We have detected that cookies are not enabled on your browser. Please enable cookies to ensure the proper experience.
Page 1 of 2 1 2 LastLast
Results 1 to 25 of 32
  1. #1
    Join Date
    May 2014
    Location
    Poland
    Posts
    229

    Performance Guide and Stuttering Analysis


    Performance Guide


    As you probably already know, LOTRO is suffering from heavy performance issues regardless of hardware used. This was main thing that was keeping me out of the game since many months, but recently I started to miss the game so I decided to take a look at LOTRO performance and see if there is any way to improve performance and make the game playable in newer areas like Rohan and Gondor, because this where serious issues are.

    Here you will find full analysis of LOTRO's performance. The source of performance issues, benchmarks for different graphics settings, stuttering analysis with frametime graphs and finally the conclusion with unfortunately not very positive findings, gently speaking.

    For detailed description on what each setting does visually, visit my Graphics Settings Breakdown.
    If you are curious about the difference in playing LOTRO on SSD, HDD and through Pendrive, check Hard Drive Comparison.


    Introduction


    So first here are some explanations about everything you are going to see in this guide. Test system, testing places, how benchmarks are made and etc.


    Testing Locations

    For benchmarking I've chosen two demanding places of different characteristics.

    First one is Harwick in Rohan, a mid-sized city with a lot of water around it which makes it very demanding place due to a lot of objects and reflections to render. This place known for heavy stuttering and performance issues, related to CPU as you will later.

    Second one are Eaves of Fangorn, very dense forest filled with with foliage, shadows and intensive atmospheric details like intensive lighting or fog. This place on the other hand is very demanding for your GPU.

    With those two places both CPU and GPU bound scenarios are tested to give you full insight on LOTRO performance.


    Testing Methodology

    There is nothing too complex here. I am using FRAPS to measure Frames Per Second and Frametimes while making sure that benchmarking runs are exactly the same, on the same settings, the same time of a in-game day and etc.

    For basic comparisons between different levels of individual settings I will use simple average and minimum framerate metrics made with FRAPS.

    For more in depth look at performance and especially this heavy stuttering that bother players the most, I will use frametimes. Frametime is the time between one frame to another on your display. This is much more complex and precise way of benchmarking that allows to precisely detect all issues that are not detected by simple average and minimum framerate tests.

    This is quite complex topic and I will leave a link to a bit more detailed but still light description, but all you need to know in this thread is this:



    This is an example of perfect frametime delivery, any spikes on the graph are unwanted and are presenting more or less significant stuttering, depending on how big and frequent those spikes are.

    Also, lower frametimes mean more frames per second (shorter time between frames = more frames fit in a second). So in this case lower is better. Lower and consistent.

    For some further explanation on why I am using this watch this video from one of the most reliable benchmarking/reviewing channels up there: https://www.youtube.com/watch?v=uXepIWi4SgM


    Testing Platform

    For testing I am using my PC with following specs:

    LG 34UC98-W
    i7 4820K 4.7 GHz
    2x MSI GTX 1080 Gaming 8G 2076/5500 SLI (only one GPU used for testing)
    MSI X79A GD65 8D
    2x4 GB 2400 MHz CL10 DDR3 RAM
    Crucial BX100/MX100 SSDs

    As you can see this is not average PC, so I needed to make some adjustments:

    Adjusting CPU is no brainer. With all the tools available I can bring my CPU to like tablet level of performance. I can turn off cores and decrease frequency as much as needed, to the point where I would end up with single core at 1.2 GHz.

    Adjusting GPU is a bit harder and needs some explaining. GTX 1080, even downclocked and underpowered as much as software safely allows is still presenting the performance a bit higher than cards like GTX 980, R9 390X, GTX 1060 or RX 480, which are also way above average, especially for MMO game that is played by people with very different setups. However my resolution is 3440x1440 and here you get only 60% of performance you would on 1080p. So after downclocking, cutting power target and taking resolution into account, all the GPU bound measurements that you will see in this thread are approximately equal to playing at 1920x1080 resolution with GTX 1050 Ti/GTX 960/R9 380.
    Last edited by Unthariel; Feb 23 2017 at 10:26 AM.

  2. #2
    Join Date
    May 2014
    Location
    Poland
    Posts
    229

    So what's wrong?


    The best thing to do when you encounter performance issues is to determine the source. This way you will know what is causing those issues, for example if it is your hardware or game/application you are trying to use.

    We will begin with very simple test in Harwick. Here are I am benchmarking the game using my GPU on three different levels of performance. Clocked at 2076 MHz which is my overclock (100% performance), 1911 MHz which is my factory default (88% performance), and 1101 MHz which is the lowest possible I could achieve with available tools (61,3% performance). As you can see different levels are performing very differently, so if GPU was my limitation, I would see very clear differences between them.



    And as you can see there no differences between those 3 very different levels of GPU performance, which means that GPU is not my limitation here.

    So let's look at CPU performance now. In this test I am Setting Affinity for lotroclient.exe process in Task Manager to different number of CPU cores. If CPU was my limitation, I would see a decrease in performance from decreasing the maximum number of cores that game can use.



    As you can see this is not a case either. There is no difference between four and two CPU cores, the differences are within the margin of error. Only decreasing the number of cores to only one gives any difference.

