Client Performance - GU03

Discussion in 'Player Support' started by codeForge, Feb 21, 2013.

  1. PaperPlanes

    Codeforge, any input on the fact a lot of people get FPS drops when turning quickly and aiming down sights? A lot of people I know have noticed this and it affects me as well. Usually when I ADS on any gun, I always drop to about 20 FPS no matter what and then it goes back up when I am fully "aimed", but will drop again and go up when I go back to hipfire.

    Stuff like this also happens a lot when I have to whip my aim around quickly, it will drop to a low rate and then shoot back up. It's pretty annoying and disorientating.

    System specs if it matters:

    2x crossfired Radeon 7850 2GB
    Intel I5 3570K @ 4.2 GHZ
    Asrock Z77 Extreme4 mobo
    16 gigs G.skill Sniper Series 1866 ram
    • Up x 1
  2. Egonieser

    Slight FPS drop, not noticeable as much, from 50-60 to 45-55 in WG/medium populated areas and standing around 30-40 in the crown with enemy platoon in the area. Not noticed any twitching while aiming as of yet, but I've not been in a full multi-platoon zerg yet cause the EU servers are a bit empty at 3AM lol.
    Guess I'll do a full stress-test tomorrow when the fights pick up, but been relatively smooth so far.
    Not playing on highest settings, have custom between ultra and low.
    Despite the wild claims going on about AMD systems giving out very bad performance, i seem to do pretty well so far.
    • Up x 2
  3. raw

    My FPS are back in completely unplayable territory again but I appreciate that you colored distant enemies in bright green so I can easily spot them.

    I had two crashes today, both of them when I switched from gunner into driver (F1) in my magrider.
    • Up x 2
  4. codeForge

    The frame rate drop when "Turning quickly" is due to some of our visibility code, which uses temporal projections to reduce the complexity of future visibility checks. When you move fast enough or turn the camera fast enough, it has to throw away pretty much all of it's caches and completely re-evaluate visibility.

    To put it more simply, as you move around in the game the visibility system has hundreds of thousands of decisions to make to determine what you can see around you. It uses your velocity and camera movement to figure out the probability of some of those things changing state visible/not visibile in the future. "If he can't see this THIS frame, and he keeps moving like he's moving, he won't see it NEXT frame either"...that sort of thing.

    The negative impact of this cache miss is one of the things we continue to optimize.
    • Up x 9
  5. raw

    Yeah, it is catastrophal. I wonder how I played games 5 years ago. That would be 2008 by the way. Some of the games you list are older than that. Sometimes I suspect post-90s coders are simply too lazy to optimize their pipeline (not PS2, that's plays in a completely different league). Or something is wrong with the C2D architecture.
    I bought this system because my old PC from 2002 with an Athlon XP somethingsomething just didn't cut it any more. It never ran anything smooth. Never had this issue with the old PC (until it grew old ofc.)
    Makes me wonder...
    • Up x 1
  6. PaperPlanes

    Wow, that is a very thorough answer. Thank you very much. You're one of the things (people, I mean) I really like about SOE's communication, you never fail to give a quality answer that makes sense to people. Thank you again.

    I'm actually going to give up my little boycott that was pretty much caused by Higby because if anything I pay to SOE benefits you, it is worth it.
    • Up x 1
  7. NotTheMomma

    Like PaperPlanes, I also really appreciate the explanation. That sounds incredibly complex, and, unfortunately, has a very profound impact on gameplay. The smart infantry player does *not* stop moving very often in this game, and, due to drop pods, the enemy can literally come from any compass direction at any time.

    If this type of forecasting is common to different areas, I could see how a poor forecast could result in a Sunderer being invisibly and inaudibly damaged when you are right beside it, for example. Like, "you can't see that Prowler, so you can't see or hear the Prowler's projectile, either".
  8. y3ivan

    Does it improve PS2 performance if distant objects such as infantry, tanks, and aircraft to be rendered in 2D. I have seen quite a few games using this method to cut the load on PCs with limited capability.

