Performance

Cr4zyb4st4rdCr4zyb4st4rd United Kingdom Join Date: 2012-08-09 Member: 155200Members, NS2 Map Tester, Reinforced - Supporter, Reinforced - Silver, Reinforced - Gold, Reinforced - Diamond, Reinforced - Shadow
Been having some lower performance recently and since performance should get better over time. I did another test.
As I have run a few "performance tests" before, I went back to the last one for data.

Now lets compare the newest patch, build 287, to that older one from build 277.
Minor detail this is a fresh install of NS2, with a wiped appdata folder because before that it was even worse.

Earlier performance is taken from here: NS2 and CPU speed scaling test.

The new 287 build numbers are from multiple runs averaged.

9e37d2b6fb.jpg
OLD AVG @ 4.7GHz: 150.53
NEW AVG @ 4.7GHz: 136.4
OLD AVG @ 4.0GHz: 132.78

So 10 builds and on average there's a 10% loss in this test alone, seeing considerably larger losses in the individual runs (up to 20% less) and noticeable impact when actually just playing the game in 6v6 where I used to hold 144+ now having issues with seeing 144 at many points.

I wouldn't have posted this, but with PDT aiming for more lights and particles because of all the performance we've apparently gained and several other competitive players who mentioned this decline in performance, I figured what the hell, someone might think about looking into it.
«1

Comments

  • Saffron_bakerSaffron_baker Sweden Join Date: 2015-06-09 Member: 205352Members
    Just for the record. Computers get older and worse as time goes.
  • FoxyFoxy United Kingdom Join Date: 2014-08-19 Member: 198032Members, NS2 Playtester, NS2 Map Tester, Reinforced - Shadow
    Just for the record. Computers get older and worse as time goes.

    Whilst technically correct, you'd already have to be running at the limits of the machine and just about meeting NS2's system requirements for this to be a factor.
  • dePARAdePARA Join Date: 2011-04-29 Member: 96321Members, Squad Five Blue
    I never fall under 180 fps. And thats on an 18 slot server.
    Cant really believe that you struggle to hold 144fps on an 12 slot one.

    Wasnt your last scaling test under "extreme conditions" with 18 bots?
    Looks like many has changed in the ENSL when bots are playing there now.
  • IronHorseIronHorse Developer, QA Manager, Technical Support & contributor Join Date: 2010-05-08 Member: 71669Members, Super Administrators, Forum Admins, Forum Moderators, NS2 Developer, NS2 Playtester, Squad Five Blue, Subnautica Playtester, Subnautica PT Lead, Pistachionauts
    edited February 2016
    I think Saffron meant the OS gets slower as time goes on - which is true - instead of implying hardware?

    This assumes a lot, though, and probably still has very minimal impact on the results.
  • Cr4zyb4st4rdCr4zyb4st4rd United Kingdom Join Date: 2012-08-09 Member: 155200Members, NS2 Map Tester, Reinforced - Supporter, Reinforced - Silver, Reinforced - Gold, Reinforced - Diamond, Reinforced - Shadow
    dePARA wrote: »
    I never fall under 180 fps. And thats on an 18 slot server.
    Cant really believe that you struggle to hold 144fps on an 12 slot one.

    Wasnt your last scaling test under "extreme conditions" with 18 bots?
    Looks like many has changed in the ENSL when bots are playing there now.

    Yes it was done with bots, that's why this one was too. I can't test different conditions as that wouldn't be a fair comparison and I don't have a build of 277 just lying around to re-test with.

    So for the purpose of getting the closest result as possible I retested my exact tests from before, settings, resolution, core count, clock count, nothing else running that isn't required. Plus averaging multiple runs.

    The fact it's clearly different in these tests to the tune of a good 10% less, in three months, while nothing else has these issues.
    Plus OTHER PLAYERS noticing worse performance in 6v6. I'm not alone in noticing this which is why I've mentioend it, if i was the only person I knew with these issues I could put it down to hardware, software, something else, but I'm not.

    That doesnt mean everyone is noticing the drop in performance only some, what causes it isn't known Im merely pointing it out with a test where I can say this is what I got in this test before and after. Not many people just have a past performance test with logs of their framerates lying around from before with the ability to reproduce it.