    This means that limitation lies in the application because it is unable to utilize your resources. LOTRO's very old game engine is able to use only two cores and is far from doing that efficiently because 80,6% of performance and adding another core results only in 19,4% performance increase. Adding more cores does nothing. This is already not a good sign because it means that we are significantly limited by software and this is never ending well.

    All right, so maybe decreasing graphics settings will help? Let's see:



    As you can see there are differences. But why not significant differences? You only need to downgrade your settings to High to get over 60 FPS, right? Well, no. Here is why:



    As you can see, frametime delivery is horrible. All the spikes on the graph are representing stutters. Bigger spike = bigger stutter. The biggest one is exceeding 120 ms, which means that our actual minimal framerate is not much higher than 8 FPS (1000/120=8,333).

    Stuttering is not solved even with Low settings:



    Here we are much closer to the target framerate, but still the minimum is around 50 ms, which means that actual minimum framerate is 20, not 65 like basic framerate measurement was showing. This is exacly why frametimes are important.

    You can see the depressing graph made on Ultra settings here.

    So, 8-20 FPS performance on high-end system with i7 processor overclocked to the max and top-end GPU? Not very positive. So let's look at something less depressing.
    Last edited by Unthariel; Feb 23 2017 at 10:25 AM.

  3. #3
    Join Date
    May 2014
    Location
    Poland
    Posts
    229


    Now it's time for some testing in Fangorn. Just like before, let's look at GPU performance scaling first:



    As you can see this time we get clear difference between high clocked profiles and downclocked one. 1101 MHz profile is showing 76,6% performance of 2076 MHz setting, so still not 61,3% like it should in perfectly GPU bound scenario, but at least we are not completely engine limited here.

    Let's look at quality presets:



    As you can see situation is looking much more positive here than in Harwick. We are getting very good scaling between settings and we are able to achieve very high average framerates, as well as minimal ones.

    But we already know that basic measurements are not showing the truth about performance, so let's look at frametimes:



    As you can see this graph is looking way better than one from Harwick. On High settings vast majority of measurements is meeting our 60 FPS target. Still there are some spikes up to 25 ms, so minimal framerate is around 40, not 79 like basic measurement is saying, but those spikes are rare and general performance is acceptable.



    Even on Ultra settings everything is reasonably tightly timed and most of fluctuations are caused by not enough power in significantly underclocked GPU that is balancing on the edge of 60 FPS target. Only one enormous stutter happened.



    On Low settings graph is looking almost perfect.

    Okay, so there may be some hope after all. We have determined the source of our performance issues and made few detailed benchmarks on general graphics quality presets, not it's time for benchmarking individual graphics settings.


  4. #4
    Join Date
    May 2014
    Location
    Poland
    Posts
    229

    Graphics Settings Benchmarks


    So let's start benchmarking individual settings. I have chosen the most impactful settings based on initial tests and my experience with the game.


    Object Draw Distance

    This is one of the most demanding settings in the game, in both CPU and GPU bound scenarios.



    As you can see this setting has very significant impact on performance in Harwick. Decreasing it will be essential for improving performance there.



    Fangorn sees relatively smaller performance gains. Also decreasing this setting in wooded areas will result in low quality pixelized imposter trees in your close viewing distance, so this is not where you should look for extra performance unless you are really desperate.


    Surface Reflections

    This is another very demanding setting, especially for CPU bound scenarios.



    Another setting to watch out for in Harwick. In general in many settlements disabling this setting is essential for improving performance. Reflections have very high cost in performance regardless of how many of them are in your view. Even small puddle or two in Bree Town can cause huge performance drop.



    Again Fangorn gains less. Visually this is also not a setting you want to disable because disabling Reflections in dark areas or during the night makes deeper waters look like black holes.

    Optimal setting for both Object Draw Distance and Surface Reflections is High, as long as you are outside of CPU bound scenarios. Unfortunately those two settings are practically the only ones that can affect your performance in CPU bound scenarios in significant manner. Decreasing others is much less effective. Visually it is hard to notice the difference between High and Ultra High for Surface Reflections. Visit Graphics Settings Breakdown for details. Object Draw Distance is the most essential in wooded areas, but for settlements and typical landscape you are not loosing all that much by keeping it on High.


    Antialiasing

    This is purely GPU bound setting.

    If you are CPU bound like in Harwick, you can use it freely if your GPU can handle it. It is not stressing for CPU.

    For Fangorn, where GPU matters, this setting has huge impact on performance:



    If your GPU is struggling, this is the first thing to decrease. For game graphics enthusiast who is looking at smaller details the difference from using antialiasing is huge, but for average MMO player it may not be that visible. Decreasing it shouldn't hurt. Also keep in mind that this setting will completely slaughter the performance on slow PCs and non-gaming laptops, so you shouldn't use it on such setups.

  5. #5
    Join Date
    May 2014
    Location
    Poland
    Posts
    229

    Landscape Draw Distance

    This is third and basically last setting that can help you in CPU bound scenarios:



    As you can see there is a lot to gain from decreasing this setting to Very Low in CPU bound scenarios like Harwick. Unfortunately this setting is very essential for visuals. One of the best things in LOTRO are those huge open landscapes that really make you feel the size of the world and decreasing Landscape Draw Distance means huge hit to your visuals. However, in settlements, during instances, raids or Epic Battles where you are focusing on your closest surroundings anyway, you can decrease this setting without a huge hit to visuals to gain so desperately needed CPU performance.



