Not signed in (Sign In)

Discussion Tag Cloud

Categories

Vanilla 1.1.9 is a product of Lussumo. More Information: Documentation, Community Support.

    • CommentAuthorC-A_99
    • CommentTimeJan 12th 2014 edited
     
    I should've moved that discussion over here yesterday, but I was too busy with, well, you guessed it: NL2 itself.

    For those who haven't been following, page 7 of this thread contains the gist of it. I'm getting very low framerates not only on Tango, but on a number of terrain coasters as well.

    Screenshot of scenic mountain coaster hitting 12 FPS.

    Specs: Radeon 6870, Phenom II x4 3.2 GHz, 8 GB DDR3 RAM.

    Settings: I still keep changing them, but I've settled on:
    - 1080p resolution.
    - Minimum LOD settings.
    - Either no shadows or lowest quality. (I didn't get a Radeon 6870 just so I could turn off shadows.)
    - Either low or high background texture and detail. (They seem to have little effect, but the quality is very noticeable a lot of the time.)
    - Either trilinear or anisotropic 8x filtering. (This should have negligible performance impact just as it did back in the day when I had a Radeon 9600 Pro 256 MB card.)
    - No AA. I realized that expecting to be able to use it is just hubris, which is unfortunate because I can handle some AA in other games with more graphical detail than NL2.


    A number of others have had performance issues, but a number of some users claim to be doing just fine on something as old as an 8800 GT. Therefore, I've requested everyone (whether you're having issues or not) to post about their lowest framerates and list which park, what settings, and what specs they're running on. However, with the clutter on the main NL2 thread, that's not going to happen. (There were a ton of "I'm running NL2 just fine" without even mentioning which park they ran and which video card they're using.) I'm hoping I can gather more info in here and get a better idea of what's going on.

    As mentioned in the main NL2 thread, the problem seems to be graphics related. The CPU appears to be utilized correctly, and I've found that Tango is not slowing down at the CPU level at all. (It's using about 50% of my Phenom II; I'd expect more if the CPU were any issue.) The problem appears graphically related, as removing all lights in the Tango scene (except for the lights on the part of the ride that animates) speeds things up considerably (to around 20 or 25 FPS [depending on resolution] on minimum settings [no shadows]). Tango has about 40 or so lights, which any decently modern 3D engine should have little or no trouble with.

    The biggest kicker here however, is that I'm not really noticing any performance difference between a Radeon 6870 and Geforce 540m, which is honestly just plain ridiculous. For those who don't know, the performance difference between a 6870 and a 540m is about the same (if not larger) as the difference between an Xbox 360 and PS4 for most games. One generation of hardware difference, no performance gain in NL2 to show for it. The 540m actually runs some NL1 tracks better, but I overlooked this and simply figured that the old OpenGL implementation happened to work better with Nvidia, but for NL2 with all these shaders, I cannot fathom why the 6870's capabilities aren't even remotely being utilized. I'll do more testing on the 540m to see how it handles 1080p instead of 720p (and yes, I tested the 6870 at 1080p), but this is pretty bad so far. I'll also have to test more to eliminate the possibility of it being a CPU issue, as the Phenom II x4 3.2 GHz and i5 2430m 2.2 GHz have similar performance according to Windows. (I'll look up some actual benchmarks later, but you basically have a case of better CPU running at lower speed (i5), vs. poorer CPU running at higher speed (Phenom II).)

    Anyway, I'd like to get an idea of how others are doing on NL2, especially with those of you who have multiple computers to test it on.

    Perhaps it's time to sell off some of the old stuff I have and start a new build; one that'll allow me to run BF4 with max settings and 1080p Eyefinity, and one that'll allow me to... run NL2 lol. Or perhaps I just need to wait for a patch, but I have no clue if the team intends to consider an update in this direction, especially if few others are having the same problems. Perhaps NL is just not playing as nicely with AMD hardware, but something as powerful as a 780 should not be hitting 30 FPS on a single terrain woodie either.

    A side issue I'm having is that the editor is capped to some 15 or 20 FPS; very annoying, but this doesn't seem to be as relevant to the graphics engine.

