Maintenance for the week of September 1:
• [COMPLETE] PC/Mac: NA and EU megaservers for patch maintenance – September 2, 4:00AM EDT (8:00 UTC) - 9:00AM EDT (13:00 UTC)
• Xbox: NA and EU megaservers for patch maintenance – September 3, 4:00AM EDT (8:00 UTC) - 12:00PM EDT (16:00 UTC)
• PlayStation®: NA and EU megaservers for patch maintenance – September 3, 4:00AM EDT (8:00 UTC) - 12:00PM EDT (16:00 UTC)

PTS and multicore?

  • deyjasagus
    deyjasagus
    ✭✭✭
    Is the multi-core support automatically enabled or is it part of the settings file?
  • ZeroXFF
    ZeroXFF
    ✭✭✭✭✭
    ZeroXFF wrote: »
    People talking about the FPS they get in Alinor are comparing apples to oranges. You have to compare the same area on PTS and live for the most adequate comparison. Go to an area where you always had issues on live and see if it's any different on PTS.

    You cannot compare apple to apple on PTS because population is significant low on PTS compare to live server. we have to wait to see if multi-core made any improvement or not.

    The population only matters within your visibility range. At certain times certain areas are just as empty on live as they are on PTS. And if that's not good enough for you, go to a dungeon. The ICP trap room and the last room of FL (standing in the middle and looking in the direction of the entrance where you jumped in) are fps killers.
  • Casdha
    Casdha
    ✭✭✭✭✭
    I took a pretty hard hit last night and dropped to 1 fps and stayed under 20 for about a half a minute with not many folks around. I was somewhere on Summerset in the first town by the default wayshrine.

    I7 6700 16GB ram and GTX 1070 running at 4k. On Live I generally never drop below 45 fps and usually stay around 60.

    I didn't notice getting much above the high 40s but it may be server load or something on the PTS.
    Proud member of the Psijic Order - The first wave - The 0.016%

  • Moloch1514
    Moloch1514
    ✭✭✭✭✭
    Running 4k on Ultra, noticed a lot of stuttering from mid 50s into low 40s when out in the trees outside Shimmerene. Looks pretty but poorly optimized, just like some of the Vanilla zones.
    PC-NA
  • Rohamad_Ali
    Rohamad_Ali
    ✭✭✭✭✭
    ✭✭✭✭✭
    Remember the PTS server has more limitations then the live server if memory serves me . The actual difference on the live server maybe more noticeable for some .
  • johnnified
    johnnified
    ✭✭✭
    I can definitely notice that load is distributed across all cores, and i'm getting constant 100+(115-120 avg i'd say) fps while running with maxed out ultra settings in Summerset.

    But i'm more about trial fps, can't really say anything without seeing that one.

    I'm running i7 7700K at 4,6GHz, GTX 1080 Ti FTW 3 with 16 gb 3200 mhz memory.

    Edit: Resolution is 2560x1440.

    Edit 2: It occasionally drops down to 70s with an average of 85 in Shimmerene, where there are loads of people atm.

    Hmm, this is very interesting. I would say our configs are fairly similar, though you have a newer i7, my 5930K is clocked at 4.6 GHz. My Titan X (Pascal) is comparable to your 1080 Ti as well. But, in Alinor and Shimmerene I was rarely getting 70 or more; only when fairly isolated.

    I wonder what the difference is. Neither my CPU cores or my GPU were anywhere near maxed out when my fps dropped.

    The difference is your CPU's. Just because you are both running at the same GHz doesn't mean you should expect the same performance. The Per-Core performance is what matters most.

    Don't forget that he never stated if he uses a SSD or normal HDD aswell.
  • Smaxx
    Smaxx
    ✭✭✭✭✭
    Remember the PTS server has more limitations then the live server if memory serves me . The actual difference on the live server maybe more noticeable for some .

    You're talking about server side things, but framerate etc. is purely local (unless you've got a weird netcode system like Blizzard where lag will kill your frame rate).

    To determine what's bottlenecking (ignoring things such as memory bandwidth): Open the game, check the game's framerate. If it's hitting 100 (and vsync is disabled or your screen is faster), then there are ressources left. Otherwiwe check your GPU utilization (Windows 10 with Fall Creators Update in Task Manager; third party tools otherwise). If your GPU is at 100%, the bottleneck. if it isn't and one core of your CPU is on full load (i.e. 100%), the CPU is the bottleneck.