    Gains in Fangorn are minimal which only proves that this setting is purely CPU intensive. Also there you are surrounded by trees and you cannot see distant landscape anyway, so if you are desperate for those 2-3 FPS, you can probably decrease this setting without any hit to visuals.


    Texture Detail

    Another purely GPU bound setting.

    There were no differences from changing this setting in Harwick. It is not affecting CPU, so if your GPU is good enough, max it out freely.

    However in Fangorn there are significant differences:



    Texture quality is obviously essential for visual quality anywhere you go and you shouldn't decrease it unless you have to. Some small decrease may not be too painful if you have small screen, but in general decreasing texture quality should be the very last desperate move.


    Texture Filtering

    Same as Texture Filtering, this is setting that is relying on your GPU.

    There were no differences from changing this setting in Harwick. It is not affecting CPU, so if your GPU is good enough, max it out freely.

    In Fangorn there are few frames to gain:



    Differences between all of those filtering options are easily detectable by enthusiast eye, but not really by average player. Sharp and Very Sharp are visibly better than Bilinear and Trilinear and this should be visible to anyone, but superiority of Anisotropic Filtering is not so easy to see. On low resolution screens Very Sharp may even look better.

    One important thing to keep in mind is that Anisotropic Filtering is handled much worse by AMD cards than Nvidia ones, so you should see much bigger gains from decreasing Texture Filtering to Sharp or Very Sharp on AMD GPUs.


    Texture Filtering

    Same as before, GPU thing.

    There were no differences from changing this setting in Harwick. It is not affecting CPU, so if your GPU is good enough, max it out freely.

    Fangorn is showing some gains:



    Normally this setting is essential, especially in wooded areas, but Fangorn is so dense that the surface there is basically one big shadow, which means that you won't be able to see those dynamic shadows for most of the time. On lower spec PCs and laptops this setting will be very taxing so use it with caution.

  6. #6
    Join Date
    May 2014
    Location
    Poland
    Posts
    229

    High Quality Lighting

    Another GPU bound setting.

    There were no differences from changing this setting in Harwick. It is not affecting CPU, so if your GPU is good enough, max it out freely.

    In Fangorn this setting requires some power from your GPU:



    You will have to visit Graphics Settings Breakdown for details but High Quality Lighting is enabling Specular Lighting and Landscape Lighting Quality and all of them combined are giving very significant boost to visual quality and you should look for extra performance elsewhere.


    Material Detail

    This setting in GPU oriented.

    There were no differences from changing this setting in Harwick. It is not affecting CPU, so if your GPU is good enough, max it out freely.

    In Fangorn this setting is not very demanding:



    Again Graphics Settings Breakdown for details, but this setting is very essential for the look of trees and should never be decreased.


    Frill Distance

    Apparently GPU bound setting, however it affects loading times quite significantly and it may take few seconds until you get proper framerate after logging in (until frills are loaded).

    There were no differences from changing this setting in Harwick. It is not affecting CPU, so if your GPU is good enough, max it out freely.

    Even in Fangorn costs of this settings are rather marginal:



    Only turning this setting off gives any sensible gains, but this means turning off the grass completely and having blank ground as a surface which you obviously don't want. Keep it as high as you can and disable only in the biggest desperation.

    I saw many posts on this setting increasing stuttering while riding war-steed and generally in Rohan and Gondor, more on that in movement stuttering analysis.


    Post Processing

    Last settings on the list, also GPU intensive..

    There were no differences from changing this setting in Harwick. It is not affecting CPU, so if your GPU is good enough, max it out freely.

    In Fangorn this settings has massive impact on performance:



    I find it hard to explain what happened here. This setting enables Surface Reflections, but I already had them turned off for testing in Fangorn to move CPU bottleneck a bit further with framerate. It enables Bloom and Blur, but I already tested those and they were giving no differences, also you can see that those affects are not heavily used in Fangorn. But anyways, this setting is very important visually as you can see in Graphics Settings Breakdown but definitely something to turn off on very low spec machines that are desperate for any performance.



    So that's it for Graphics Settings Benchmarks. The most impactful settings have been tested for both CPU and GPU intensive scenarios to fully investigate what you can do to improve performance.
    Other settings were showing very small differences in performance so benchmarking them was pointless. Although keep in mind that it may vary depending on your hardware.

    If you are struggling for CPU performance in settlements or during instances, which we all are because there is no CPU available in the world with so much single-threaded performance to fix all of those extreme performance issues in LOTRO in CPU bound scenarios, you should look at Object Draw Distance, Surface Reflections and Landscape Draw Distance. But don't expect miracles, even high-end CPUs overclocked to the maximum are falling short badly here, no exceptions. Or rather I should say that LOTRO's engine is falling short in delivering acceptable performance even on high-end systems.

    If you are lacking GPU performance, options for decreasing settings are almost countless. There are many impactful settings that can be decreased without huge hit to visuals. You should start from Object Draw Distance, Texture Filtering and Antialiasing.

  7. #7
    Join Date
    May 2014
    Location
    Poland
    Posts
    229

    Movement Stuttering Analysis


    One thing that can drive anyone crazy in games, even those who do not have any expectations or standards for framerate and smoothness, is stuttering. Low but consistent framerate is somewhat bearable and can be used to with time if necessary, but stuttering is something that can really make people throw their PCs or monitors out of the window.

