I would really appreciate advice on configuring it to get the optimum performance put of it.i'm using window xp.And i would like to know why i'm getting pretty low score with 3dmark2k1,i get about 1.8k,but i search and compare,similar systems gets abt 2K and above,btw i'm usingt amigamerlin 3.0.

you were the guy with the modified mac voodoo5 pci, right? here's the problem:
1: PCI gets a little less performance than the AGP varation, because the PCI bus is limited to -i bet- 133MB/s while the AGP bus is not
2: A modified MAC Voodoo5 gets nearly full performance, but - caused by the modification - you'll never get averages other voodoo5 owners will get

Ok,i check it with my seller,he told me that this is the pc version,i shall take his word for it,as he has a favourable rating on ebay.So that is why i can't understand why i get only 1.8k for the 3dmark2k1.My cpu is a athlon xp(palmino) 1700+@1618mhz with 320 sdr memory on a ecs k7s5a.Would really need help to rememdy the situation.

2 pixel per clock rendering=enable
3d filter quality=Normal
Alpha blending=smoother
D3D guardband clipping=enable
Anti aliasing=fastest performance
Edge-aliasing=disable
level of details=0
Maximum buffered frames=6 pending buffers
rendering color-depth=software controlled
scan-line height=1 line
speed setting=fastest(v-sync off)
trilinear texture filtering=disable
Z-buffering optimization=enable
[This is my setting in 3dfx tools for Direct X]

Ok firstly, forget 3D Mark benchmarks, they're Direct3D benchmarks and you'll never get a true performance comparison.  The early versions run on the 'Max Payne' game engine, and I don't consider whether my V5 runs 'Max Payne' at a decent fps to be a true test of it's abilities.  The latest version is a DirectX 9.0 benchmark, again pointless.  Run an OpenGL benchmark or use Q3.  If you're primarily after D3D performance though, you've definately just gone out and bought the wrong card.

Secondly yes the PCI bus has limited bandwidth which is shared with other PCI bus devices, if you have a SCSI controller for example and an ethernet nic and more you will see lessened performance to AGP, but not to a very great degree.  patiences post covers this fully, I need not elaborate.

Thirdly once a Mac Voodoo5 is flashed with a PC BIOS is will work just as a PC Voodoo5, no difference in performance.

i know, the two statements i gave don't really influence the game speed of the card - but if we refer to 3dmark they definitely do. as i don't know how PhoZy configured his system i also can't refer to (system configuration (that means drivers, loaded topics and so on..) is -beside that nobody is interested in- a very important point concerning benchmarks, believe me!)

@ PhoZy:
i've tested all 3dmark versions with a voodoo3 3000 and a voodoo5 5500. here's what i've learned:
3dmark and voodoo doesn't go well together. the better way to count your game performance is fraps (only opengl and direct3d)

first,i have a radeon 9000pro which i use for gaming.Secondly i brought this v5500 because i'm curious abt them and that is why i brought them.Ok, i know that voodoo and 3dmark don't mix well.And lastly i'm new to 3dfx products and would appreciate help in tunning this voodoo5500.And i'm using directx 9.0b wil that decrease my 3dmark score in 3dmark2k1?(my card that number on the top right hand corner 210-0414-001-a,would this string of number let me know whether this is a mac version or a pc version)

http://www.falconfly-central.de/cgi-bin/yabb/YaBB.pl?board=drivers;action=display;num=1064400920

:)

personally, i've now changed to directx9.0b but have not tested it with 3dmark...just recognized some textures in ut 2k3 weren't garbage anymore.
well..in win9X you'll get a 3dfx tuning tool:
http://www.falconfly.de/tools.htm -> Misc. Tools / Utilities -> 3dfx OverClocking Tab
this tool allows you to overclock you voodoo5 very safely (if an error occurs, the tool automatically restores defaults), but it works only under win9x! for winnt you can try VoodooControl V1.81 Beta (same link&section)
i don't know what the number on the board refers to, maybe somebody else can help...by the way, what says the sticker on the little chip above the first VSA-100? it should tell you the bios-version.

wrote on 20.10.03 at 08:04:49:

210-0414-001

That is the FCC product code number.

wrote on 19.10.03 at 14:52:25:

Maximum buffered frames=6 pending buffers scan-line height=1 line This is my setting in 3dfx tools for Direct X]

6 pending buffers  :o, you're wasting memory. Don't use more than two. And I think 1 line interleave isn't the optimal value. Maybe 20 lines would be better.

wrote on 19.10.03 at 14:52:25:

2 pixel per clock rendering=enable 3d filter quality=Normal Alpha blending=smoother D3D guardband clipping=enable Anti aliasing=fastest performance Edge-aliasing=disable level of details=0 Maximum buffered frames=6 pending buffers rendering color-depth=software controlled scan-line height=1 line speed setting=fastest(v-sync off) trilinear texture filtering=disable Z-buffering optimization=enable [This is my setting in 3dfx tools for Direct X]

Hm, indeed...

I believe (but don't quote me on that) that 4 lines are Standard for ScanLine Interleave.

3D Filter Quality should be on "High" and Alpha Blending "Sharper" for maximum Visual Quality (performance loss is neglectable compared to the gain in Quality) IMHO.

And 6 Pending Buffers really is a waste of memory, and can induce severe lag during Gameplay.

Batracio is absolutely correct, 1 Pending Buffer (=Double Buffering) is ideal for Gaming, everything higher was just intoduced to slightly increase some Benchmark scores.

WoW,thAnx guys changing the amt of buffer to 1 it really increase my performance,so far i have tested out on the fill rate test(multi-texturing)and it jumped frm 568 to 597.5mtexels.And abt the scan line stuff.should i put 4 or twenty?

after some tweaking in the 3dfx tools,i found out that changing the number of buffer back to 6,does not decrease performance but however changing the the scan line height to 128 increases performace by abit.(now in the multitesture test,it jumped to 600mtexels) ;D

while my setting were on scan line height 128,i tried to test unreal touranment 2003 on,and i started getting this weird effect,is when u get shot the whole screen should turn red,but however this time stripped pieces of red(horizontal),and i can see lines far apart frm each other on the ground and after a while the game crashed.So i decided to change it to height of 4,only the former problem remains but this time the gap is much smaller.I would like to know whether is this the problem with increased height,or is this a bug? :-/