    For most common setups in ESO's case the CPU should pretty much always be the limiting factor.
  • exeeter702
    exeeter702
    ✭✭✭✭✭
    ✭✭
    Smaxx wrote: »
    Remember the PTS server has more limitations then the live server if memory serves me . The actual difference on the live server maybe more noticeable for some .

    You're talking about server side things, but framerate etc. is purely local (unless you've got a weird netcode system like Blizzard where lag will kill your frame rate).

    To determine what's bottlenecking (ignoring things such as memory bandwidth): Open the game, check the game's framerate. If it's hitting 100 (and vsync is disabled or your screen is faster), then there are ressources left. Otherwiwe check your GPU utilization (Windows 10 with Fall Creators Update in Task Manager; third party tools otherwise). If your GPU is at 100%, the bottleneck. if it isn't and one core of your CPU is on full load (i.e. 100%), the CPU is the bottleneck.

    For most common setups in ESO's case the CPU should pretty much always be the limiting factor.

    Framerate is not purely local. Communication between the server and the client can have an effect since the server determines what textures, lighting, characters etc etc have to be loaded and then sent to be rendered, once it is rendering, fps really revolves around the gpu (local), but if enough things are being tasked on the cpu from the server you will take a perdormance hit and since variables are always in play in mmos, cpu is always more burdened than the gpu.
  • Nestor
    Nestor
    ✭✭✭✭✭
    ✭✭✭✭✭

    I wonder what the difference is.

    Addons can have a huge effect on frame rates in this game.

    Edited by Nestor on April 17, 2018 9:55PM
    Enjoy the game, life is what you really want to be worried about.

    PakKat "Everything was going well, until I died"
    Gary Gravestink "I am glad you died, I needed the help"

  • JasonSilverSpring
    JasonSilverSpring
    ✭✭✭✭✭
    Nestor wrote: »

    I wonder what the difference is.

    Addons can have a huge effect on frame rates in this game.

    Agreed, but I have no addons installed for PTS. Also, I am noticing that no core is close to 100% like I would experience in live when I cannot max framerate. Neither CPU or GPU are maxed out. I believe the theory of network delays holding up framerate to ensure smooth viewing of other players may be at issue here. If so, hopefully the network part is more responsive in live.

    If it is this slow in live I will not be happy. I guess I will have to continue to use the Adaptive Settings addon to dynamically reduce draw distance in cities. I was really hoping multi-threaded improvements would eliminate that.
  • SirAndy
    SirAndy
    ✭✭✭✭✭
    ✭✭✭✭✭
    Smasherx74 wrote: »
    ESO only uses 1 core. Always has and always will.

    What a load of BS ...
    poke.gif


    This is from a post of mine from 2016:
    https://forums.elderscrollsonline.com/en/discussion/comment/2988276/p1
    shades.gif


    post-2-1463537041.png



  • infraction2008b16_ESO
    infraction2008b16_ESO
    ✭✭✭✭
    Managed to download the client, did a couple of run throughs of Davon's Watch, Elden root & the new hub.

    https://youtu.be/CBy1wji2G48

    In short despite my bad ping to the PTS being a factor, I'd say there's not a lot of difference with thread utilization. Pretty much the same as on live, I don't think they've actually changed much.
  • profundidob16_ESO
    profundidob16_ESO
    ✭✭✭✭✭
    I can definitely notice that load is distributed across all cores, and i'm getting constant 100+(115-120 avg i'd say) fps while running with maxed out ultra settings in Summerset.

    But i'm more about trial fps, can't really say anything without seeing that one.

    I'm running i7 7700K at 4,6GHz, GTX 1080 Ti FTW 3 with 16 gb 3200 mhz memory.

    Edit: Resolution is 2560x1440.

    Edit 2: It occasionally drops down to 70s with an average of 85 in Shimmerene, where there are loads of people atm.

    Hmm, this is very interesting. I would say our configs are fairly similar, though you have a newer i7, my 5930K is clocked at 4.6 GHz. My Titan X (Pascal) is comparable to your 1080 Ti as well. But, in Alinor and Shimmerene I was rarely getting 70 or more; only when fairly isolated.

    I wonder what the difference is. Neither my CPU cores or my GPU were anywhere near maxed out when my fps dropped.

    your max overclock 4.6Ghz is your limiting factor. I Succesfully predicted the 10-30% reported fps increase in the following thread a while ago:

    https://forums.elderscrollsonline.com/en/discussion/404462/update-18-true-multi-core-cpu-support#latest

    Note that the people that will see any increase are the ones who were previously bottlenecked by the single core performance of their (weak) processors. Especially low and mid range AMD cpu's will be affected most.