    On a positive note, NL2 seems to have a pretty low memory footprint (I think it's using around 1 GB?), and runs in x64.
    • CommentAuthorgronji2004
    • CommentTimeJan 12th 2014
     
    Ive noticed changing all the settings in NL2 to the lowest, then if using an nVidia card, using the control panel and setting the display settings to the ones you require in the nvidia control panel, as opposed to in game seems to make it run better.
    •  
      CommentAuthorL Bosch
    • CommentTimeJan 12th 2014 edited
     
    System specs: i7-2670QM @ 2.2GHz, 64bit, 16GB RAM, AMD Radeon HD 6650M 2GB
    NL2 settings: AA Quality 2, LOD near, Anisotropic 8, Shadows 4, Background max, Advanced Effects On.

    In the Tango Park, looking at both Tango and Olympia Looping, I get 6 FPS, 16% CPU load (4 out of 8 kernels in use by NL2), 90% CPU load (with all settings to minimum, I get 7 FPS)
    In The Forest Mountain Park, I get 10 FPS, otherwise same load percentages.
    • CommentAuthorC-A_99
    • CommentTimeJan 12th 2014 edited
     
    ^ Thanks. That sounds about in line with what I got when testing on an older box with a Radeon 5770.

    Posted By: gronji2004Ive noticed changing all the settings in NL2 to the lowest, then if using an nVidia card, using the control panel and setting the display settings to the ones you require in the nvidia control panel, as opposed to in game seems to make it run better.


    The only settings I can try this with are AA and AF. I already keep AA off, and AF should not make any noticeable performance impact, as setting it in NL2 itself doesn't really do so. I doubt forcing it on via AMD's control panel would make much difference. Guess I could test it when I have time, but it sounds like a waste of such.
    • CommentAuthorgronji2004
    • CommentTimeJan 12th 2014
     
    Posted By: C-A_99^ Thanks. That sounds about in line with what I got when testing on an older box with a Radeon 5770.

    Posted By: gronji2004Ive noticed changing all the settings in NL2 to the lowest, then if using an nVidia card, using the control panel and setting the display settings to the ones you require in the nvidia control panel, as opposed to in game seems to make it run better.


    The only settings I can try this with are AA and AF. I already keep AA off, and AF should not make any noticeable performance impact, as setting it in NL2 itself doesn't really do so. I doubt forcing it on via AMD's control panel would make much difference. Guess I could test it when I have time, but it sounds like a waste of such.


    Hmm strange. Im using GeForce GTX 560 TI and all the settings are available to change. Depends what drivers etc i guess ...
    • CommentAuthorC-A_99
    • CommentTimeJan 12th 2014
     
    Generally, you can't change quality settings that are internal to the program (like LOD, shadows, etc.), but looks like Nvidia has built something built-in now where it does. (Geforce Experience, I think it's called.)
  1.  
    I'm considering going back to the GTS450 and returning the GTX760 if this is really the case that it is not utilizing much GPU power available. I have not swapped back to try yet but I think I may have jumped the gun on getting the new one. The devs have been kind of quiet about acknowledging the performance issues and whether they're being worked on, but obviously something isn't right when old low end laptop graphics cards are getting the same performance as brand new $300 ones.

    I'd also like to hear people's ideas on what is an efficient workflow with such slow load times. Not sure anyone is there yet but checking your detail on support work (such as rotating beams) is going to be pretty painful if that isn't improved on. Like in the other thread I would have been in and out of the simulator 5-6 times checking and changing work in the time it takes to render NL2. Perhaps I will create another thread for this to not clog this one up.
    •  
      CommentAuthorride_op
    • CommentTimeJan 12th 2014 edited
     
    We are all for a discussion about how to optimize your park for improved performance. However, it is not helping to compare NoLimits 2 to Battlefield 4 or any other 3D game. Battlefield 4 is not simulating roller coasters with amazing detail and NoLimits 2 is not a 3D shooter with carefully controlled scenes and scripted events. Both applications are doing completely different things.

    The number one influencer on how fast a park runs is YOU, the park creator. This holds true since NoLimits 1. Build a medium sized coaster in flat land and it will run great. Build a few big rides on complex terrain with a dense forest of 3D trees and flat rides with over 6000 flashing lights and your computer will struggle. Tango park shows what you can do with scripting and that's it. It is not a park optimized to run fast. On Forest Mountain Park I went overboard with the trees, my fault. It runs well enough but I never expected it to run at 120 fps with all detail settings maxed. It runs well enough to enjoy the rides and that's why we included it.

    We have a ton of optimization built in to help preserve as much detail as we can. We will continue to improve and make optimizations to the engine as we go along. That's our promise to you.

