To be fair I don't have enough degree of competency to give you strict answer, but I don't sure how well is depth buffer working on scenes with tons of objects and very fast distance changes. It's just technique. As long as I understand it's can be easily used for landscape/units/buildings rendering. E.g you can see not LOD, but fake flat planet from distance which get more geometry depend on distance. Anyway it's probably not something Uber will use.
Oh well.. our problems may soon be over: http://qz.com/108920/the-materials-brea ... mes-faster :mrgreen:
Comprehensive thoughts on this: - it isn't optimized yet. Making the game load and play will in the end work theoretically on integrated graphics client side. - it is designed to work lighter on weak computers while stronger ones can carry weaker computers in the same games, making medium planet systems plausible standalone. -since it will run light for a user yet carry heavily on a server, the public servers and the community servers can run off of powerful I9 64g ram dual graphics liquid cooled computers which i believe you can get a bunch of people chip in and buy one or rent one by the month. All in all, this game isn't asking for more than the common game asks for, battlefield and all...
My computer runs this alpha BETTER then supreme commander and my computer is hardly top notch. Operating System: Windows 7 Home Premium 64-bit (6.1, Build 7601) Service Pack 1 Processor: Intel(R) Core(TM) i3 CPU 530 @ 2.93GHz (4 CPUs), ~2.9GHz Memory: 6144MB RAM Graphic Card: NVIDIA GeForce GT 220 Give them some time to adapt/optimize the game to different hardware and i doubt you will have any issues.
I still don't understand this. I had an almost identical setup, though with a 9800 GTX+ instead of a GT 220 (they are essentially the same thing, 9800 is just twice as big), with a better processor, and I was getting around 2 FPS at the start of the game. It's as though the computer gods favor you.
Thats because sup com was never optimised properly as promised and that is exactly what I am worried about.
It does seem a bit hit and miss with the specs / performance. I should get at least 20FPS going by some of the setups discussed around here, but I tend to sit pretty at around 2-3 late-game.
Do you understand what game engine is? It's just bunch of code, you can change everything if you have source code. Main supcom problem was because of synchronous network architecture. It'a why SupCom only able to use two processor cores: one for rendering and one for simulation. PA is completely different, it's engine designed with understanding of SupCom problem.
The computer gods definitely favour you. Running on an i5 3570K @ 3.4GHz with 16GB RAM and a GTX 460 with 768MB VRAM, PA would go from 30fps at start to ~3fps in a big game with hundreds of units. Upgrading the GPU to a GTX 760 (4GB VRAM) made all the difference in the world; 120fps at start (VSYNC locked I reckon) dropping to ~80 once I died and was in spectator mode. Personally I reckon you need 2+GB VRAM to avoid any shared memory slowdowns (i.e. the GPU needing to use computer RAM because it has filled up the available VRAM). Unfortunately 2+GB VRAM = $$$ whether you go nVidia or AMD Having said that, running on an AMD E350 HTPC (dual core 1.8GHz, 4GB RAM and HD 6310 GPU) previous builds have been running at 15fps at game start, dropping to a few fps later (all graphics settings on low). That is damn good for mildly optimised code on a comparitively slow CPU / GPU combination trying to render at 1920 x 1080. Your i3 & GT 220 should have the edge on that box speed-wise. Actually, since this is still Alpha, Uber can and have made major changes to the engine. The initial build did not differentiate internally between celestial objects, causing your base to appear on the moon or sun if you changed focus. It now mostly does, because the intent is to have multi-planet battles. The orbital layer is not fully implemented yet, nor is the body -> body transit, when they are that will again be (effectively) a major engine update. So from an optimisation perspective, there can still be some major render pipeline or sim updates pushed out, and that is entirely at the perogative of Uber. This is not a finished product getting the last layers of polish, this is the raw basics being built up from the ground, and I fully expect the finished game to have several major differences from the current implementation, just because better ways were found to do things.
So if a game doesn't run perfectly at alpha then it will never run well, despite loads of optimization?
Even as a hobbiest coder, I have to ask: is that a serious question? Please tell me you are trolling Bhaal there..... Code optimization is something not just reserved for graphics processing, but EVERYTHING in a program system. The "engine" as it were isn't even really complete at this point, since they're still adding features. Problem with adding stuff is that most times it's added as something standalone (at least in the few things I've done), and the optimization comes later finding all the common ground among modules. Hell, they prolly have a person working full time on this kind of code cleanup. If not, come the beta they should
The fundamental core design of PA allows much greater scaling both up and down that a synchronous design like SupCom. We haven't realized the down scaling at all yet but it's going to be possible to run the game with something like a "thin client" that's been discussed in other threads. In SupCom you were literally limited to the speed of the slowest peer which you can't get around. Client performance in PA is strictly about rendering which can be simplified and or improved in speed for that client. Bottom line, we are setup to provide good performance we just have plenty of optimization and feature work to do before we get to that point.
I asked it in a rather sarcastic way because I get what you said, I just see so many people see an unfinished product and judge it as a released one. Hell the game has become so much more playable in the couple months its been in alpha, and like the devs said they haven't even started heavy optimization yet.
Phew! Highlighting the important word here: plenty. Performance right now prolly more depends on drivers and hardware interactions than anything codewise right now. Or at least that would explain the lower performance on certain cards that are better in terms of "raw stats". Also, optimization efforts don't have to end at the retail release gate, right?
Premature optimization is the root of all evil. So it's good that they are not going all crazy about it yet.
I have every faith in the abilities of this crew. Forged alliance proves it, and as for the performance lets have the discussion now, lets not wait until its to late. That is what this process is about.Every view is valid.
