I think so, but am not sure of the command. When you get into a game, hit your zoom button and pan around 360 but pan kinda slowly. This will force terrain caching for almost the entire played area of the map. The structure deal is something we have to deal with as it's inherent to the map and our vid cards and drivers. It has to do with a video card thing called occlusion culling. No cards have a real handle on occlusion culling even today. Google for occlusion culling demos and download one, run it and see what occlussion culling is all about. Well ok actualy the drivers could do a better job of culling but the driver makers have never really dug into it as much as they could. And mappers still can bring cards to their knees by not mapping as they should.
edit: occlusion culling means your video card doesn't waste time rendering anything that's not in frustrum/view for that frame. When you pan across an area in game that has a structure above ground or below, and your fps drops or your card stutters, that's because the parts of the structure that are unseen are still being rendered needlessly. There's a fine line between what you want the card to render and not render. Say you're walking into an entranceway to a building. Do you want to wait for the card to render the insides before you step in or after? My guess is before. This is why there's a fine line in culling scenes. But the game makers and card makers could do a lot better, I mean after all this is free performance we're talking about here.
By the way, have you tried another vid driver, perhaps an older one? Back in the day when the r300 core (ati 9700pro) came out, there were preferred drivers for t2 that everyone knew of. Cat 3.4 rings a bell, but will not work on your card.
The way I select a driver for a card, especialy one that is no longer made, or the actual architecture has been replaced like the ati1950 (and 8xx) series was replaced by todays cards, and the 7xxx series was replaced by the 8xxx series in nvidia's case, there are huge differences in core architecture. From vector to scalar rendering. The new cards and drivers take the vector game code of older games and transform it into scalar that the card can work with. You think this would be laggy having to transform vector into scalar code but the new scalar cards are freaking awesome at it and work very well in most cases. In your case with that 800, you're running a vector card on a vector game and that means that newer drivers aren't optimal for your card.
Scalar cards have huge numbers of cores that dynamicaly share the workload, and can be tasked on the fly with new operations. Vector cards have hard coded instruction sets for the most part with few programmable aspects. Think DX9 and older hardware. This means that when your vector card needs to draw textures it uses the texture rendering engine and the pixel shader is doing absolutely nothing because it has no pixels to shade, and this idle shader hardware is wasted when it could be helping to texture. In such a case a scalar card creates pixel shaders and texture engines out of the cores it has whenever it needs them. For example: this to-be-rendered frame is mostly pixel shader code so the scalar card programs and devotes most cores to shading, the next frame needs mostly textures, so it sets the cores up to create textures and not for shading. This scalar method is far more efficient, and this is the primary reason why the scalar cards kick the poo out of the vector cards, even with the scalar card at lower core and memory speeds than the vector card.
There are some superb vector cards around, the 8xx series to the 1950 in the ati lineup, and the 6800 series to 79xx in nvid.
I would look into the date that the next series of cards after yours came along, the 1950 I think was like the last vector card from ati. Go a bit before that date and select a driver that is no younger than that date to ensure that the driver has most optimisations for your card series yet no code for other architecture cards. Dl that driver and install it. This means your driver is optimised for your specific card. No additional code was written into that driver for different architecture cards.
Case in point: let's say I have a nvid 6800gt (wich i do) but the newest card is some kind of 290gt or some other. The issue is the 6800gt is a vector rendering card whereas the 290 is scalar. There is no updated code in the newest nvidia drivers specificaly for my 6800, just the 290.
The latest drivers from nvid that will run my 6800 are over 100 megs of compressed code, while the drivers that were actualy written for the 6800 when it was state of the art are some 34 megs. So if you run that new driver on older vid cards, you're not only adding additional code that may or may not be placed in memory at run time that is completely unnecessary, you actualy have installed a driver that is likely slower than the first few drivers written when your card was state of the art.
This long winded deal can be drawn down to this: use the last driver before your card was replaced by a new type. I can't help myself on certain subjects because I am a hardware freak, and a freak in general.
I just now saw this thread. I'm on an ATI Radeon 4870X2 card as well, running Vista 64-bit with 8gb of DDR2 system ram. I have edge-detect 24x anti-aliasing, maximum quality mip-mapping, and 16x anisotropic filtering all turned on via the Catalyst Control Center. I, on average, get 150 fps in Tribes. I've never experienced any choppiness whatsoever. Have you updated to the latest drivers? And are you using a piece of third-party software (like EXPERTool) to set your fan speeds up? Maybe your card is overheating.
So I uninstalled the card completely, and reinstalled it with the original install disc's catalyst drivers.
The difference was night and day. I used to have drops down to 10-20 fps, now I'm at a solid 60 almost 100% of the time. We'll see how things go with different maps as the servers get more populated.
To be honest nothing in the "Graphics" submenu made much of a difference for me. The "Texture" submenu is where I found the biggest changes. Unfortunately anything but max or near max makes the game look like crap.
Interior textured fog should be disabled if you'd like best performance. I have never missed fog inside buildings and doubt you will either.
Speaking of older drivers, ati cards, and 3GHz p4 systems, my old 3GHz p4/9700pro system played tribes and tribes2 flawlessly. I ran the 5.11 cats, and was very impressed with the performance. I ran 16x aniso and 2xaf at 1024x768 @ 100Hz refresh vsynched. It was the system I had always been wanting. But it would drop frames when there was lots going on in scene. Your 800 is the great granchild of the 9700, and has the potential for much better performance than my 9700 and the 60fps you're getting, my guess is near 400fps in open areas. Someday I'd like to build another p4 sys, very reliable and gamed fine.