    What we have no control over is what you make and how you make it. It is up to you to employ the various optimization techniques and methods included in NoLimits 2 to help the simulation run well. If you tangle together 4 coasters with over 6000 feet of track into a giant ball of yarn with each coaster running 5 trains don't cry to us about how Battlefield 4 runs great and NoLimits 2 is a slide show of 0.5 fps. That's on you!

    So please feel free to share system specs and compare the performance you get. Feel free to make BETTER optimized tracks and teach each other how to go about that. Then you are helping the community learn to build better parks and that is what we are all here for!

    -Ride_Op
  2.  
    ASUS U46E
    Windows 7 64-bit
    8GB RAM
    Intel i7-2640M @ 2.80 GHz (quad-core, 8-way threading)
    Intel HD 3000 Graphics (64 MB dedicated VRAM, unknown max shared)
    1366x768 resolution

    All Windows visual effects turned off.
    All NoLimits visual settings turned off or set to minimal.
    (Trilinear textures, no shadows, lowest render distances.)
    (FPS Limit: 45 FPS)

    -

    Tango/Olympia Looping

    Looking at Tango: 3 FPS
    Looking at Olympia: 11 FPS

    -

    Forest Mountain Park

    Overview: 23 FPS


    Turnaround: 37 FPS


    Suspended: 34 FPS
  3.  
    Fair enough, but that says nothing about my second point which is load times. And this has nothing to do with our content, as my boulder dash recreation which is currently 900' of track with simple autosupports is taking 5+ minutes to freeze to check anything. My CPU is admittedly a few years old, but it was AMD's top of the line at the time so it shouldn't have trouble rendering supports for 900' of track (especially since the wood supports aren't that much more detailed than they were in NL1 as far as shapes/polys). My 15k support ride loaded into NL1 in about 15 seconds.

    It also says nothing about the integrated graphics card that AJ just posted getting pretty similar performance as me.

    Phenom X4 955 Black @ 3.7
    8gb DDR3 PC1600
    GTX760
  4.  

    The devs have been kind of quiet about acknowledging the performance issues and whether they're being worked on, but obviously something isn't right when old low end laptop graphics cards are getting the same performance as brand new $300 ones.

    They're most likely watching the forums and collecting data / samples. The last thing they want to do is issue a wrong statement if they were going to make one.

    •  
      CommentAuthorride_op
    • CommentTimeJan 12th 2014
     
    Re-posting to ensure everyone reads this beneath the large screenshots that were just posted:

    We are all for a discussion about how to optimize your park for improved performance. However, it is not helping to compare NoLimits 2 to Battlefield 4 or any other 3D game. Battlefield 4 is not simulating roller coasters with amazing detail and NoLimits 2 is not a 3D shooter with carefully controlled scenes and scripted events. Both applications are doing completely different things.

    The number one influencer on how fast a park runs is YOU, the park creator. This holds true since NoLimits 1. Build a medium sized coaster in flat land and it will run great. Build a few big rides on complex terrain with a dense forest of 3D trees and flat rides with over 6000 flashing lights and your computer will struggle. Tango park shows what you can do with scripting and that's it. It is not a park optimized to run fast. On Forest Mountain Park I went overboard with the trees, my fault. It runs well enough but I never expected it to run at 120 fps with all detail settings maxed. It runs well enough to enjoy the rides and that's why we included it.

    We have a ton of optimization built in to help preserve as much detail as we can. We will continue to improve and make optimizations to the engine as we go along. That's our promise to you.

    What we have no control over is what you make and how you make it. It is up to you to employ the various optimization techniques and methods included in NoLimits 2 to help the simulation run well. If you tangle together 4 coasters with over 6000 feet of track into a giant ball of yarn with each coaster running 5 trains don't cry to us about how Battlefield 4 runs great and NoLimits 2 is a slide show of 0.5 fps. That's on you!

    So please feel free to share system specs and compare the performance you get. Feel free to make BETTER optimized tracks and teach each other how to go about that. Then you are helping the community learn to build better parks and that is what we are all here for!

    -Ride_Op
    •  
      CommentAuthorride_op
    • CommentTimeJan 12th 2014
     
    Regarding load times, we're aware of it. We've been using NL2 for a long time now and are used to it. We're aware that it takes awhile due to many things that are being processed. Among other things, on wooden coasters the engine is analyzing all of the individual wooden beam segments and is combining/optimizing them into longer posts when possible. This just takes awhile. We're always working on improving things, but again we're pushing around more polys than other games. We have to work with anything you can put together and then optimize it for you. The wooden structure is auto generated with many supports configured with higher and lower LOD priorities. It's a numbers game here. The more you throw at it the longer it'll take to freeze and process.

