lol so by my calculations, 40+ REAL people, dont make me show ya screen shots of em! lol - your 20-25 people puts me......at least 10-15 up!

Im doing a victory lap! Woot!

Like I said above, I expect this is simply a case of dedicated gamers probably spend more on their PC's than the rest of us and it really does depend on who your pitching your game at.

At the other end of the spectrum of course, I have clients (non game related), education sector, who are not even running P4 chips, most are on PII's and a few lucky ones have PIII's.

So lets face it its a broad market, pick your niché

The same peak point should also apply if one has a 1.5 GHz processor and Radeon 7000, or a 4 GHz processor and Radeon X1800. It's the balance that matters. The performance graph is based on my system which is sort of balanced. Memory, the hard drive speed, and programs running in the background also play a factor. When doing the experiment I closed whatever I could to free up memory which meant that I had to go offline to free up 140 MB memory from Firefox (the big one) and other programs I had running. I took the fastest values I could.

In otherwords, stick to models of about 1500 to 5000 polygons for best results. I need to run an experiment to find the peak with more precision than in the graph. The graph just gives the full range and isn't focused on anything special, just the limits of DirectX (and rendering speed).

Note: I'm assuming "balanced" as in the card and processor being of the same release dates (as in when they were first available). I don't know when the first 3.0GHz processors were released, but they should be somewhere around August 2004, which is in the range my video card (Radeon 9600 XT) was released.

It would be interesting to draw from a nice range of 'balanced' PC's to get an idea of how the graph would be expected to go. For example across 3 balanced platforms starting with a P1 lowest spec through to a P4 High spec, some kind of benchmark tool.

The data could possibly then provide a rolling expectation of performance chart that could allow you to plot your title and cross reference the recommended PC spec for your title.

Very interesting and certainly more research is needed in this area.

If I had such computers available, I'd do so, if it was legal though (the part of installing 3DGS on other computers without a team license is the big one). I've given my test subjects so you can test it yourself. All you need to do is take note of the kps value and average it out ignoring sudden drops. Then, given the numbers, I could plot them on that logarithmic graph I made. For some, especially with 3.8 GHz processors and the very latest video card, they could be getting 12 million polygons per second as their peak (a semi-random guess). My motherboard only has AGP available (no PCI-express like the newer cards only seem to have) so I can't get the newer cards without first swapping the current motherboard.

The hypothesis is that, with the processor, video card, and video card drivers from the same date of creation, the peak is going to be roughly the same. When an old processor is used with a new video card, the peak will be much higher (~6K) and when a new processor is used with an old video card, the peak will be much lower (~800). It's only a hypothesis though and not tested. I don't have such a test environment to doing this with so it'll be difficult for me to analyze and verify it.

Quote:

---

12 million polygons per second

---

Though obviously peak polygon throughput is a theoretical domain, the newer cards and CPU combos ought to be capable of significantly more than that..I could say a GeForce 7800 and a 3.8 ghz cpu ought to be able to process several 100 million polygons. of course this would be untextured or at least without shaders.

The upshot of this is that raw triangle power is becoming less relevant. Pixel and vertex processing is key now, well as mesh batching.

I don't have such cards so this is only a guess I made. It doesn't hurt to run an experiment and find out, however.

Whoa wait, wait, wait, you guys. Shouldn't we stay on topic?

This topic, as wishdragon have started, is about choosing which engine (Torque or 3DGS) is right for him, not about Poly counts and Performance or average users who have computers with a 2-3.4GHZ processors. Right now, there are modern processors that don't go beyond 4GHz, as Intel have tried that sometime in 2003 and it failed; thus, Intel have started to scale back their clock speed and improve performance of the processors, on which AMD have already done, who have provided a lower clock speed like 2.4 or 2.6 GHz, but provides a medium to high-end performance. So clock speed has nothing to do with performance but in order to accommedate the clock speed, chipmakers will have to add more transistors (a switch which holds a value of 0 for off and 1 for on), increase cache size, and any other techniques to help improve performance of a processor and lower energy consumption. So in a conclusion, higher clock speeds don't mean higher performance. Plus, higher GPU clock speed doesn't equal to higher performance, either. You need to have a good understanding of how CPUs/GPU's work.

OK. Let's go back on-topic. Please?

@wishdragon:

Quote:

---

I have right now narrowed my search for what I want between two engines.

The one here, and the coming Torque Shaders Engine.

The plan is to make a mmorpg.

Those of you that have used this engine and have used the torque as well, what do you prefer about each and why?

---

If you're new to game developement, go with 3DGS.