Anybody have an idea of percentage of 3d accelerated video cards used by your customers.

I have been developing and planing to develop 3d games for some time. But I am not sure what percentage of customers could run my games (not have one yet). My concern is rendering around 10000 polygons at 25 fps. Therefore I need some info about customers' video card profiles.

Most customers won't be able to manage that much, according to my logs (they are posted somewhere in this forum but I can't get the search feature to find them).

To be sure of getting the vast majority of customers the baseline is a TNT, but the T&L capable cards have a sizeable chunk.

Cas :)

Those logs ibn the other thread are subjective though. It all depends on your target audience. If you develop a FPS game then your target audience is likely to have higher capabilities than if you develop a puzzle game.

There's still the problem that a lot of shareware buyers choose shareware because of the generally lower hardware requirements though. The trick is in the marketing. Ensure you get your game noticed by the kinds of people that would have the hardware and interest for the game. (EG games magazines, action game websites, etc).

Just for reference, the basic Q3 engine has a recommended on-screen limit of 10,000 triangles. In Q3:TA they increased this limit in accordance with hardware growth to 15,000 triangles.

Then a couple of years ago was Return to Castle Wolfenstion which set the limit at 20,000. Most recently was Enemy Territory which recommended no more than 30,000 triangles. However Enemy Territory was plagued by people complainaing that their mid-range systems were choking on it.

10,000 is a scene is not excessive by todays standards for commercial games, but might be excessive for a shareware title. Don't forget - the Q3 engine was optimised by one of the world leaders in game engine development and so will likely be 3-5 times more efficient than any of the freeware engines out there.

As an example, the Irrlicht engine is very nice for Q3 stuff. But when playing Q3 on my current hardware I can manage up to 400 frames per second. Running in Irrlicht I only get about 150-200. Irrlicht is a great engine, but like most freeware engines it's not that highly optmised.

TNT is fine for me. I tested and got reasonable values as 30 fps with 10000 shaded and textured polygons at 640x480. And I guess most people have at least a TNT nowadays.

Your 10,000 polygon microbenchmark is probably a horribly degenerate best-case scenario (the use of which nvidia are famous for touting just how fast their hardware is). In real life you are likely to be really dealing with 100-1,000 sets of 10-100 polygons, all of very different sizes and using different blending modes and textures. The TNT chokes on that pretty quickly.

Cas :)

Originally posted by ergas
TNT is fine for me. I tested and got reasonable values as 30 fps with 10000 shaded and textured polygons at 640x480. And I guess most people have at least a TNT nowadays.

There is big percentage of relatively new PC's with built in graphics like those from Intel's 810,815-845 chipsets and those "cards" are worse than TNT 1 with 16MB when performing 3D. I recently tested my game with some low end cards and for example - Diamond's TNT1 with 16MB is performing 2.5 times better than 810E and about 1.5 times better than 845G ! This, of course depends on the game but...

Oh yeah, those on-board chips are a right b****rd. Not only are they pretty crappy to start with, but their driver support is terrible too.

The Nforce's onboard GF2MX cards aren't too bad, but the intel (and other) ones are a joke. Alas, because they are cheap and require no inserting of cards into the precious computer (especially off the shelf ones that have huge stickers saying "opening this PC invalidates your warranty") the average Joe has no idea what a pile he has.

It's almost tempting to splash around big sign saying "we do not support X cards for any game". That might wake a few poeple up. Alas, cheap is always preferable to good - which is why we have a shot at making some money ;)

You know, though, a shareware game will typically sell for several years. So it may be more important to you to make a game that won't show its age really badly in years to come, than to make a game that won't choke on the *current* worst-case hardware.