    So in this part of a guide we are going to take a look at stuttering specifically with frametime only analysis to see if there are any particular settings that are causing or at least increasing stuttering in intensity or frequency. Also there will be some tests on movement speed, if it is increasing this unwanted effect.


    Movement Speed

    First let's look at on foot movement speed, walking and running through Harwick:



    As you can see, graphs are looking similar. And this is very good thing, you will why below.

    Now let's look at how movement speed in affecting performance on War-steed:



    This time tests are made around the city, not inside, and you can already see big improvement in frametimes.

    But again, graphs are the same.

    And now here is why this is good. Decreased movement speed means that completing the test run will take much longer. In case of Walking versus Running on foot, Walking test took 2.5 times longer than Running one. In case of Walk vs Gallop on War-steed, exactly 2 times longer. Which means that Walking graphs should be 2-2.5 times wider. If graphs are looking the same, this means that while walking we got the same amount of stutter but over the course of 2-2.5 longer time, which means 2-2.5 reduced stuttering frequency. Of course this is not fixing anything, stuttering is still happening and even in the same places for the most part, you are just increasing the time you need to get from one stuttering point to another, which decreases stuttering frequency and makes gameplay feel smoother. Again, this is nowhere near fixing anything, more like lying to yourself, but if you cannot change the world then the only thing that is left is to pretend.


    Stuttery Settings?

    Thanks to previous test we already know that stuttering is happening in the same places regardless of your movement speed. So now it is time to check if you can affect stuttering with graphics settings while maintaining the same movement speed.

    Settings that have a lot to do with object/landscape loading are Object Draw Distance, Landscape Draw Distance and Frill Distance. Also Model Detail, but this one is not showing any changes to performance.

    So here are few tests, also from around Harwick, first on Ultra with Object Draw Distance and Surface Reflections on High, then three tests with three settings mentioned above reduced to minimum individually, and at the end final test with all three settings reduced to minimum at the same time:



    I have to say that the consistency of those graphs is very satisfying. All major stutters are happening in exactly the same moment. Unfortunately regardless of settings used. Only decreasing all three settings at once gave marginal difference and few of out many major stutters are now inside the scale and around 50-60 ms. Not like it changes something, those are still heavy stutters.


    I think we are done with collecting data about performance, graphics settings and general game behavior, now it is time to use all of that to make our desperate attempt to fixing performance



    Slow Hard Drive?

    Here I a took a ride from West to South Bree Gate on SSD and HDD to check if there is any difference in stuttering between two drives.



    As you can see the difference is very significant. On HDD there are over 3 times more major stutters than on SSD. This is a huge difference.

    Unfortunately I don't have anything faster than typical SSD to test it further, but I tried to stress my SSD with benchmarking tool and playing the same time, and there was no difference in stuttering, so I assume that reduced stuttering on SSD vs HDD is not caused by Write and Read rates, but by much smaller access time. It would be interesting to try the game from RAM Disk, but I have only 8 GB of RAM right now and I will be upgrading my platform once Intel X299 is out, probably late summer 2017.



    FPS Lock

    Just a quick test to check if locking your FPS can help:



    And as you can see it doesn't help at all.

    Last edited by Unthariel; Feb 28 2017 at 08:59 AM.

  8. #8
    Join Date
    May 2014
    Location
    Poland
    Posts
    229

    Fixing The Performance


    Now it is time to sum everything up and use it in practice.



    Maximizing CPU Performance

    As we learned at the beginning, LOTRO cannot utilize your CPU resources properly and is not using much more than single CPU core.

    If your hardware cannot be fully utilized, you should maximize the performance of the part of hardware that can be used. In this case you need to overclock your CPU:



    As you can see, in LOTRO gains from CPU overclocking are enormous. Gains in performance are almost linear to increase in frequency. For example by increasing CPU frequency from 3.5 GHz to 4.7 GHz, by 34%, we gained over 27% of average performance. Minimum framerate is improved by 50%. We already know that gain in minimum framerate doesn't mean much because stuttering is happening regardless of you hardware and settings used, but CPU overclocking is definitely essential for achieving good performance in LOTRO.

    RAM frequency is also very important:



    As always additional memory frequency comes in handy during CPU intensive scenarios, in this case giving us over 16% increase in performance, despite the myths spread around the web that memory frequency has little to no difference which misinforms players and make them choose inferior components that are limiting potential of their PCs.

    Now final graph on this matter showing the difference between non-overclocked CPU with basic frequency on RAM and overclocked CPU with high frequency on RAM:



    You can see that the difference is just as huge as any difference from overclocking can be. Overclocked CPU paired with fast RAM is performing 42% better than stock clocked CPU paired with slow RAM.

    Important: Keep in mind that memory timings are very important too. This memory frequency comparison was made on the same timings, so only frequency was changed.

    Of course if you are GPU limited you can also overclock your GPU. Gains won't be that huge, usually from 5 to 20% depending on GPU architecture used, but it is recommended to take the most out your hardware.

    If CPU overclocking is not available to you, still you should make sure that you are taking the most of your machine. Search the web for any ideas how to improve CPU performance on your particular PC, turn off any power saving features and set all power options for maximum performance, close unnecessary applications that may take CPU performance away from you if you own CPU with only two cores, check your temperatures and see if there is any thermal throttling, especially if you own a laptop, and if there are any issues with temperatures then give your laptop to service for cleaning and replacing thermal paste.



