
- sfft gets an survey of older driver releases (does anyone remember in which way sfft33 was better than 32?)
- 3dfx users could give better assistance for coming driverreleases (ok thats like point 1)



So the final question is: IS THERE ANYONE BESIDES ME, WHO SUPPORTS THE IDEA OF PUBLIC RELEASENOTES?

[Rolo01]

Ok,
release notes are a good idea. But what if they are not possible ?
Lets say, a driver release shows some improvement in some games.
So SFFT tries to improve it some more, by changing the driver code in line 578 ( just an example ) and tweaking somewhere else.
No one knows what the effect will be, maybe its worse than before, maybe better.
I personally have no benefit from knowing what SFFT changed in the source code.
Just my 2 cents...