What are the best NVIDIA control panel settings for gaming, for a GTX 1060 / Mac Pro 12-core?

Discussion in 'Mac Pro' started by PowerMac G4 MDD, May 9, 2017.

  1. PowerMac G4 MDD macrumors 68000

    PowerMac G4 MDD

    Joined:
    Jul 13, 2014
    Location:
    At the intersection of Conch and Coral.
    #1
    Hi,

    So, I recently purchased a GTX 1060 6GB for my Mac Pro. I use it in Mac OSX, but I also have a Windows 7disk for my favorite Windows-only games.

    I know I am trying to game with Xeons... but my performance should be much better than what I've been seeing. I like this intensive game, BeamNG.Drive (real-time physics simulator), and it just so happens that lower graphics settings make things worse, and - oddly enough - less intensive maps lag more than more intensive ones. At least the lag that I received while I still had my GTX 660 was... well, consistent.


    What are the best NVIDIA settings for a card like this, and do they differ depending upon the CPU? I've followed a few tutorials for those using typical Windows PCs, but those didn't really help.
     
  2. BillyBobBongo macrumors 68020

    BillyBobBongo

    Joined:
    Jun 21, 2007
    Location:
    On The Interweb Thingy!
    #2
    I think that it is currently only you and I that have the GTX 1060 in our systems. I see that you are gaming in Windows since you mention BeamNG.Drive.

    First question, are you running at 1080 or 1440? I have a Dell UltraSharp U3415W (3440 x 1440 pixels) and I too have noticed some worse performance at lower settings i.e. using a 1080 resolution. However this does not happen in all games. Witcher 3 performs better at 1080 than at 1440 however, the FPS in Guild Wars 2 takes a serious nose dive at 1080. It doesn't always seem to be logical. In Assetto Corsa, Project Cars and F1 2016 lowering the resolution increases FPS significantly.

    I allowed the GeForce Experience program to optimise the settings for my games, and then I flicked the resolution to either 3440 x 1440 pixels or 2560 x 1800 pixels. I also downloaded and ran the EVGA Precision X OC program and ran it during testing as it provides an on screen overlay with FPS etc. If you don't have an EVGA card then Steam has a built in FPS counter that you can use.

    It all seems a little hit and miss and I wonder if the driver in Windows isn't quite optimised for it. Worth noting is that BeamNG.Drive is in early access at the moment, so that might also be influencing things.

    In general I'm quite happy with my 1060, although I am considering taking it back to the store and going for a 1070 so that it can cope with my 1440 screen a bit better. That said all games I am playing perform better on the GTX 1060 compared to my old HD 7950, with the exception of Guild Wars 2 that seem to prefer the HD 7950.

    Perhaps you can post some FPS numbers from of some other games and we can compare notes.
     
  3. PowerMac G4 MDD thread starter macrumors 68000

    PowerMac G4 MDD

    Joined:
    Jul 13, 2014
    Location:
    At the intersection of Conch and Coral.
    #3

    Yeah, I have a separate Windows 7 disk for my games. Amongst them, I have BeamNG.Drive, GTA IV, GTA V, Rigs of Rods, and Minecraft. I am running a 1080p Dell as my main display and a 2002 Apple HD Cinema display as my secondary monitor. (The Apple monitor runs at an impressive 1920x1200 pixels.)

    I tried tweaking things within the NVIDIA control panel. I tried the default settings, quality settings, and more. Right now, I think most are still default. Since I have an ASUS card, I downloaded the tweaking software that's meant for it, and it seems to have some overclocking options... but I only tried the "gaming" option, and I don't know how well that really helped. Also, I don't even want to have to overclock my GPU. I just want it to work as it does.

    Speaking on behalf of BeamNG, I noticed something odd about its performance. One of the default maps, East Coast USA (a heavy map with lots of trees and shadows), runs nearly perfectly, with every setting on and at its highest, and in full-screen 1080p. Meanwhile, Grid Map (not the empty one; the one with stunt ramps and such, which is otherwise a very light-weight map and is one of the only ones I'd drive on when I had my 660) is a TOTAL hog. With the same settings, I get horrible stuttering and lag. I have no idea why it's like that. Lowering settings (as I mentioned) does NOT help at all. The map lags in the same way, no matter what. I wonder if I need to re-install my game or something.