    Those with bleeding edge singlecore overclocks will see close to no improvement in fps, even though process monitor will clearly show a better spreading of the load across all cores

    I truly believe this is the best they can do concerning this rather complex problem. For instance the typical classic challenge in mmo's of buffering, prioritizing and synchronizing of network packets to make everything seem to be happening in realtime sync for everyone is a very cpu heavy task that cannot simply be split off into different cores.

    reducing load on the main utilized core is a task 100 times more complex then it would seem to an average player and these results are completely in line with a proper attempt imho

    Makes sense and we'll explained. I think ZOS should communicate the expectations better though. They indicated that the benefit would be best seen on middle to high end PCs.

    My biggest concern was that I was struggling to get 60 fps in Summerset city areas. I was hoping to get at least 60.

    Also, this is at 1440p on ultra-high preset.

    If it helps you, I believe that when ZOS said "benefit will be best seen on middle to high end PCs" they probably meant anything up to high end pc's. In other words: on low and middle. And even that is indeed not accurate as you pointed out because your graphics card doesn't have to be high end in order to reach max fps in the scenario you describe. Specifically in the summerset city area The single-core performance (regardless of whether you play on 1440p or 4K resolution) of your cpu will determine your fps . For instance I'm steady in Summerset between 80fps min and the max cap of 100fps that I deliberately set in place for myself in order to stabilize continuous fps and frametimes.

    In your particular case the only way to raise your fps further in these area's would be to switch to better hardware (cpu and it's cooling)


    Also take into account that right now the new multicore changes on PTS are really in public beta. Everything works but there's an annoying bug where all frametimes and fps drop in a moment where everything seems to hang. On the highest end configuration It's only a 1-2second hickup that occurs every 5-10 minutes randomly but just to show you that definately devs are still going to have to make further changes. When they will troubleshoot this hickup problem and it turns out to be caused by 1 of the tasks they have currently split of from the main core onto the others they may conclude they have to roll back that action and put that load back in the main utilized core...
  • ZeroXFF
    ZeroXFF
    ✭✭✭✭✭
    Managed to download the client, did a couple of run throughs of Davon's Watch, Elden root & the new hub.

    https://youtu.be/CBy1wji2G48

    In short despite my bad ping to the PTS being a factor, I'd say there's not a lot of difference with thread utilization. Pretty much the same as on live, I don't think they've actually changed much.

    This isn't a real benchmark, you have to record identical scenes on both PTS and live and make a side by side comparison. Also, the displayed utilization is inaccurate, you have to look at what the actual threads of the game do, because if your OS is swapping the game threads between 2 different cores within the polling period, even though it takes 100% of a single core, it will show up as 50% on both. The way you did it you won't see a difference that is less than 50%.
  • Jonno
    Jonno
    ✭✭✭
    I have i7 7700k @5ghz
    16gb ddr4 3200
    Gtx 980

    I get 400 fps in the middle of nowhere (up from 200)
    I get 150 ish in busy towns
    I get 50+ in the new trial
    PC / EU
    |Chimaira
    |Thats What She Said
    |Call Of The Undaunted
    |Unfinished Business

    TinkerBell - Orc Stamplar - Tick Tock Tormentor
    Aelin - Bosmer Stamplar - Tick Tock Tormentor
    Prıncess - Dunmer - Mag NB - Divayth Fyr's Coadjutor
    Cher Lloyd - Dunmer - Sorc - Divayth Fyr's Coadjutor
    Luna Lovegøød - Dunmer - Templar - Dro-m'Athra Destroyer
    Leíghton - Redguard - Stam DK - Divayth Fyr's Coadjutor
  • JasonSilverSpring
    JasonSilverSpring
    ✭✭✭✭✭
    Jonno wrote: »
    I have i7 7700k @5ghz
    16gb ddr4 3200
    Gtx 980

    I get 400 fps in the middle of nowhere (up from 200)
    I get 150 ish in busy towns
    I get 50+ in the new trial

    What are you video settings, specifically your draw distance? What resolution?
  • Marabornwingrion
    Marabornwingrion
    ✭✭✭✭✭
    ✭✭✭✭
    I'm really interested about opinions of people who were running trials on PTS.

    Is there any difference in fps and latency, for example vHoF 4th boss on Live vs. PTS with Summerset
  • infraction2008b16_ESO
    infraction2008b16_ESO
    ✭✭✭✭
    ZeroXFF wrote: »
    Managed to download the client, did a couple of run throughs of Davon's Watch, Elden root & the new hub.

    https://youtu.be/CBy1wji2G48

    In short despite my bad ping to the PTS being a factor, I'd say there's not a lot of difference with thread utilization. Pretty much the same as on live, I don't think they've actually changed much.