    Tweaking Graphics Settings

    Once your hardware performance is maximized, you can start tweaking graphics settings.

    The most impactful settings were already mentioned and explained, so here is just a short summary.

    The most CPU intensive settings are:

    Object Draw Distance
    Surface Reflections
    Landscape Draw Distance


    Decrease them if you are suffering from performance issues in settlements or during instances.

    The most GPU intensive settings are:

    Object Draw Distance
    Surface Reflections
    Antialiasing
    Post Processing Effects
    Texture Detail
    Texture Filtering
    DX10 Dynamic Shadows
    High Quality Lighting
    Material Detail


    Decrease them if you are suffering from performance issues on landscape or in wooded areas.


    Decreasing Movement Speed

    As we learned in stuttering analysis part of this guide, heavy stuttering is happening regardless of settings or hardware used, but stutters are always happening in the same places. If you move slower, it will take more time to get from one stuttering point to another, which will make the game visually smoother because of reduced stuttering frequency. Again, this is not a fix to stuttering, just decreasing negative effects of something that cannot be changed anyway. So don't move at full speed at all times and don't hit everything on your way, it will not only improve your's and other player's immersion, but you will get much less frequent stuttering.


    Using Fast Hard Drive


    As you saw in testing, SSD reduces stuttering very significantly, over 3 times. For LOTRO specifically, upgrading from HDD to SSD is a true game changer.


    Conclusion

    So here it is, detailed analysis on LOTRO performance and on what you can do to improve it.

    LOTRO is really an extreme example of poor performance, especially in Rohan and Gondor. Framerates are low and stuttering is close to unbearable. As you saw at the beginning of this guide I am using two GPUs in SLI, so I am dealing with not the most consistent framtime delivery and some microstuttering on the daily basis, and this was the compromise I was willing to take to be able to play my games at their full visual potential, so it is not like I am requiring perfect smoothness, but what is happening in LOTRO is really, really extreme.

    Engine is really old and limited. Paired with irresponsible development for Rohan and Gondor without taking those limitations into account, it creates the situation where game performance is so poor that it is close to unplayable for players with any performance standards and requirements.

    Another issue is what this game actually requires from your hardware. If you read system requirements for LOTRO, you will find a list of old and slow components that are supposed to be able to play this game well. While in fact, as you saw in my testing, there is not much playing this game without expensive overclocked CPU with high frequency RAM, which also require good motherboard and cooling. And knowledge that average MMO player definitely doesn't have.

    So can you make LOTRO run well? No, regardless of hardware and settings used. Can you make it playable? It's relative. Can you improve performance? Yes, in many ways, but effects are limited and all major issues still persist.



    So what PC do you need?

    At the end I will leave a short guide for building PC for LOTRO specifically. What is important, what to choose and why.

    CPU: Intel i3-7350K or i5-7600K
    For CPU you need to focus on single-threaded performance. My 4.7 GHz overclocked 4820K you saw in testing is achieving around 165 points in single-thread test in Cinebench benchmark. The latest Intel Kaby Lake processors can do a bit over 200 when overclocked, so there is a healthy amount of performance to gain. They are presenting much better clock speeds than previous generations and increasing the common overclock from 4.4-4.5 GHz to 4.9-5.0 GHz, so this is your way to go for LOTRO. Upcoming Ryzen CPUs will feature more threads but single-threaded performance will be lower.

    GPU: Adequate to your screen resolution
    GPU power needed is dependent on your resolution. LOTRO is not using a lot of VRAM so any 2 GB card will be very enough. Performance wise, you probably should make a typical choice for your resolution based on reviews and recommendations from reviewing/benchmarking sites and channels.

    RAM: As fast as you can
    For Kaby Lake CPUs it will be DDR4 memory. Find the highest clocked one with the lowest timings that you can fit in your budget. 8 GB is very enough for LOTRO, so buy 2 stick of 4 GB to run the in dual channel.

    Motherboard: Z170/Z270 motherboard with support for RAM frequency of your choice
    Z170 motherboards may require BIOS update for which you will need previous generation CPU, so better stick to Z270. Make sure that it supports RAM frequency that you chose.

    Hard Drive: SSD or Pendrive
    Check Hard Drive Comparison for more, but switching from HDD to SSD will give you considerable boost to comfort of playing thanks to much faster loading times and first of all hugely reduced stuttering frequency. Fast Pendrive is also a good and easy alternative that performs very similarly to SSD when in comes to in-game loading times. Running the game on HDD is not recommended because of much more stuttering than on faster hard drives.
    Last edited by Unthariel; Aug 06 2017 at 09:32 AM.

  9. #9
    Join Date
    May 2014
    Location
    Poland
    Posts
    229
    Updated with SSD vs HDD and FPS lock stuttering analysis.

  10. #10
    Join Date
    Feb 2007
    Location
    USA
    Posts
    5,848
    It would be interesting to try the game from RAM Disk, but I have only 8 GB of RAM right now and I will be upgrading my platform once Intel X299 is out, probably late summer 2017.
    I once tried LotRO in a RAM drive. Though I did not benchmark, to my eye there was no noticeable difference between running it from RAM and running it from SSD. I was surprised.

    I think the sub-1ms access time of just about any SSD has done about all that is possible to do where stuttering/hitching are concerned.

    Incidentally, I didn't notice any difference when moving from a SATA SSD to a PCI-E NVMe SSD either.