    As for GTA V, I was able to put every single setting in the game at its highest, and I could run it at what seemed to me like a consistent 40-60 FPS, in 1080p and in full-screen, which is just fine. I did get a few lag spikes here and there, however.

    Minecraft was the odd one here. The first time I tried it (and this is totally vanilla Minecraft, without any optimizations), I put it on highest settings and achieved as much as, like, 150 FPS. Even more insane is that, on lowest settings possible, it ran at over 400 FPS. However, coming back to the game later on (after messing around in BeamNG), I noticed that, on highest settings, it didn't want to pass about 90 FPS. In fact, it dipped as low as 36 FPS sometimes. Then, on lowest settings possible, it definitely was NOT over 400 FPS. Could be that the server I was on was experiencing some lag, but I doubt it. It runs on a nice internet connection, and I was the only one on at the time. After some time, I think I put the rendering settings on medium-high and tried again - this time, with an FPS cap back on it. Solid 70 FPS. No idea why there was fluctuation there. I'm wondering what this is all about.

    GTA IV is HORRIBLY optimized, so I wasn't surprised at its lag spikes and occasional stuttering... but I really was surprised that it wasn't playing substantially better with my 1060 than it was with the 660. Just like with BeamNG, I was expecting this GPU to just crush the game. (Although, it's understandable, with IV. I had a realistic cars mod that's kinda glitchy, and the game itself was never optimized well to begin with.)

    I am hoping that it's the fault of the drivers and not an issue with the card. I'm also wondering if things would be better on Windows 10; though, Windows 10 + Bootcamp are known to be problematic on Mac Pros. I'll see...




    Oh, and I'll try and compare FPS and such over the weekend. I'm pretty busy with school this week, and I basically don't boot into Windows until the weekend. I was only on it today because I was concerned about my card. I'll make sure to find an FPS meter for each and every game.
     
  4. BillyBobBongo macrumors 68020

    BillyBobBongo

    Joined:
    Jun 21, 2007
    Location:
    On The Interweb Thingy!
    #4
    Cool stuff. I'll keep tweaking settings and we'll compare notes again later. Worth noting is that I actually game in Windows 10. I find it to perform more than adequately, although my only point of reference is gaming in macOS.

    Like yourself I'm not interested in overclocking my card, I purely used the software to monitor FPS and GPU temperatures. I also think it's the drivers at fault.
     
  5. Monyx, May 10, 2017
    Last edited: May 10, 2017

    Monyx macrumors regular

    Joined:
    Oct 24, 2005
    Location:
    Australia
    #5
    I'm looking to go from a gtx 660 to 1060 6Gb as you did, but no gaming - motivation which is hard to quantify is improved Adobe Premiere and Media Encoder performance for editing and encoding of 4K HVEC media. Do you happen to have any insights into Mac OSX apps performance? I've poured over the specs comparing it to my existing gtx660 and just not sure if there'd be a appreciable difference after the changeover - I note that rescaling media say 4K --> 1080p is a fair bit faster as you step up to 1070/80, but otherwise CUDA performance in PC Adobe Pr between 1060/70/80 is almost identical according to Puget gfx systems tests.
     
  6. PowerMac G4 MDD, May 13, 2017
    Last edited: May 13, 2017

    PowerMac G4 MDD thread starter macrumors 68000

    PowerMac G4 MDD

    Joined:
    Jul 13, 2014
    Location:
    At the intersection of Conch and Coral.
    #6

    The card actually does make OSX - the general GUI - feel snappier. Basically, it feels like I'm using an iMac whose GPU is actually meant to be in a Mac and is running proper drivers. Unfortunately, I haven't had much time to play around Photoshop and Illustrator. I have used Photoshop once since receiving my new card. I cannot tell for sure, because I didn't do anything too intensive, but it seemed as though the general application was running better. Navigation throughout all applications I have used just seems snappier and more fluid; although, this is all from an observational standpoint. I have only done one benchmark so far (although, I can report that the scores were much better).