    -Ride_Op
    • CommentAuthorzaz76
    • CommentTimeJan 12th 2014
     
    Posted By: ride_opRegarding load times, we're aware of it. We've been using NL2 for a long time now and are used to it. We're aware that it takes awhile due to many things that are being processed. Among other things, on wooden coasters the engine is analyzing all of the individual wooden beam segments and is combining/optimizing them into longer posts when possible. This just takes awhile. We're always working on improving things, but again we're pushing around more polys than other games. We have to work with anything you can put together and then optimize it for you. The wooden structure is auto generated with many supports configured with higher and lower LOD priorities. It's a numbers game here. The more you throw at it the longer it'll take to freeze and process.

    -Ride_Op


    Would it be possible to cache some of those calculations? For example, say I'm done editing my park and ready to upload to the exchange, could I make a "finalized" version of my park with all applicable optimizations pre-computed? I understand that this would likely make editing it in the future difficult (i.e., if you optimize away many of the auto-generated wooden supports), but you could keep two versions of the park (e.g., an "editable" version and a "finalized / ride-only" version).
    • CommentAuthorC-A_99
    • CommentTimeJan 12th 2014 edited
     
    I ran some side by side tests with my desktop and laptop. Same monitor, running side by side and observing the framerate. Again, the specs:

    Desktop: Phenom II X4 955, 8 GB RAM, Radeon 6870
    Laptop: i5 2430m, 6 GB RAM, Geforce 540m

    Both systems run Windows 7 x64 Professional, both systems have similar background tasks running. The task list is almost identical. Both NL2 setups are at identical video settings, and the day/night cycle is also the same. According to PassMark, the Phenom II x4 955 just barely outperforms the i5 2430m, although the i5 is more efficient and is doing more per clock cycle. I also already mentioned the performance difference between the two GPU's, which you can look up to verify (current PassMark scores put the 540m at 700, and the 6870 at 2571), but the gist of it is that it's essentially one console generation's worth of performance difference. This performance difference is consistent in every single 3D application I've run except for...

    NL as well as NL2 apparently. I've only casually observed this in NL1 where I noticed in some tracks that NL1 was running better on my laptop than desktop. The side-by-side comparison I just made here shows this to be the case for NL2 as well. I tested Scenic Run and Forest Mountain Park.

    The Geforce 540m is getting up to 10 FPS (big difference at 20 and 30 FPS) more than the Radeon 6870. This is the only application doing this, although granted, I have not tried observing the performance differences between these two systems in other OpenGL applications. A quick Google search shows that OpenGL performance in general doesn't quite favor either AMD nor Nvidia cards, but it's clear that NL2 is not liking a good number of the AMD cards.

    Which, forgive me for calling that ridiculous, but as someone who has used a decent number of 3D applications on both systems, but does not know the intricacies of the OpenGL and Direct3D API's, this looks ridiculous on the surface that such a shader-heavy application would be running better on the worse video card of the two. I guess I'll try to test out more OpenGL applications such as Blender or Doom, and see if I need to start sending emails to AMD telling them to fix up their OpenGL performance. In the meantime, it would be nice to hear some additional details as to why this could be the case.

    That said, I do realize this performance disparity is low on the list in comparison to Mac support and addressing the crashes and bugs that other users have been having. In comparison, NL2 has been remarkably stable on every system I've tested so far, although I've not gone in-depth into the editor on any system other than my desktop so far.
    •  
      CommentAuthorride_op
    • CommentTimeJan 12th 2014
     
    We're already doing that once a coaster is frozen. While the coaster freezes those optimizations are calculated and stored in the file. There's still more going on when a park is loaded into memory to run in play mode that makes things slower. This is where people with SSD drives will see a big boost. The faster that data can go from your hard drive and into memory the faster the park will load.

    -Ride_Op
    •  
      CommentAuthordjd711
    • CommentTimeJan 12th 2014
     
    My laptop uses a caching SSD, the first time a park is loaded it takes a while but then any times after that the SSD uses it's data instead of the standard hard drive.
    •  
      CommentAuthorride_op
    • CommentTimeJan 12th 2014
     
    One other note...the first time you load of a train style you have not viewed before there is additional processing going on. This is a one time thing and afterwards you won't have that delay when you view the same style.

