Performance rating needs fixing

MoFoMoFo Join Date: 2013-09-09 Member: 188047Members
So am I alone in thinking that the incorrect performance rating is causing a problem in NS2?

Servers with downright horrible performance (often dropping below 10%) are populated 24/7/365.. Some of these server owners are even adding more unstable servers...

I think changing the performance to be an average over time, instead of an instant reading, would help immensely as it would help players (especially new ones) to see which servers they should not join. Plus it would help filter active players onto more stable servers. I'm sure I'm not the only one who has been frustrated when the only populated server in an acceptable ping range is one with horrible performance.

When I was new I thought the performance rating was the coolest feature I'd ever seen in a browser. Unfortunately I quickly realized that It was worthless. I'd join servers at 100% only to experience some of the most unstable gameplay I'd ever seen. I almost quit early on due to the frustrating lag and warping issues, but I stuck it out because Alien was so fun to play. (at the time)

Now of course I've learned to click on details and watch it for a few minutes to get an accurate reading, but players shouldn't have to go through these steps. It seems like this would be such a simple change for them to implement, so I'm making this thread as a plea to UWE. Hopefully if enough people agree they will take notice and do something.

Comments

  • Omega_K2Omega_K2 Join Date: 2011-12-25 Member: 139013Members, Reinforced - Shadow
    edited November 2013
    Yes, it should be an avarage, but the with implementation they currently use it would also be easy to lie about the server's performance (as a side note), though I don't think it's happening atm.
  • SamusDroidSamusDroid Colorado Join Date: 2013-05-13 Member: 185219Members, Forum Moderators, NS2 Developer, NS2 Playtester, Squad Five Gold, Subnautica Playtester, NS2 Community Developer, Pistachionauts
    It's already been attempted to be fixed (at least by me) the problem is it's nearly always at 100 and if it does dip below 30 ticks it's only for a half second, so averaging it does no good. Plus, you would have to average every server in the list, which is a LOT of averaging, just think about that. That, and actively tracking it while you are playing, just adds more workload the game has to do. It's not as easy as "oh yeah just average over time that's easy", it's a little more complicated than that. They get the current server rate, divide it by 30 and multiply by 100, which is also not that exploitable.
  • shonanshonan Join Date: 2013-01-28 Member: 182562Members, Reinforced - Shadow
    Omega_K2 wrote: »
    Yes, it should be an avarage, but the with implementation they currently use it would also be easy to lie about the server's performance (as a side note), though I don't think it's happening atm.

    This could be averted by letting players report servers to blacklist if they see them cheating.
  • BensonBenson Join Date: 2012-03-07 Member: 148303Members, Reinforced - Shadow, WC 2013 - Shadow
    edited November 2013
    I have said this every time player retention comes up.

    NS2 already has a reputation for high system requirements, and many new players playing on these (most) servers, may just confuse poor server performance with poor game performance.
  • ezekelezekel Join Date: 2012-11-29 Member: 173589Members, NS2 Map Tester
    SamusDroid wrote: »
    It's already been attempted to be fixed (at least by me) the problem is it's nearly always at 100 and if it does dip below 30 ticks it's only for a half second, so averaging it does no good. Plus, you would have to average every server in the list, which is a LOT of averaging, just think about that. That, and actively tracking it while you are playing, just adds more workload the game has to do. It's not as easy as "oh yeah just average over time that's easy", it's a little more complicated than that. They get the current server rate, divide it by 30 and multiply by 100, which is also not that exploitable.

    listen here mr 800 mb upload, make it happen
  • |DFA| Havoc|DFA| Havoc Join Date: 2009-08-07 Member: 68375Members, Reinforced - Shadow
    Agree 1000% with OP.
  • RoobubbaRoobubba Who you gonna call? Join Date: 2003-01-06 Member: 11930Members, Reinforced - Shadow, WC 2013 - Shadow
    instead of simply averaging, wouldn't it make more sense to hold the lowest value obtained in the last 10 minutes? So if a server drops to 20 ticks, for the next 10 minutes it will have 67% performance (unless it drops even lower). Yes, this requires monitoring, so that's less easy, so you'd need the 'broadcast performance' to be a server-side non-override-able value in addition to current tickrate.
  • VitdomVitdom Join Date: 2012-04-30 Member: 151345Members, Reinforced - Supporter, Reinforced - Silver, Reinforced - Gold, Reinforced - Diamond, Reinforced - Shadow, WC 2013 - Shadow
    edited November 2013
    Roobubba wrote: »
    instead of simply averaging, wouldn't it make more sense to hold the lowest value obtained in the last 10 minutes? So if a server drops to 20 ticks, for the next 10 minutes it will have 67% performance (unless it drops even lower). Yes, this requires monitoring, so that's less easy, so you'd need the 'broadcast performance' to be a server-side non-override-able value in addition to current tickrate.
    Since this also requires saving all sampled values(unless min completely resets to 100% after 10 minutes), I think some kind of statistical averaging system would be better while at it and it eliminates the case where there is only 1 drop to 0%.

    Why not just sample tickrate for like 5 seconds and accumulate the intensity of the drops squared and then SQRT it, kind of how RMS-averaging works, which would work like a rating system for performance. 0 would be best, sqrt(30^2 * 30 * 5) = 367 would e.g. be worst, meaning it is convertible to a 0-100% display. Larger tickrate drops would weigh more than smaller ones. Instead of completely resetting the performance rating to 100% after the update time, it has already calculated the new performance.

    It is more accurate than the current system since it provides a useful weighted-average rating over time, also without having to save all samples.
  • Ghosthree3Ghosthree3 Join Date: 2010-02-13 Member: 70557Members, Reinforced - Supporter
    edited November 2013
    43634-stalin-dear-diary-today-OP-was-2N1J.jpeg

    Unfortunately this has become increasingly rare recently.
    @Ghosthree3 This isnt a image bourd. So remember the spoiler tags when posting images please.
  • NordicNordic Long term camping in Kodiak Join Date: 2012-05-13 Member: 151995Members, NS2 Playtester, NS2 Map Tester, Reinforced - Supporter, Reinforced - Silver, Reinforced - Shadow
    I am not a server operator. If I was a server operator, I would not want my server permanently shunned because it dropped to 20 ticks for a half second once a day. Even the most powerful servers have minor drops, just a lot less often.
  • lwflwf Join Date: 2006-11-03 Member: 58311Members, Constellation
    edited November 2013
    BNS2 already does this. :)

    It may be ugly, but it does everything described so far. It's just a toy though and not exposed in any meaningful way (but you can easily check it on the "server variables" tab on servers here, but it's not frequently updated... nor does it run on any servers with bad performance) such as the server browser... This can't be done by modding, UWE would of course have to implement something like that in the engine. As a mod (or anywhere near Lua) it's trivial for servers to cheat.

    It picks the longest frame (lowest FPS) every 100 ms, makes slow frames weight more (25 out of 30 FPS is just 83.3%, but from a playability perspective it's more like 50%) and retains a memory of past longest frame which it presents as a "performance" value between 0 and 100.
Sign In or Register to comment.