  9. tacticalretreat

    Massive fps loss for me, even medium battles are unplayable, biodomes back to 15 fps.
    • Up x 1
  10. Bloodmyth

    I feel for you guys who seem to be running this game on a calculator, albeit a scientific one ;) I wonder what the average gaming rig is these days but I see a few posts from ppl with machines that should have died years ago but they somehow expect ps2 to run for them, do us a favour and stop holding my game back lol.

    P.s I'm only being an **** coz I bought a new pc, yay me :)
    • Up x 1
  11. permafrost

    So the question I have is why PhysX and why not PhysX 3.0 if you are going to use PhysX? Why not a OpenCL physics engine that supports both AMD and NVIDIA? Also, if you are going to use PhysX, how about a creating a PhysX wrapper for OpenCL?
    • Up x 2
  12. Lucius Trio

    Wow, that was an entirely useless post which contributes absolutely nothing to an outstanding issue.

    Some of us greatly exceed the requirements of this game for even high settings, and still get horrible performance.

    i7-2600k @ 4.2
    2x ATI 6970s (2GB GDDR5 each)
    16GB DDR3 @ 1600MHZ
    AsRock P67 Extreme 4 Mobo
    Patriot Wildfire 128GB SSD
  13. Miggs

    For those of us who have re-discovered that it can be quite an engaging enterprise (I wouldn't say "fun", it's more like "interesting") fiddling with game and system settings to try and trace just what setting has what effect on our game fluidity, something that you can access that provides identical surroundings really should have been in there from the start for a PC game that needs to perform on such a varied client platform harware base.

    Nice to hear something like this is in the pipeline at least.
  14. Miggs

    So what we should notice is that if we rotate through 360 in a loction slowly (compared the rotating quickly in that same situation) we should see less of an FPS hit on our systems as we are giving the CPU more time to update the items that are visible to us?
  15. BigBlackenstein

    codeForge are we ever going to see a fix the Desynch/Warping issue that a lot us are experiencing, or are we left to be dust in the wind of a new player surge? I hope this is the appropriate place to ask, some kind of response from the Devs after three weeks of silence would be peachy. Keep up the good fight.
  16. Cryless

    @codeforge can u explain us somehow what exactly are u guys doing when u optimize something? this is something I asked myself for quite some time now :)
  17. SuperLexatron

    Any update on this desync/warping issue? It's been 3 weeks now and 2 weeks since the last response, and haven't been able to play the game since. People in the main thread (myself included) ( ) are starting to get a little impatient and frustrated at the lack of dialogue between them and the devs, and the lack of any assurance this problem is a priority on the list (and since it affects so many players, I would have thought it should ne - 3 of my friends have no quit because this was the 'icing on the cake' for them.' :() and the fact that patches are now being brought out, with the new stuff to buy, new added/change features when a lot of your customers are still suffering from GU2... very frustrating indeed.

    Would appreciate a response as you have given these other people on a far more trivial matter.
  18. BenYeeHua

    Profiling, then find out which process time is the most, then optimize it.
  19. iller

    Yeah and that makes perfect sense.... but how much of it is actually Serverside where it will do mid-end users some good? (Ie: sounds like really high-traffic Level3 cache stuff that only helps expensive CPUs & would have a strong bias for +perf on Intel processors only). Nor does it help us around really "Porous" Bases or ones with too many doors & windows. Will this other "Normalizing Method" feature also be accompanied by more powerful .INI settings to finally turn most of the Eye-Candy off somewhere between the Server and our CPU? I fear without those, we're only going to discover that NOTHING we do "on the rollercoaster" will really matter.

    I don't have these problems so much in GW2 in similar sized battles because their Umbra engine culls most of this stuff before it even gets Cached. CPU optimization is only a very small part of the formula there.
  20. Blackoth

    I would say that i gained a few FPS this patch, i play with all settings on high except particles and shadows at medium and i saw about 43fps last night, didnt get into a large fight tho. will try to evaluate tonight