    -Ride_Op
  5.  
    It won't be a direct comparison, as I'm just trying out the demo for now (I'm waiting until the first or second update to buy), but maybe this will show another angle.

    Here's my oldish laptop, totally factory, which I've been playing the demo on.
    Windows 7 Home Premium 64-bit
    ATI Radeon HD 4200 approx. 1.8 GB total memory (I don't know how much is shared etc.)
    AMD Phenom II N830 triple core @ 2.10 GHz
    DirectX 11
    4 GB RAM

    NL2 settings:
    disabled antialiasing
    turned font-smoothing to simple
    trilinear textures
    turned off shadows
    background texture and detail to low
    limited FPS to 60 (though I should lower it)
    ...using a window.

    On Wilderness Park, my FPS was from 13-45, lowest near the water and overviews, highest on the lift and when there wasn't much water or track in view. Forest Hills Park was 11-26 with the same characteristics, and Hybris 12-37. In Hybris, FPS fluctuated quite jarringly...totally different from the other two.

    I'm admittedly a noob in these matters. I see all these scary numbers and names and am lost.
    • CommentAuthorPaulm12
    • CommentTimeJan 13th 2014
     
    Alright, so I'm on a Desktop: 12 GB RAM, Intel i7 8-core @ 2.67, Radeon HD 4560 w/ 1 GB
    I'm also running two monitors, with different resolutions, so I assume that could be taking up a lot of GPU.
    I run the game with everything on lowest settings, except draw distance which is at "low" ( I prefer performance and load times over graphics). Most of the time the framerare does not drop below 20-30 fps. The tango park obviously slows down quite a bit, maxing out at around 23-30 or so and getting as low as 7-10 at certain spots, but that's to be expected.
    To put things in perspective, I could run NL1 on max settings (with some antialiasing) and it would rarely, if ever, dip below 100 fps, even on parks with tons of models. Also, loading never took longer than a second or two.
    NL2's performance is quite good to me (it doesn't blow me away by how optimized it is, but then again it's a simulator optimized for creation and not necessarily running pre-made, optimized parks) because the light, reflection, and water all look great.
    But I am more concerned with load times than performance right now, so I was wondering if some of the graphics settings also increased or decreased load times.
    Hey Ride-Op, if you get the chance, how much will an SSD speed things up in terms of loading parks? Like if I was to get one and put the parks or the application there, would it, say, double the speed in which the park loads?
    • CommentAuthorTOGO Fan
    • CommentTimeJan 13th 2014 edited
     
    Here is my computer specs.

    <a href="http://imgur.com/e12iNXj"><img src="http://i.imgur.com/e12iNXj.jpg" title="Hosted by imgur.com"/></a>

    Should I get a more powerful video driver?
    More ram?
    Better processor?
    Better CPU?

    I have a Nvidia GeForce GTX 550 Ti version 311.06

    Should I upgrade that or get a better unit?
    • CommentAuthorC-A_99
    • CommentTimeJan 13th 2014
     
    Posted By: Paulm12
    Hey Ride-Op, if you get the chance, how much will an SSD speed things up in terms of loading parks? Like if I was to get one and put the parks or the application there, would it, say, double the speed in which the park loads?


    Based on his previous response...

    Posted By: ride_opWe're already doing that once a coaster is frozen. While the coaster freezes those optimizations are calculated and stored in the file. There's still more going on when a park is loaded into memory to run in play mode that makes things slower. This is where people with SSD drives will see a big boost. The faster that data can go from your hard drive and into memory the faster the park will load.

    -Ride_Op


    ...it seems very difficult to even get an estimate.

    @TOGO Fan: Any word on how well NL2 is running?

    _____________________________

    Anyway, thanks to everyone who responded. I did get one report on another forum about a 7870 (and i7-975) with "adjustments" running Tango well, which suggests it may not be a general AMD issue, but something more isolated. Either way, it's clear that more research is needed. It's also possible that CPU is playing a larger role than it appears to be, but I'm having trouble finding anything solid to strongly back up either side, other than the Task Manager's CPU usage readings from a single system test.
  6.  
    Here's an annoying one I don't think anyone's mentioned. Why do things stop loading when you minimize the program? The load times would be far less annoying if it could do it in the background. Is this a windows thing where the process needs priority, or just the way it works?
    • CommentAuthorC-A_99
    • CommentTimeJan 13th 2014 edited
     
    ^ I usually keep NL2 in a window if I know I'm going to be doing other things while it loads, and fullscreen doesn't work nicely with alt-tab. However, I always just leave the window open without minimizing, and whatever else I'm doing just gets overlaid on top of that window. That may be the best workaround available for you.

