@jombo23 (View Post)
Well if you wanted to prove you are an asshole, you in fact just proved the opposite. I'd suggest editing out that ending, you don't need to put one word on every line.

You have some major false assumptions in your post, which make it mostly invalid. For one, a 3d TPT would not have 140 million particles. That is just unreasonable, of course nothing would support that many at once. I imagine a 3d TPT-like game as having a smaller width / height, and just having extra layers. It would probably be a few million particles.

It would not take days to render a 10 million tpt particle simulation. The simulations you are thinking about have ultra-realistic physics and are designed to simulate real-world particles. TPT is just a game, the water physics are not realistic at all.

The particle grid right now takes 12-13MB by my calculations. Although I somewhat remember it was more like 20MB. For a new 3D version you can throw out pavg[0], pavg[1], flags, and possibly some others. In a simulation the size of a few million particles, that's only 1GB. Definitely a lot, but if you removed further properties and optimized it (we don't need 4 bytes for everything) it would go lower.

You try to claim that RAM would be the limiting factor for FPS, but that just isn't the case. Every single byte of the particle data is not accessed every frame. We don't care about the .tmp2 of DUST, and we definitely don't care about anything where there are empty particles. In addition, stuff would get put in the ram cache which has much faster access than actual RAM. If coded properly, most accesses would hit the cache.


Those fluid simulations you talk about are ultra-realistic. They simulate many properties TPT doesn't do. Even collision TPT does not do properly. It is particle order dependent, meaning particles look to move into a spot where there is just an empty space at that moment. It doesn't do the entire movement calculation at once or use real world formulas.



"Id also like to add in the fact that it has nothing to do with time, difficulty, or money"
It has everything to do with time, difficulty, and money. I imagine it would take a team of experienced programmers months or years to build this from scratch. It would take a lot of skill, it would have to be ultra-optimized to run at a reasonable speed. It would want to take advantage of multi-core CPUs and GPUs, neither of which TPT does now. It might not even work out. But it is never a good idea to claim something is "impossible".

@TPT_PL (View Post)
Mostly useless properties intended for QRTZ, TUNG, and I think one other element to break during rapid pressure changes. It is an array of two floats, even though only one float is needed to do that. It was repurposed for PIPE, STOR, and VIRS to store extra information when storing particles