Planetside 2 Optimization Issues

Discussion in 'Player Support' started by Jimster480, Dec 6, 2012.

  1. Jimster480

    I have been watching the forums over the last week or so and studying who is having the most issues wth FPS. And based on processor architecture and instruction sets I believe that alot of your optimization issues come from your use of a brand new version of SSE that most processors do not support. Because your new optimizations make the FPS go up on my room mates machine and he is running a Sandy Bridge i5 OC to 4.2. But my FPS do not change at all on a Phenom II x6 1045T. Not only does it not use all the cores (for either of us, its limited to 2 threads). But it really seems like an SSE issue, even my room mates brother has a older i7 940 or 960 at 3.X GHZ and he gets the same FPS that I have. And it doesnt use either of our GPU's (only my room mate gets full power). And based on the forum posts everyone with Core2, Phenom II, Athlon II, Core i (first gen), cannot get any FPS. Now Sandy Bridge is only about 10% faster than its predecessor and Phenom II is matched aswell (although not clock for clock, but its not that much different). You all need to expand the amount of threads that the game can take advantage of aswell as try compiling it with a older version of SSE (2 or 3 would cover most PC's, although 3 is much better but you could also setup a platform-specific build system). I am a software dev myself (C++, ASM, etc) and I have seen problems like this before and suggest you all try it.
  2. Badname3073

    The coders did implement fallback to SSE3 and less, since your processor can run the code. But SSE4.1 adds many useful arithmetic instructions on packed integers, and that's what the game needs, so why not use it? If AMD chose to implement some half-***** SSE4a on their chips which nobody cares about, why should it impact Intel users?
  3. LordMondando

    If this were true, why do people with upper end FX chips seem to not be having problems. I base this on the comparably low number of reports from FX uses, and i've seen no one with a FX8300 or greater complain.

    Just trying to add something constructive before this descends into a AMD ARE **** thread.FX
  4. Badname3073

    Oops, mutually exclusive statements detected:
    LOL
  5. Badname3073

    Because FX supports SSE4.1 and SSE4.2 :D
  6. LordMondando

    Yes, you are correct. My mistake.

    Either way and I say this not merely as a Phenom user. But given that family of chips popularity this the gaming community if this is the case then unless they want to alienate a sizable proportion of the market. Something will have to be done in regards to this.

    Not to mention as the O.P notes, if this is true, everyone but 2nd/3rd gen and fx 'i' users are in effect out in the cold.

    You want to push some 'lol why should intel suffer argument' be my guest. I wouldn't run a MMO predicated on the basis of only being able to really 'sell' it to people who bought a high end intel or AMD in the last 2 years.
  7. Badname3073

    The only way to do this is to **** parts of code which use SSE4.1. Because you cannot magically speed up scalar code to work like vectorized code.
  8. LordMondando

    I have no illusions that its easy. I'm simply trying to argue that if this is the case and that in effect prehaps as much as 40% (hopefully no ones going to contest that) of the userbase is just going to not be able to run the game properly. Then PS2 might be in trouble.
  9. Badname3073

    Again, what you do not realize is that there exist certain computational limits for each chip. The chips with SSE4.1 can do more game-relevant computations than those without. What you seem to lack is understanding that reality is real. You cannot just utter some words to magically "make" computational tasks to execute faster on chips with legacy extensions.
  10. Badname3073

    But that is what the computational problem is, can you comprehend it? The game model is the top-most formulated problem. Then, people implement code to compute the problem. The formulated problem is such, that SSE4.1 instructions speed up computations. That's it, that's reality. If you want to change the game model, it will be a different game then.
  11. ben4345

    My only guess is that SOE is being paid/bribed to make this game upnplayable unless you have the latest hardware from CPU and GPU developers. To try to help boost computer hardware sales.
  12. Jimster480

    Well if you knew anything you would see that people with i7 first gens and other Core systems cannot even play. So unless you have sandy bridge you are fkd? Games should not be built that you cannot play them unless you spend a ton of money on an overpriced intel core series? They should do multiple builds as I said. And yes I am a software developer I have owned a company that does so for many years, and I primarily make low level stuff such as cheats and I work with things like optimization issues in my mini engines (since primarily in my cheats I do not do engine hooks). So yes It does seem to be a problem with SSE 4.1/4.2 implementation. And this needs to be fixed.
  13. XenoScifi

    IMO hasn't this been an issue with gaming for years now? Bad optimization on older processors? The ideal that you have to upgrade to run a game you probably should be running without issues on most current machines?

    I am running a Phenom II dual core with a HD5750 series card on all medium settings, and for the most part the game runs well enough. After playing for awhile the game really starts to poop out on me ie: constant stutters/freezes and enemies/allies not rendering quick enough.
  14. Jimster480

    Not really no. It happens with GPU's sometimes with the newest DX but now CPU's generally. Things are built to support everyones CPU's with better optimization usually for newer ones.
  15. ben4345

    Define "newer"
  16. Badname3073

    If you were really a software developer, you would have known that apart from the formal AVX support, Sandy Bridge is practically no different from Nehalem.
    If you were really a software developer, you would have understood that there are computational tasks which new extensions compute faster than old extensions, and there is nothing anyone can do about it.
    If you were really a software developer, you would have known that implementation of different extension instructions are not different "builds", these are actually different pieces of code, with different asm mnemonics or c intrinsics.
    ya ya
    If you were really a software developer, you would have understood that the situation is directly opposite - there does not seem to be a problem with SSE4.1 extensions, since the code with these extensions is in fact faster than that without.
  17. Badname3073

    FYI, Jimster does not know what he is talking about.
  18. Jimster480

    I am a software developer and I can tell you that Sandy Bridge's difference is that it supports 4.1 and 4.2 as compared to The older i7's. And being a software developer I know that you can build the code with different compiler optimization options having the compiler implement newer versions of SSE and it cannot build for all of them at once, thus making it different builds. And being a software developer I can see that with 4.1 Extensions it does run faster, but there are only 4.1/4.2 extensions and if you do not have it, then it is completely unoptimized. Thanks for being a troll and a *******. Congrats on spending alot of money on a new Intel machine, but not everyone is you and nobody wants to be you. So instead of trolling the forums you should be open to idea's that makes the game playable for everyone before it dies out like Shattered Horizon did due to lack of optimizations.
  19. Badname3073

    All i7 support SSE4.1 and SSE4.2.
    What you do not realize is that there is a whole world beyond your little -ftree-vectorize on a simple scalar code. There are special types for packed data structures, special instructions for them, and many other things. It's just as I thought - you are an -O3 -ftree-vectorize guy, I've seen many of your kind :D
    What you do not realize, is that it is not "unoptimized", it is simply limited by the lack of hardware extensions. There is nothing to "optimize".
    You are welcome.
  20. Jimster480

    And while it is true that the older i7's claim to have this optimization it clearly doesnt work since I know many people who have these CPU's and they are in the same boat as the rest of us. So if it was implemented as intel claims or works as they claimed then all these people with first gen i5 and i7's wouldnt have the same opimization issues.