    Edit: The editor is able to load stuff for me while minimized.
    • CommentAuthorTOGO Fan
    • CommentTimeJan 13th 2014
     
    I get 17 FRPS when looking at the Olympic looping coaster and 17 or 18 FRPS when looking at the Tango.
    I have advanced effects off
    Shadows off
    LOD distance very near
    Background texture and detail to lowest
    Texture filter quality to bilinear
    Limit framerate to 90
    Full scene aliasing to off
    No triple buffering
    No V sync.
  7.  
    Posted By: TOGO FanI get 17 FRPS when looking at the Olympic looping coaster and 17 or 18 FRPS when looking at the Tango.

    Holy hell where can I get those kind of specs lol. Those are the best I've heard so far!
    • CommentAuthorTOGO Fan
    • CommentTimeJan 13th 2014
     
    So can I upgrade my computer to get 40 or 50 FRPS with the tango and Olympia looping?
    • CommentAuthornannerdw
    • CommentTimeJan 13th 2014 edited
     
    Full screen 1080p and everything at lowest settings/off:
    23 fps looking at Olympia Looping
    21 fps looking at Tango
    24 fps aerial view of both rides in Forest Mountain Park
    35 fps aerial view of Scenic Run

    CPU: Intel Core i7-2630QM 2.0-2.9 GHz
    Ram: 8 GB
    GPU: Nvidia Geforce 560M

    I typically run in windowed mode, and make the window smaller when I'm in simulator mode. There really needs to be a way to make the editor resolution independent from the simulator. I don't care what framerate I'm getting in the editor; I just need all the screen real-estate I can get. When I'm in the simulator, I wouldn't mind turning the resolution down to 720p.
    •  
      CommentAuthorHZGarfield
    • CommentTimeJan 13th 2014
     
    Track Tube in NL1 can seen as pentagonal prism´╝î How many polys does NL2 's Track Tube have? We may have an option on this
  8.  
    Dell XPS15 laptop from 2011 running Windows 7
    CPU: Intel Core i5-2430M @ 2.40 GHz
    RAM: 6 GB
    GPU: Nvidia GeForce GT540M

    Running it as a full size window on the 1366 x 768 screen, Advanced Effects off, limited to 30 fps, antialiasing off, all other settings at default it runs pretty smoothly with a little lag here and there and around 10-12 fps when looking straight at the Tango with bits of Olympia looping in the background.
  9.  
    So I just decided to check out the Tango Park to gauge my performance. Here are my specs:

    AMD FX-4300 OC@4.5GHz
    8 GB DDR3 1600MHz RAM
    NVidia GTX 560 1GB

    Running at 2560x1440 resolution with all settings maxed out, including antialiasing, I get about 10fps average when looking at the entire scene, with both Tango and Olympia Looping in view. FPS will randomly drop down to the 2-3 range, but it normally hovers between 10 and 15 whenever Tango was in view. When I was looking at just Olympia Looping I averaged about 25, with bits as high as 40 and 50 fps depending on how many of the lights were in view.
    • CommentAuthortplowman91
    • CommentTimeJan 13th 2014
     
    I am telling you- AMD just sucks. They love to have performance and driver issues let alone use a lot of power. NVIDIA; I have never had an issue with drivers, performance, or high power. They use proper power levels and have always noticed less complaints with gaming performance. If you have AMD and you're complaining about low FPS, switch to nVidia... you won't be disappointed.
    •  
      CommentAuthorrcrider4
    • CommentTimeJan 13th 2014
     
    I would if I could. Unfortunately, I have a laptop, so that is kind of out of the question without buying a new one. I still get pretty good performance on everything, but not quite what I expect from a 2 year old, $1000 high performance laptop.
    • CommentAuthorC-A_99
    • CommentTimeJan 13th 2014
     
    Posted By: tplowman91I am telling you- AMD just sucks. They love to have performance and driver issues let alone use a lot of power. NVIDIA; I have never had an issue with drivers, performance, or high power. They use proper power levels and have always noticed less complaints with gaming performance. If you have AMD and you're complaining about low FPS, switch to nVidia... you won't be disappointed.