    This isn't a real benchmark, you have to record identical scenes on both PTS and live and make a side by side comparison. Also, the displayed utilization is inaccurate, you have to look at what the actual threads of the game do, because if your OS is swapping the game threads between 2 different cores within the polling period, even though it takes 100% of a single core, it will show up as 50% on both. The way you did it you won't see a difference that is less than 50%.

    First I never said it was a benchmark but a run through, pretty pointless benching live vs pts because live is going to have more people running around in hubs for a start.

    Secondly if you look at what's happening CPU 3 (2nd core first thread) is what the main bulk of the game engine is running on, all other threads are generally 10-20% utilized. CPU 4 (2nd core 2nd thread) is actually one of the lesser ones because CPU 3 is taking up most of the whole cores utilization.

    Here's an example of what's happening:

    xkZN70c.png

    Not very multicore when you compare it to say battlefield 1 or another DX11 game with good threaded utilization.

    Oh and this is live:

    J9iPDx3.png
    Edited by infraction2008b16_ESO on April 18, 2018 10:35AM
  • Zymcio
    Zymcio
    ✭✭✭
    Hello,

    I made some tests.

    Summary PTS was better like 5 FPS vs Live. (But in PTS is less ppl and no addons)
    Performance

    1.Increased the amount of CPU cores used by the client, which will improve the overall framerate for players on mid- to high-end CPUs. This will be most notable in PvP or graphically intense scenarios.

    2.Improved character compositing, resulting in faster character loads and less framerate spikes.

    3.Fixed issues with character loading that would cause the all black stand-in character models to persist for longer than expected.

    4.Made some improvements to the issue where the framerate could decrease over time, most notably while in Cyrodiil.

    5.Implemented additional fixes to prevent long load times across the game.

    6.Fixed various issues that were causing players in your group to appear invisible.

    1. i dont see any differences in monitoring about CPU (it is working ?)
    2. True but need to be tested in live version
    3. True no more black :P
    4. Dont know need to test
    5. Well no differences again (ESO on SSD)
    6. Dont know need to test

    Most ppl on chat said to me they dont see any differences in FPS...
  • technohic
    technohic
    ✭✭✭✭✭
    ✭✭✭✭✭
    I want to check it out. I know they said dual core and quad core would be minimal. Once I am up and running, I will have a 6 core i7 so Im going to want to run both again and see where its at.
  • ZeroXFF
    ZeroXFF
    ✭✭✭✭✭
    ZeroXFF wrote: »
    Managed to download the client, did a couple of run throughs of Davon's Watch, Elden root & the new hub.

    https://youtu.be/CBy1wji2G48

    In short despite my bad ping to the PTS being a factor, I'd say there's not a lot of difference with thread utilization. Pretty much the same as on live, I don't think they've actually changed much.

    This isn't a real benchmark, you have to record identical scenes on both PTS and live and make a side by side comparison. Also, the displayed utilization is inaccurate, you have to look at what the actual threads of the game do, because if your OS is swapping the game threads between 2 different cores within the polling period, even though it takes 100% of a single core, it will show up as 50% on both. The way you did it you won't see a difference that is less than 50%.

    First I never said it was a benchmark but a run through, pretty pointless benching live vs pts because live is going to have more people running around in hubs for a start.

    Secondly if you look at what's happening CPU 3 (2nd core first thread) is what the main bulk of the game engine is running on, all other threads are generally 10-20% utilized. CPU 4 (2nd core 2nd thread) is actually one of the lesser ones because CPU 3 is taking up most of the whole cores utilization.

    Here's an example of what's happening:

    xkZN70c.png

    Not very multicore when you compare it to say battlefield 1 or another DX11 game with good threaded utilization.

    Oh and this is live:

    J9iPDx3.png

    It may not be very good, but there is definitely an improvement, just like I said in the thread I made about this.

    Also, something is wrong with your task manager display, are you sure you're showing this comparison while in the same place in the game? Because my CPU load doubled on PTS. Furthermore, the PTS client has 60 threads (vs live 41) with more of them doing a significant portion of the work.