    As always, thanks for taking the time to put numbers to all our guesswork and shenanigans.
    "I went to the trouble and expense of driving my old Honda Civic into a lake to try to prove to you that it won't float. . . only to have you respond: 'Hmmm, interesting. . . would you be willing to drive a Honda Accord into the lake?'"

  11. #11
    Join Date
    May 2014
    Location
    Poland
    Posts
    229
    Quote Originally Posted by Hurin View Post
    I once tried LotRO in a RAM drive. Though I did not benchmark, to my eye there was no noticeable difference between running it from RAM and running it from SSD. I was surprised.

    I think the sub-1ms access time of just about any SSD has done about all that is possible to do where stuttering/hitching are concerned.

    Incidentally, I didn't notice any difference when moving from a SATA SSD to a PCI-E NVMe SSD either.

    As always, thanks for taking the time to put numbers to all our guesswork and shenanigans.
    Thanks for reply, I was really curious about that. I have to say that I am quite surprised with how much of a difference in stuttering frequency there is with SSD vs HDD, I didn't expect over 3x difference.

    As for the rest, I am not reading much of this forum, I have been out of here for months now, but I know how many myths are going around, so busting all of this with numbers is certainly useful, at least in theory, because I don't really believe that people will now suddenly listen to reason. I really don't like this word, but performance in games and tech in general is a closed world with definitive rules, there are certain metrics that have certain effects and there are strict standards to fulfill, if something is not performing then it is not performing, period, it's not like in the real world where everything is relative and nothing is true.

  12. #12
    Join Date
    Jun 2010
    Posts
    2,322
    Quote Originally Posted by Unthariel View Post
    As for the rest, I am not reading much of this forum, I have been out of here for months now, but I know how many myths are going around, so busting all of this with numbers is certainly useful, at least in theory, because I don't really believe that people will now suddenly listen to reason. I really don't like this word, but performance in games and tech in general is a closed world with definitive rules, there are certain metrics that have certain effects and there are strict standards to fulfill, if something is not performing then it is not performing, period, it's not like in the real world where everything is relative and nothing is true.
    "Oh but I never, ever lag! The game performs just fine in Ultra!!!" -forum poster who has never gotten past Dunland/Rohan.
    Chromite/Grumbletocks/Grumbletunes/Schrade on Landroval, Chromite on Arkenstone, Appendage on Brandywine

  13. #13
    Join Date
    May 2014
    Location
    Poland
    Posts
    229
    Quote Originally Posted by ChromiteSwiftpaw View Post
    "Oh but I never, ever lag! The game performs just fine in Ultra!!!" -forum poster who has never gotten past Dunland/Rohan.
    That's not the worst thing, theoretically you cannot blame them that they haven't been to Rohan and Gondor yet, at least as long as they are not making definitive statements about how the game is running without testing it throughout. All of this stuttering talk may be hard to understand for someone who is in for example Evendim or somewhere else where performance is almost perfect. The worst thing is when people with no clue are starting spread their reasonless assumptions around, I don't mean guessing or such, but making definitive statements with no proof and no testing whatsoever.

  14. #14
    Join Date
    May 2014
    Location
    Poland
    Posts
    229
    Thread disappeared from forum again, refresh.

  15. #15
    A few things:

    1 Was wondering about effect of changing priority of LOTRO to "high' in Windows task manager.
    2 SSD or RAM disks are nice but-- I thought this is primarily what Texture Cache Size is supposed to help with. Putting more helpful stuff in RAM for fast access...? I've seen guides where people set it to zero ==> perhaps because HDD loads texture *more often* leading to increased stuttering...??? I'd love to see a little bit about this.
    3 I've also seen guides which recommend switching to DX10 or DX9 to improve performance. Would be interesting to see if that is a red herring.

    Windows 7
    AMD Phenom dual core 3.2ghz
    8gb RAM 800mhz
    Radeon 6950 1gb <== faster than CPU
    1Tb SATA drive
    Last edited by MagicVoice; Jul 10 2017 at 12:01 AM.

  16. #16
    Join Date
    May 2014
    Location
    Poland
    Posts
    229
    Quote Originally Posted by MagicVoice View Post
    A few things:

    1 Was wondering about effect of changing priority of LOTRO to "high' in Windows task manager.
    2 SSD or RAM disks are nice but-- I thought this is primarily what Texture Cache Size is supposed to help with. Putting more helpful stuff in RAM for fast access...? I've seen guides where people set it to zero ==> perhaps because HDD loads texture *more often* leading to increased stuttering...??? I'd love to see a little bit about this.
    3 I've also seen guides which recommend switching to DX10 or DX9 to improve performance. Would be interesting to see if that is a red herring.

    Windows 7
    AMD Phenom dual core 3.2ghz
    8gb RAM 800mhz
    Radeon 6950 1gb <== faster than CPU
    1Tb SATA drive
    1. Honestly, I was trying to use this option since forever, on different PCs, from complete potatos to current monsters, and with different games. I haven't seen any benefit of that in my life, not even once.
    2. Maybe it does something. Problem is that it is not giving immediate effect and testing over time introduces too many variables. But it doesn't seem to have any affect on visual quality so maybe it is good to keep it at 0 just in case.
    3. There is no difference in performance between DX9, 10 or 11 on the same settings. The difference comes from settings available. DX10 doesn't have DX11 Interactive Water and DX11 Ambient Occlusion, and DX9 doesn't have DX10 Dynamic Shadows, DX10 Distant Landscape Lighting and something else I don't remember, and of course it doesn't have anything from DX11. So if you set for example Ultra preset on all three APIs, you are not getting the same settings. But those settings that are missing in DX10 and DX9 are GPU oriented, disabling them can help only on PC with slow GPU. It won't help with the major flaws of the game described in this guide, as it is proved that major stutters are occuring in exact the same moments regardless of settings used.