    540m/Intel integrated graphics on my laptop can't output HDMI properly. (The color range gets compressed and any attempt to fix results in white pixels flickering.) Never seen this on any AMD card. A quick Google search not only shows OpenGL issues getting fixed, the use of AMD chips in the PS4 and the GL extensions they're working on also indicate that they've fixed things up for OpenGL.

    As far as comparing them goes, you'll find both sides have better deals and performance on some cards than others, as well as applications performing better or worse no matter what. Unless you have some references to link to, it just sounds like you're trolling.
    • CommentAuthorsprog
    • CommentTimeJan 13th 2014
     
    Just thought I'd post up mine to add to the data you've got.

    Current setup:
    Intel i5-2500K (Quad Core) 3.3GHz
    8Gb DDR3 1600MHz RAM
    AMD Radeon HD 7700 Series (2Gb GDDR5 VRAM)

    Set my NL2 settings to the same as yours in the first post (No AA, Tesxture anisotropic 8x filtering, minimum LOD + BG Detail + Textures + No visual effects, 1080p full screen), these are my results from Tango:

    Looking at Tango averages 34fps
    Looking at Olympia Looping averages 21fps
    Overhead shot averages 14fps
    Average CPU Usage = 25% across all 4 cores
    Memory usages sits at about 3.5Gb

    And results from Forest Mountain Park (tried to aim for similar views to Arly, for comparison):

    Overview averages 28fps
    Turnaround averages 60fps
    Suspended averages 90fps
    Average CPU usage is again the same
    Memory usages sits at about 3.5Gb




    Also, just for a laugh I maxed out all settings, results are:

    Tango average 11fps
    Olympia Looping average 11fps
    Overview averages 11fps

    Overview of Forest Moutain averages 21fps
    Turnaround averages 33fps
    Suspended averages 35fps

    Again, all 4 CPUs hover around 25% and Memory is around 3.5Gb

    Hope that all helps!
    • CommentAuthor1NT4M1N
    • CommentTimeJan 13th 2014
     
    Why can't NL2 leverage far more GPU and CPU power?
    • CommentAuthorTOGO Fan
    • CommentTimeJan 13th 2014
     
    My Nvidia GeForce GTX 550 Ti version 311.06 is now getting a huge update. I manually checked it for updates and it seems to have found a huge one.
    • CommentAuthor1NT4M1N
    • CommentTimeJan 13th 2014
     
    Posted By: TOGO FanMy Nvidia GeForce GTX 550 Ti version 311.06 is now getting a huge update. I manually checked it for updates and it seems to have found a huge one.


    Updates, where?
    • CommentAuthorTOGO Fan
    • CommentTimeJan 13th 2014
     
    Go to device manager, select display adapters, select your device, and right click and select update driver software.
    • CommentAuthormacevhicz
    • CommentTimeJan 13th 2014
     
    Here's a question which I know has been asked before, but I haven't seen an answer to, and cannot see any info regarding this on the nolimitscoaster.com page.

    Is NL2 a 64 bit program? Ie, would it be beneficial to have 64 bit Windows?
    • CommentAuthorole
    • CommentTimeJan 13th 2014
     
    On ATI/AMD Graphics, NL2 is doing things slightly different in Window mode atm.

    NL2 was developed on Nvidia Hardware (550) and for whole time of 8 years I never experienced a single bug in their drivers. Last year I bought a AMD card only to test NL2 on that hardware and it took me around 5 minutes to detect some bugs in some core DX9 functionality. As a workaround I had to synchronize each frame with the CPU for each frame on ATI in Window mode. In Fullscreenmode, the driver bug did not appear to happen, so there is a lot relaxed synchronization resulting in higher frame rates. It is up to anyone if he prefers ATI over Nvidia or other way around, but if there are serious bugs in drivers there is not this much I can do.
    • CommentAuthorC-A_99
    • CommentTimeJan 13th 2014
     
    Thanks for the update and info. There must be other bugs going on and/or things we're missing, since most of the performance discrepancies I've noticed between the 540m and the 6870 happened in fullscreen.
    • CommentAuthorole
    • CommentTimeJan 13th 2014
     
