[]$[Homie]

some do, though normally the driver settings app, uses one setting/slider for both D3D and OGL

[mitsubishi]

They shouldn't I said  , if they do.

[L3mMinG (old)]

Quote

lowering to 1024x800 and i would say it perfectly playable (this was on a local setup, not online with ppl and fights going on)

The problem is that all the fighting/models/particles get drawn in areas of the map that shouldn't be rendered but are being rendered as well. So handling the whole uptown is 1 thing, handling it with 16 people getting their gun off is another

Quote

it is related, but in practise for gaming Trilinear is no good because of the extra frame buffered, this really slows an older grfx card/cpu combo and forces Vsync on making things even worse

(beside theres no Hardware T&L in Q3, hence the hard CPU hit)

Trilinear forces vsync ? that sure is news to me 

Would it be extremely hard for the ioQuake3 people to make quake3 actually use hardware T&L, thinking about it logically they would only have to add "some replacement functions here and there" ?

[]$[Homie]

ok talking this further with Lemming, it seems my memory lets me down

its triple buffering NOT tri-linear filtering that requires vsync to be enabled
...................................................................

"The disadvantage of VSync is, that the total frame rate can never exceed the refresh rate of the monitor. So if your system is set to use 60 Hz in the given resolution, the frame rate can never exceed 60 frames per second. But since the monitor isn't capable of displaying more than 60 distinct images per second, a higher frame rate doesn't do much good anyway.

Another disadvantage is, that if the game doesn't use three buffers for rendering the scene (a so called triple buffer), the frame rate cannot fluctuate freely. Only integer fractions of the monitor's refresh rate (i.e. 1/1, 1/2, 1/3, 1/4, 1/5 etc.) are possible. In the case of the refresh rate of 60Hz, the only possible framerates would be 60fps, 30fps, 20fps, 15fps, 12fps..."
....................................................................

so from that we can assume that vsync WITHOUT triple buffering can cause wild FPS changes, so if u have vsync enabled u MUST run triple buffering as well. triple buffering WILL hurt older cards/cpu's as another frame has to be stored in the vRAM
....................................................................

"Memory Usage and Triple Buffering

Triple buffering seems like the perfect solution to cure low frame rate, however, it does not come free. Enable triple buffering also means it requires 50% more frame buffer spaces. Under certain conditions, it can also negatively impact your gaming experience. A simple example can illustrate this potential problem. Let's say we are running a game at 1600x1200 resolution. Each pixel needs 32 bits to store the information, 1600x1200x32 = 61,440,000 bits, converting it to megabytes, it equals to 7.32MB. To use double buffering, it requires 14.64MB video memories; to use triple buffering, it requires 21.96MB video memories. Heck, it's only 21.96MB, what's the big deal? Modern video cards have 256MB video rams onboard. You are certainly right, it does not seem like memory usage would be an issue. However, once we enabled 4x FSAA, the number gets inflated really fast. With 4x FSAA enabled, there are 4 times more pixels being sampled, thus you need to multiple 21.96MB by 4 and it's whopping 87.84MB! That's more than 1/3 of your total video ram. If a game requires 200MB of space to store textures, light maps, bump maps, normal maps......etc., you are going to have a very bad gameplay experience with lots of pauses caused by hard drives accesses. This is just something to watch out for, if you find the game accessing the hard drive too frequently, it can be an indication that triple buffering is eating up too much of your video memories."
...............................................................................

if u are running a 64Mb card or even a 128Mb u could run into serious performance problems with vsync and triple buffering enabled, either run vsync and no triple buffer, or better still just dont enable triple buffering or vsync at all, and set the fps cap to 60 (which is what i do)

Trilinear Filtering does employ a small performance hit, so if squeazing the most performance out of ur card is priority stick to bilinear. for ATI users still enable geomerty instancing as that will help also

[~SG~Juno]

Quote

Each pixel needs 32 bits to store the information

I thought 24 bits

[L3mMinG (old)]

Could somebody try using r_finish 1 ? This has to do with parallelization so I've read. Might help with that affinity stuff too.

[mitsubishi]

I remember that var from when I was editing an autoexec script..

I've written

set r_finish "0" // *1 drops fps a bit* "1 may improve input lag" // it seems from the code, it calls glFinish from opengl: http://www.mevis.de/...l/glFinish.html someone from raven: "the GLFinish command will then make the app wait until the rendering device has completed all the commands it has been sent up til then."

[]$[Homie]

I must be one of the lucky ones, I have a dual cored AMD64 AM2 X2 6000+ and setting affinity makes no difference to my rig

on a side note but still on topic, Lemming u were talking about input lag in irc, theres some interesting notes in this article (about ATI Tray Tools settings) with regards to queue length and how it can affect input lag. its the bit about Flip Queue Size under Anisotropic/Trilinear Filtering Optimizations

[/eVo/CRIMENY]

Quote

I thought 24 bits

I thought it was 24+8 check bits...so both are valid?

[]$[Homie]

isn't it that TFT displays have 24bit output not full 32bit?

the textures themselves will still be 32bit and that's what's being rendered regardless if the monitor is a CRT,LCD or monochrome