  17. #17
    Join Date
    Jul 2016
    Location
    UK
    Posts
    1,502
    The only cpu related option that works ( a bit) is outside of the game; setting the game to use a specific core or pair of cores; you get to pick whichever cores the OS and background tasks arent using, which gives you an extra 5-15% cpu time, a slightly reduced amount of cpu induced stutter and slightly smoother game-play. However most stutter is server connectivity issues, so the improvement is minor, subjective, and only apparent when the cpu core is running at 80% or above.

    As mentioned above, DX9 makes the game slightly smoother, at the cost of some eye candy; TBH, the only place I notice the lack of DX11 is in Mirkwood, where the distant trees are awful blocks of green; for some reason the tree in Lothlorien dont suffer the same awful effect.

    The server situation is the cause of 99% of all issues, the "new" servers dont run the game very well, and the fact they are all in the same physical location means that 90% of players have a sub-optimal amount of lag in reaching them, which magnifies the server induced lag.

    As long as this is true, there is going to be no real improvement in how the average player experiences the game; it might run perfectly well for a dev sat in an office next to the server farm, but they want to ship them 4,000 miles away, so they can see what it is like for the rest of us.

  18. #18
    Join Date
    Jun 2010
    Location
    Alabama
    Posts
    6,046
    Quote Originally Posted by MagicVoice View Post
    A few things:

    1 Was wondering about effect of changing priority of LOTRO to "high' in Windows task manager.
    2 SSD or RAM disks are nice but-- I thought this is primarily what Texture Cache Size is supposed to help with. Putting more helpful stuff in RAM for fast access...? I've seen guides where people set it to zero ==> perhaps because HDD loads texture *more often* leading to increased stuttering...??? I'd love to see a little bit about this.
    3 I've also seen guides which recommend switching to DX10 or DX9 to improve performance. Would be interesting to see if that is a red herring.

    Windows 7
    AMD Phenom dual core 3.2ghz
    8gb RAM 800mhz
    Radeon 6950 1gb <== faster than CPU
    1Tb SATA drive
    Texture cache size is exactly what it says it is. There is a fixed minimum size. There is a fixed maximum size. Set to zero, the size is fixed. This may help some people with crashing due to memory usage at the expense of more disk accesses that happen more frequently. The game default for this is 0.5, which lets the cache grow from the minimum size to half-again its minimum size. If you have enough RAM, this increased memory usage lets you hit the disk less. So if you have a slow disk and a lot of memory, you may get better performance.
    "No sadder words of tongue or pen are the words: 'Might have been'." -- John Greenleaf Whittier
    "Do or do not. There is no try." -- Yoda
    On planet Earth, there is a try.
    Indeed, in a world and life full of change, the only constant is human nature (A is A, after all :P).
    We old vets need to keep in mind those who come after us.

  19. #19

    Game Engine Speed

    Thanks for all replies.

    Does decreasing game engine speed (i'm guessing this is a CPU load), improve performance / reduce stuttering?

  20. #20
    Join Date
    Sep 2010
    Posts
    226
    Quote Originally Posted by MagicVoice View Post
    Thanks for all replies.

    Does decreasing game engine speed (i'm guessing this is a CPU load), improve performance / reduce stuttering?
    At least it won't improve your FPS, it gets worse when you lower game engine speed so imho make sure your game engine speed is maxed.

    What bothers me most, I just upgraded my GFX card to Vega 56 and it just basicly cruising at idle when all the settings are maxed because this game is so CPU bound. GPU clock is staying at 300-400MHz and usage tops sometimes to 60% with thouse clocks MAx clock with this card at stock is around 1400MHz so there is lot of resources available. Only place where I see 100% usage is in character selection screen.

    So what this game needs is really really high overclocked CPU. My i5 4670K is runnning at 4,3GHz so 900MHz OC and it cannot provide studder free gaming experience with lotro. When I was running stock 3,4GHz it was lot worse ofcourse.
    Thing is that there ain't any CPU what this game would require to give you good gaming experience and utilize your expensive GFX card.

    Game engine would require serious rewrite at this point to fix all performance issues it currently has and I'm afraid this is not gonna happen for lotro.

  21. #21
    Join Date
    Jul 2016
    Location
    UK
    Posts
    1,502
    In my thread (quite a while back now), I suggested the reason LOTRO is so bad on "newer" cpu's is the changes in cpu architecture, and highlighted the physical differences between the AMD Phenom 2 range and the AMD FX range to explain why the FX sucks at LOTRO.

    I have found out something else; in fact I am a bit embarrassed at how I missed it; the Phenom 2 has HUGE L1 and L2 cache sizes compared to the FX, which severely restrict how much data you can cram through it; no wonder you need a 1GHz over clock to get anywhere near the equivalent P2 performance!!