    Okay, let me first deal with the critical bugs. I am collecting information right now to fix serious incompatibility issues and crases. After the critical bugs have been fixed, I will try to look again into ATI performance. I have a 5xxx series card, not sure which one, but it was a cheap low end one and performance was okay compared to its class of card.
    • CommentAuthor1NT4M1N
    • CommentTimeJan 13th 2014 edited
     
    My worst fear is that we will never see AMD GPU performance match their Nvidia counterpart. This is going to cause issues in the community, since anyone with an AMD GPU might as well not bother running NL2, as it performs terribly at present.
    • CommentAuthorC-A_99
    • CommentTimeJan 13th 2014
     
    When looking things up about AMD issues, I found some post from 2010 or 2011 stating that AMD had mostly fixed their problems by the 4000 series. Best guess I have for this is that these issues were fixed in DirectX 10 and 11, while some issues were left in DX9. I suppose it's worth asking whether these issues are present in DX 10 or 11, as DX 9 is quite old.

    Anyway, I appreciate your time to comment in depth about these issues; this was mainly what I was looking for initially. Thanks. In the meantime, I'll continue asking around and experimenting (I still have some friends to ask to test the demo), and I heard of some good results that came from overclocking. Not my kind of territory, but it's winter and I have a fat and powerful house fan I can place next to my rig. =)

    Also, if there's any additional information/data anyone could gather that would be helpful, do let us know.

    Regarding the critical bugs, it seems like some users are getting them with Intel graphics, but I've been unable to reproduce any on the Intel i5 2430m in my laptop. I'll jump back in and test around again.
    • CommentAuthormacevhicz
    • CommentTimeJan 13th 2014 edited
     
    Well, I've gotten the demo of NL2 running on my Macbook via Parallels. It was clunky, and one change fixed that, which was TURNING OFF WINDOWS AERO!

    Looking at the Dive Machine station with Aero on (default), was very clunky (fps about 4). Turning off Aero, and now I see no lag at all.

    Sooo glad to have turned of AERO, now NL2 is running like a dream!
  10.  
    Posted By: macevhiczSooo glad to have turned of AERO, now NL2 is running like a dream!

    Can confirm. Every time I post screenshots of something people always ask why I'm "running Windows 98", but the performance is noteworthy.

    If you go to Performance Information & Tools in the Control Panel, you can disable tons of other visual effects as well that also help boost performance, but Windows Aero is the big one.
    •  
      CommentAuthorUncleArly
    • CommentTimeJan 13th 2014
     
    It didn't seem to make a difference for me unfortunately.
    • CommentAuthorC-A_99
    • CommentTimeJan 13th 2014 edited
     
    I forgot to mention, I always have Aero off by default due to the clunky mess it is. I don't need constant flashing back and forth to the ugly basic blue whenever alt-tabbing something. Anyway, I replaced the ugly basic blue with this scheme, and it looks a whole lot better than blue as well as that awful Windows 98 legacy mode that I'm pretty sure MS only keeps so as to not upset the luddites. But hey, if you prefer something else, I'm not judging, I just think Windows Basic Blue and the 98 style look awful. (Now, Windows XP Silver was quite nice.)

    Aero's the only one with potential to make a difference in performance on modern systems. Disabling something as basic as moving windows that follow the cursor won't help unless you're on some 300 MHz Pentium (in which case, you probably wouldn't even be able to browse these forums).
    • CommentAuthor1NT4M1N
    • CommentTimeJan 13th 2014
     
    Posted By: C-A_99When looking things up about AMD issues, I found some post from 2010 or 2011 stating that AMD had mostly fixed their problems by the 4000 series. Best guess I have for this is that these issues were fixed in DirectX 10 and 11, while some issues were left in DX9. I suppose it's worth asking whether these issues are present in DX 10 or 11, as DX 9 is quite old.

    Anyway, I appreciate your time to comment in depth about these issues; this was mainly what I was looking for initially. Thanks. In the meantime, I'll continue asking around and experimenting (I still have some friends to ask to test the demo), and I heard of some good results that came from overclocking. Not my kind of territory, but it's winter and I have a fat and powerful house fan I can place next to my rig. =)

    Also, if there's any additional information/data anyone could gather that would be helpful, do let us know.

    Regarding the critical bugs, it seems like some users are getting them with Intel graphics, but I've been unable to reproduce any on the Intel i5 2430m in my laptop. I'll jump back in and test around again.


    I also have two different Intel HD graphic chipsets on two laptops and don't have an issue. I have an issue with the AMD GPU's.