    I've gone from practically constant 200fps in 6v6 to dropping below 100 in some cases? Yeah that's not normal hardware or software degradation.

    On the plus side I tested with the atmospheric territories mod and only noticed almost identical framerates easily within margin of error.
  • dePARAdePARA Join Date: 2011-04-29 Member: 96321Members, Squad Five Blue
    edited February 2016
    Take 2 servers, each with an score over 20.
    Im sure you will find enough players who are saying: "Ah, server 1 is crap. Hitreg is bad there"
    As far as i know EVERY server with an score over 0 has the same hitreg.
    So i would call this a classic placebo.

    And about testing with bots.
    Maybe the bots are "better" now compared to previous builds and thats why they need more CPU time?
    And maybe thats why its a bad idea to compare different builds with bots?
    Cause without bots you wont see a difference?

    But im sure some devs can enlight us here.
  • Saffron_bakerSaffron_baker Sweden Join Date: 2015-06-09 Member: 205352Members
    Hopefully the texture optimization Obraxis is working on will help this issue!
  • CalegoCalego Join Date: 2013-01-24 Member: 181848Members, NS2 Map Tester
    Interested to see if this has to do with Mem usage from Skins.
  • LocklearLocklear [nexzil]kerrigan Join Date: 2012-05-01 Member: 151403Members, NS2 Playtester, NS2 Map Tester, WC 2013 - Shadow
    coolitic wrote: »
    xDragon wrote: »
    Your eye can't see past 30 fps anyways.

    I can noticeably see the difference between 60 and 75.

    Your statement makes me want to cry.

    Are you a wizard?
  • Soul_RiderSoul_Rider Mod Bean Join Date: 2004-06-19 Member: 29388Members, Constellation, Squad Five Blue
    xDragon wrote: »
    Your eye can't see past 30 fps anyways.

    Your brain can interpret 1000 or more frames per second, although human range of perception in difference tends to top out around 150 in Armchair tests.

    You cannot say eyes cannot see above 30 fps, as your eyes are a constant stream source.

    I thought this myth died of old age when I was a youngster...
  • FrozenFrozen New York, NY Join Date: 2010-07-02 Member: 72228Members, Constellation
    Unless you can g-sync your eyes to your monitor's refresh cycle, you're going to see the difference between frame rates.

    The more frames per second, the lower the chance that you perceive the change.
  • SquishpokePOOPFACESquishpokePOOPFACE -21,248 posts (ignore below) Join Date: 2012-10-31 Member: 165262Members, Reinforced - Shadow
    Higher frames also feel better. Aiming on 120hz monitor at 120fps is almost dreamlike
  • dePARAdePARA Join Date: 2011-04-29 Member: 96321Members, Squad Five Blue
    @xDragon
    You should stop programming gamecode after your last comment.

  • Cr4zyb4st4rdCr4zyb4st4rd United Kingdom Join Date: 2012-08-09 Member: 155200Members, NS2 Map Tester, Reinforced - Supporter, Reinforced - Silver, Reinforced - Gold, Reinforced - Diamond, Reinforced - Shadow
    edited February 2016
    dePARA wrote: »
    Take 2 servers, each with an score over 20.
    Im sure you will find enough players who are saying: "Ah, server 1 is crap. Hitreg is bad there"
    As far as i know EVERY server with an score over 0 has the same hitreg.
    So i would call this a classic placebo.

    And about testing with bots.
    Maybe the bots are "better" now compared to previous builds and thats why they need more CPU time?
    And maybe thats why its a bad idea to compare different builds with bots?
    Cause without bots you wont see a difference?

    But im sure some devs can enlight us here.

    People are not complaining about the Hitreg (well they are but im not on about that)
    The issue is the FRAMERATE regardless of server is WORSE.
    The stuff above is all local tested anyway and would not change it in anyway.

    The bots are all "frozen" they are doing nothing they're just there as extra entities with no actual movement/aiming calculations their impact is considerably reduced.

    edit, i meant to edit the above post rip
  • CalegoCalego Join Date: 2013-01-24 Member: 181848Members, NS2 Map Tester
    xDragon wrote: »
    Your eye can't see past 30 fps anyways.