    This might also explain how another forum members experiments "disproved" my theory; by underclocking his i7, he wasnt really comparing like for like, the i7 cache memory access speed would have stayed the same, so wouldnt have hit any limits on how fast it fed the data to the cores. (His average frame rate results were also skewed by the fact his huge gfx card gave a huge max fps, but the minimum was exactly the same - so of course his average was higher).

    I havent had much time to read up, but comparing Core 2 Duo and current "i" series architecture may throw some extra light on the subject.

    Ré huge gfx card idling while running LOTRO at max, the ONLY difference between an ancient Radeon 1GB 6670 and the latest cards is peak speed, and that is more likely down the the increased amount of VRAM; the minimum speed blips stay almost exactly the same, so 1GB 6670, 4GB R7265 and 8GB RX470 OC all give the same minimum frame rate +/- 2fps in the not-spots around Bree; while max fps is 70, 120, 160+ respectively.

  22. #22
    Join Date
    Sep 2010
    Posts
    226
    Quote Originally Posted by Yarbro View Post
    so 1GB 6670, 4GB R7265 and 8GB RX470 OC all give the same minimum frame rate +/- 2fps in the not-spots around Bree; while max fps is 70, 120, 160+ respectively.
    They all give same minimum because game is so CPU bound. My example where GFX was cruising idle was from Minas Tirith. When I dropped object draw distance and some other CPU intensive setting, my GFX card came also alive and it started to run around 1200MHz and load around 50%.

    Also it differs alot where you are how game performs. Inside runs much better than outside.

  23. #23
    Join Date
    Jul 2016
    Location
    UK
    Posts
    1,502
    Quote Originally Posted by JiiPee View Post
    They all give same minimum because game is so CPU bound. My example where GFX was cruising idle was from Minas Tirith. When I dropped object draw distance and some other CPU intensive setting, my GFX card came also alive and it started to run around 1200MHz and load around 50%.

    Also it differs alot where you are how game performs. Inside runs much better than outside.
    I am not so sure cpu GHz makes a difference to the minimums, I have an overclocked 4.6GHz hexacore AMD FX chip with 1866Mhz DDR3 RAM, and my daughter has a quad core 3GHz Xeon chip with ancient 800MHz DDR2 RAM, yet they give the same* minimums in the hotspots around Bree, even her previous set-up, a stone age 2.2Ghz Core2 with 533Mhz DDR2 gave the same minimums. In fact, I have run several chips from stock 3.2-3.4Ghz speeds right up to 5.1GHz overclocks, and the minimums have stayed the same, even when the core LOTRO is using only shows 60-80% usage.

    * +4 fps is the most I have seen, and usually no more than +2.

    I agree with the last part though, inside is ALWAYS OK; you can be inside a building in Bree with 160+ fps showing, then go through the doorway and it is like running into cold treacle - from 45 down to 18fps.

  24. #24
    Join Date
    Sep 2010
    Posts
    226
    Quote Originally Posted by Yarbro View Post
    I am not so sure cpu GHz makes a difference to the minimums, I have an overclocked 4.6GHz hexacore AMD FX chip with 1866Mhz DDR3 RAM, and my daughter has a quad core 3GHz Xeon chip with ancient 800MHz DDR2 RAM, yet they give the same* minimums in the hotspots around Bree, even her previous set-up, a stone age 2.2Ghz Core2 with 533Mhz DDR2 gave the same minimums. In fact, I have run several chips from stock 3.2-3.4Ghz speeds right up to 5.1GHz overclocks, and the minimums have stayed the same, even when the core LOTRO is using only shows 60-80% usage.

    * +4 fps is the most I have seen, and usually no more than +2.

    I agree with the last part though, inside is ALWAYS OK; you can be inside a building in Bree with 160+ fps showing, then go through the doorway and it is like running into cold treacle - from 45 down to 18fps.
    Well i know in fact that game is very CPU bound. I had MSI Afterburner pushing overlay data about GPU and CPU usage and I saw CPU usage around 30-40% which means one of my 4 cores was hitting 100% and I confirmed that when I manually assigned affinity for lotroclient 1 core and it got hit 100%.
    So lotro client ain't really using more than 1 core. I do get little better performance when I set affinity to 2 cores, even setting affinity to 3 cores may give little more performance but when I assing 2 cores, thouse cores never hit 100% what reveals that client just use one core. Reason why performance does increase when 2 or more cores are assigned is that windows is swapping tasks to different cores depending of core load, but most of the time it's not doing that good job and assign jobs for core that core where lotro is running and if affinity is locked, windows won't swap lotroclient to some other core.
    Or maybe it's just by design that scheduler likes to cycle tasks to all cores.

    And here is screenshot what I took when I locked lotroclient affinity to cores 3 and 4



    As you can see:
    - Vram is not an issue, only little over 1GB in use
    - GPU is not an issue, 10-30% ping pong usage
    - System memory ain't no issue, I have 16GB and lotro client is using like 2-3GB max

    Only issue is lotro client doesn't support more than 1 CPU cores, so only way to boost client performance is to get CPU what clocks like crazy.

  25. #25
    Join Date
    May 2014
    Location
    Poland
    Posts
    229
    It seems like there was some improvement made in recent months in terms of performance. There is no improvement to sheer framerate and game is still as CPU limited as it was, but stuttering seems to be less frequent and all of those violent stutters are reduced.

 

 
Page 1 of 2 1 2 LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  

This form's session has expired. You need to reload the page.

Reload