You might also do well to enable triple buffering. Vsynch usualy renders better performance if triple buffer is enabled, but sometimes if you have a huge resolution, triple can take up all your video ram. If you have plenty of vid card ram you can enable triple and enjoy the benefit.
These links will hopefully keep me from writing another book here:
I have noticed with both Tribes and Tribes II, this happens with later versions of ATI video card drivers. This issue was happening to me as well until I installed ATI Catalyst v7.10. This is the latest driver I could find that fixed the problem.
I do not know if the same could be said for Vista machines, I am running XP.
Comments
edit: occlusion culling means your video card doesn't waste time rendering anything that's not in frustrum/view for that frame. When you pan across an area in game that has a structure above ground or below, and your fps drops or your card stutters, that's because the parts of the structure that are unseen are still being rendered needlessly. There's a fine line between what you want the card to render and not render. Say you're walking into an entranceway to a building. Do you want to wait for the card to render the insides before you step in or after? My guess is before. This is why there's a fine line in culling scenes. But the game makers and card makers could do a lot better, I mean after all this is free performance we're talking about here.
By the way, have you tried another vid driver, perhaps an older one? Back in the day when the r300 core (ati 9700pro) came out, there were preferred drivers for t2 that everyone knew of. Cat 3.4 rings a bell, but will not work on your card.
The way I select a driver for a card, especialy one that is no longer made, or the actual architecture has been replaced like the ati1950 (and 8xx) series was replaced by todays cards, and the 7xxx series was replaced by the 8xxx series in nvidia's case, there are huge differences in core architecture. From vector to scalar rendering. The new cards and drivers take the vector game code of older games and transform it into scalar that the card can work with. You think this would be laggy having to transform vector into scalar code but the new scalar cards are freaking awesome at it and work very well in most cases. In your case with that 800, you're running a vector card on a vector game and that means that newer drivers aren't optimal for your card.
Scalar cards have huge numbers of cores that dynamicaly share the workload, and can be tasked on the fly with new operations. Vector cards have hard coded instruction sets for the most part with few programmable aspects. Think DX9 and older hardware. This means that when your vector card needs to draw textures it uses the texture rendering engine and the pixel shader is doing absolutely nothing because it has no pixels to shade, and this idle shader hardware is wasted when it could be helping to texture. In such a case a scalar card creates pixel shaders and texture engines out of the cores it has whenever it needs them. For example: this to-be-rendered frame is mostly pixel shader code so the scalar card programs and devotes most cores to shading, the next frame needs mostly textures, so it sets the cores up to create textures and not for shading. This scalar method is far more efficient, and this is the primary reason why the scalar cards kick the poo out of the vector cards, even with the scalar card at lower core and memory speeds than the vector card.
There are some superb vector cards around, the 8xx series to the 1950 in the ati lineup, and the 6800 series to 79xx in nvid.
I would look into the date that the next series of cards after yours came along, the 1950 I think was like the last vector card from ati. Go a bit before that date and select a driver that is no younger than that date to ensure that the driver has most optimisations for your card series yet no code for other architecture cards. Dl that driver and install it. This means your driver is optimised for your specific card. No additional code was written into that driver for different architecture cards.
Case in point: let's say I have a nvid 6800gt (wich i do) but the newest card is some kind of 290gt or some other. The issue is the 6800gt is a vector rendering card whereas the 290 is scalar. There is no updated code in the newest nvidia drivers specificaly for my 6800, just the 290.
The latest drivers from nvid that will run my 6800 are over 100 megs of compressed code, while the drivers that were actualy written for the 6800 when it was state of the art are some 34 megs. So if you run that new driver on older vid cards, you're not only adding additional code that may or may not be placed in memory at run time that is completely unnecessary, you actualy have installed a driver that is likely slower than the first few drivers written when your card was state of the art.
This long winded deal can be drawn down to this: use the last driver before your card was replaced by a new type. I can't help myself on certain subjects because I am a hardware freak, and a freak in general.
lol
Shit
The difference was night and day. I used to have drops down to 10-20 fps, now I'm at a solid 60 almost 100% of the time. We'll see how things go with different maps as the servers get more populated.
Thanks for you help, much appreciated.
Speaking of older drivers, ati cards, and 3GHz p4 systems, my old 3GHz p4/9700pro system played tribes and tribes2 flawlessly. I ran the 5.11 cats, and was very impressed with the performance. I ran 16x aniso and 2xaf at 1024x768 @ 100Hz refresh vsynched. It was the system I had always been wanting. But it would drop frames when there was lots going on in scene. Your 800 is the great granchild of the 9700, and has the potential for much better performance than my 9700 and the 60fps you're getting, my guess is near 400fps in open areas. Someday I'd like to build another p4 sys, very reliable and gamed fine.
These links will hopefully keep me from writing another book here:
http://en.wikipedia.org/wiki/Triple_buffering
http://www.ocworkbench.com/2006/articles/DXtweaker/
I have noticed with both Tribes and Tribes II, this happens with later versions of ATI video card drivers. This issue was happening to me as well until I installed ATI Catalyst v7.10. This is the latest driver I could find that fixed the problem.
I do not know if the same could be said for Vista machines, I am running XP.