The memory consumption of emPolygonizer4 can grow quite fast. The more input particles and the higher the level of detail the more memory it will consume. emTopolizer2 however meshes things bit by bit which results in neglectable memory consumption no matter how your level of detail might look.
So with heavy setups emPolygonizer4 can easily go up to 10 Gig and more of RAM whereas emTopolizer2 will only use a couple of hundred megabytes. Speed wise it depends a little on how the input looks like (amount of particles, spacial distribution, ..), but on average emTopolizer2 is at least twice as fast. In most cases it will be even 2-10 times faster. But in my opinion the speed is not as important as the memory consumption. Fact is that with emPolygonizer4 you will eventually have memory problems and with emTopolizer2 you won't. Here a concrete example that I just tested. Note the memory consumption. Hardware: Sony Vaio Laptop, Intel i7m, 8 Gig Ram. OS: Windows 7 64 bit. Software: SI 2012 SAP A.) The input was a small liquid simulation with roughly 2000 particles. The level of detail was set to 50, which resulted in a mesh with about 2.4 million polygons. emPolygonizer4: time: 30 seconds. max. RAM: 2.6 Gig. emTopolizer2: time: 10 seconds. max. RAM: about 0.15 gig. B.) Same input, but this time with liquid filaments enabled which results in roughly 270000 particles. The level of detail was set to 50, which resulted in a mesh with about 1.5 million polygons. emPolygonizer4: time: 55 seconds. max. RAM: 2.7 Gig. emTopolizer2: time: 26 seconds. max. RAM: about 0.2 gig.