    And yet again, the population only matters if the characters are being rendered. Go benchmark in a dungeon if you're worried that a difference of 2-3 players in visibility range are going to have a significant impact (ICP trap room or FL last boss room when looking in the direction of the entrance are FPS killers). But just running around randomly and saying there is "no difference" is very misleading. Like I said, you won't see a difference of less than 50% just looking at it.
  • Jonno
    Jonno
    ✭✭✭
    Jonno wrote: »
    I have i7 7700k @5ghz
    16gb ddr4 3200
    Gtx 980

    I get 400 fps in the middle of nowhere (up from 200)
    I get 150 ish in busy towns
    I get 50+ in the new trial

    What are you video settings, specifically your draw distance? What resolution?

    Max particles max particle distance max view distance sub sampling high everything else lowest im using 4k resolution downscaled to 1080p with DSR
    Edited by Jonno on April 18, 2018 4:46PM
    PC / EU
    |Chimaira
    |Thats What She Said
    |Call Of The Undaunted
    |Unfinished Business

    TinkerBell - Orc Stamplar - Tick Tock Tormentor
    Aelin - Bosmer Stamplar - Tick Tock Tormentor
    Prıncess - Dunmer - Mag NB - Divayth Fyr's Coadjutor
    Cher Lloyd - Dunmer - Sorc - Divayth Fyr's Coadjutor
    Luna Lovegøød - Dunmer - Templar - Dro-m'Athra Destroyer
    Leíghton - Redguard - Stam DK - Divayth Fyr's Coadjutor
  • Zymcio
    Zymcio
    ✭✭✭
    TA DAM

    Same settings Live vs PTS but

    1. SET FPS CAP 120 / USAGE CPU 54% GPU 66%!!

    Live 96 FPS (most time 90)
    Screenshot_20180418_184320.png

    PTS 90 FPS (most time 90)
    Screenshot_20180418_182452.png

    2. SET FPS CAP 200 USGAE CPU 60% GPU 75%!!

    Live 125 FPS (most time 125)
    Screenshot_20180418_185622.png

    PTS 133 FPS (most time 125)
    Screenshot_20180418_190226.png

    Why tell me why ESO dont wanna use 90~100% of CPU ??
  • NewBlacksmurf
    NewBlacksmurf
    ✭✭✭✭✭
    ✭✭✭✭✭
    I noticed higher FPS only when most effects were off.....around 90 FPS but with everything on ultra or very high its maybe 5-10 FPS higher 65-68ish


    i7 8700k
    32 gig of ram
    1 TB SSD HD
    GEForce 1060 (6gig)
    -PC (PTS)/Xbox One: NewBlacksmurf
    ~<{[50]}>~ looks better than *501
  • Zymcio
    Zymcio
    ✭✭✭
    Well summary

    live vs pts = same...
  • MLGProPlayer
    MLGProPlayer
    ✭✭✭✭✭
    ✭✭✭✭✭
    I'm not really noticing a difference on the PTS. 40-60 FPS in towns. 60-144 in open world. GPU usage is still low.

    R5 1600 + GTX 1080 + 16 GB DDR4 2666 Mhz.
    Edited by MLGProPlayer on April 18, 2018 7:17PM
  • Octopuss
    Octopuss
    ✭✭✭✭✭
    I really hoped they finally brought the Vulkan API support, but no :(
    I did notice no general system lag when logging in or out, so there definitely are some improvements.
    No improvement to framerates thus far, though.
    Core i7 3770K

    Smasherx74 wrote: »
    ESO only uses 1 core. Always has and always will.
    Perhaps you shouldn't post about things you have absolutely zero idea about. You look stupid.
  • MLGProPlayer
    MLGProPlayer
    ✭✭✭✭✭
    ✭✭✭✭✭
    Some more detailed feedback.

    GPU usage is around 25-39% at all times (still haven't seen it crack 40%). This is with a GTX 1080 and R5 1600.
  • Zymcio
    Zymcio
    ✭✭✭
    Well i have GTX 1050ti max was 75% then max for someone with GTX 1080 will be like ~38%...
  • Kadoin
    Kadoin
    ✭✭✭✭✭
    ✭✭
    I think the bad framerates and loss of performance have to do with lighting, specifically ray casting. Everywhere my performance suffers, there is buggy (?) ray casting going on. I've actually screenshotted quite a few areas where it is the worst. From what I've seen, the problems occur on :

    A. Fire effects (only fire effects reflected on the ground or a reflective wall/mirror)
    B. Any lighting effect on a reflective surface, adjusting the camera at a different angle alleviates the frame drop
    C. Reflective lighting effect on the character's armor

    Wherever the problems occur though, I've seen buggy textures that look low-resolution compared to the rest of the scene. I'm also playing in ULTRA quality specifically to better identify these places. Playing on low resolution does not improve performance, and I also have one core bloated and the rest evened out.
Sign In or Register to comment.