    ...am I seriously the only one here who thought he was making a joke?

    Don't worry @xDragon, some of us understood you.
  • turtsmcgurtturtsmcgurt Join Date: 2012-11-01 Member: 165456Members, Reinforced - Supporter
    edited February 2016
    i've been having lower performance recently as well, especially noticeable for me as I tend to get really bad input lag when I go below 120 fps, which has been happening more and more the past couple patches. edit: in 6v6
  • AbsurdonAbsurdon Germany Join Date: 2015-02-09 Member: 201274Members
    I suffer from lower performance as well. I know it's not recommended but I'm using the D3D11 renderer cause it gives me an avg of ~35 fps (160-200 with mostly round 185) more than the D3D9 one (120 - 170 mostly around 160). or atleast it used to. now it seems even more (tested it a few days ago to see if there is still a huge difference). But still over the last few months (and builds) i've lost about 8-10 fps in avg (looking at it while runing around the map in pregame 6v6). I'd not swear that this lil amount is noticeable for me ingame but it's there. In case u come up with old windows and stuff: I just reinstalled right b4 christmas.
    The increased performance on D3D11 compared to D3D9 renderer might be due to me using windows 10 and a GTX 980Ti so there might be benefits in terms of optimisation or something (dunno the technical details, just describing what is, think i heard something like d3d12 has performance tweaks 4 d3d11 applications). CPU is i7 4770 @~3.6 GHz when ingame btw and it'S the bottleneck in performance terms in my system.
  • SquishpokePOOPFACESquishpokePOOPFACE -21,248 posts (ignore below) Join Date: 2012-10-31 Member: 165262Members, Reinforced - Shadow
    I took it at face value. Well done
  • SupaDupaNoodleSupaDupaNoodle Join Date: 2003-01-12 Member: 12232Members
    xDragon wrote: »
    Your eye can't see past 30 fps anyways.

    But the games must produce that 30fps consistently. Otherwise your eyes notice stuttering on-screen. If your PC can consistently produce 60fps or more, THEN you will perceive a smooth flow of visuals.
  • GoldenGolden Join Date: 2004-09-01 Member: 31169Members, NS1 Playtester, NS2 Playtester, NS2 Map Tester, WC 2013 - Silver, NS2 Community Developer
    xDragon wrote: »
    Your eye can't see past 30 fps anyways.

    But the games must produce that 30fps consistently. Otherwise your eyes notice stuttering on-screen. If your PC can consistently produce 60fps or more, THEN you will perceive a smooth flow of visuals.

    Could it be... he was JOKING?!?!?!?!!?!??!?!?
  • dePARAdePARA Join Date: 2011-04-29 Member: 96321Members, Squad Five Blue
    edited March 2016
    Golden wrote: »
    xDragon wrote: »
    Your eye can't see past 30 fps anyways.

    But the games must produce that 30fps consistently. Otherwise your eyes notice stuttering on-screen. If your PC can consistently produce 60fps or more, THEN you will perceive a smooth flow of visuals.

    Could it be... he was JOKING?!?!?!?!!?!??!?!?

    Well, maybe.
    Or maybe not? who knows.
    I saw so many laughable comments about how many fps you need in countless forums, and these posts where serious.
    Maybe thats why i cant laugh about such a joke anymore.
    Too many people have zero clue what there talking about.
  • LocklearLocklear [nexzil]kerrigan Join Date: 2012-05-01 Member: 151403Members, NS2 Playtester, NS2 Map Tester, WC 2013 - Shadow
    Really? If 14yos on twitch know this myth and troll in chat about it, I think you can expect MOST people to know better and to simply be joking.
  • Soul_RiderSoul_Rider Mod Bean Join Date: 2004-06-19 Member: 29388Members, Constellation, Squad Five Blue
    Believe me, the amount of people in this world who believe that to be true means you have to publicly denounce it every time it is mentioned, joke or not, or more idiots end up believing it.
  • SquishpokePOOPFACESquishpokePOOPFACE -21,248 posts (ignore below) Join Date: 2012-10-31 Member: 165262Members, Reinforced - Shadow
    Trump for president
Sign In or Register to comment.