    I actually feel as though the 1070 is a better value; granted, even a 1060 is a large improvement over the 660. I didn't just buy my 1060 for games - I very much wanted my 12-core Mac Pro to really feel like a 12-core Mac Pro, and I realized that my non-flashed 660 was - while it's still a good card - dragging my OSX performance down a bit. So, the 1060 has made it better. I would recommend upgrading from your 660, regardless. Only thing, though, is that the Stacks animations STILL are choppy when transparency is enabled. I have no idea why. I know it's not a flashed card, but it should power through these animations. In Sierra, I set the 'Limit animations' feature, for this very reason. Otherwise, the card is doing fine.

    One thing to know, though, is that GTX 10xx drivers are not very good at the moment. There's actually a bug with the 1060 where it glitches out (just in OSX) upon having the Mac Pro wake from sleep. I have to re-log to fix this. I reported it to NVIDIA, and they were able to replicate this issue; although, I do not know at all when a driver update is going to be released. This is something you should know if you're looking to get one of these cards. (Also, if you get a 1060, skip the 3GB version; go with the 6GB version.)
    --- Post Merged, May 13, 2017 ---
    Billy, I got around to testing out BeamNG today, with the in-game FPS meter enabled. The results are puzzling. My game's FPS isn't necessarily that bad; it's more of a case of the game itself stuttering! Look below:

    1. USA East Coast: full settings > default car
    -Min FPS: 32
    -Max FPS: 47
    -General FPS: 42
    (These were just kind of oddly low, but not stuttering. It gets worse.)

    2. USA East Coast: low settings > Covet
    -Min FPS: 43
    -Max FPS: 66
    -General FPS: 53
    (Game FPS was fluctuating a ton, though. It would shoot up and drop down very quickly. No stuttering yet, but just not consistent - even with Vsync.)

    3. USA East Coast: normal settings > Covet
    -Min FPS: 55
    -Max FPS: 67
    -General FPS: 55-61
    (Okay, THIS was weird. This is when I discovered that the FPS was fine and wasn't fluctuating... yet the entire game was stuttering, as though, perhaps, my CPUs were at fault; or maybe my game installation is at fault.)

    4. Grid map: first full settings, then low settings, and then normal settings > default car all throughout
    -53 min; 59 max; 56 general
    -51 min; 60 max; 59 general
    -50 min; 59 max; 57 general
    (THIS is where I became totally convinced that a glitch was occurring. My FPS was totally fine and not fluctuating badly whatsoever; yet, the entire game was stuttering so incredibly badly - even while my FPS was as high as 60 frames. So, there seems to be something odd going on here that is allowing the game to look like it's getting 20 FPS, when it's actually exhibiting fine frame rates.)

    So, in essence, I gain FPS by lowering the settings, but, consequently, the game exhibits increased stuttering. And then, of course, some maps stutter no matter the settings. (Also - on a side note - I'd expect to be getting better frame rates in general.)
     
  7. PowerMac G4 MDD thread starter macrumors 68000

    PowerMac G4 MDD

    Joined:
    Jul 13, 2014
    Location:
    At the intersection of Conch and Coral.
    #7

    Are there any specific tests that you want me to do? I'd like to help out. I actually work in Illustrator, over the summer; although, that's nearly two months away. For now, I could do tests of your choosing. As far as my own work goes, I'd have to wait until summer to start comparing my 660 to my 1060, in terms of Illustrator CS6 performance.
    --- Post Merged, May 19, 2017 ---

    Hey, have you tested out frame rates yet?
     
  8. BillyBobBongo macrumors 68020

    BillyBobBongo

    Joined:
    Jun 21, 2007
    Location:
    On The Interweb Thingy!
    #8
    I've had to put a new SSD on order, as my windows installation is a bit full at the moment, so that I can install more games to test. Once I've got that installed I'll be back with some numbers.
     
  9. PowerMac G4 MDD thread starter macrumors 68000

    PowerMac G4 MDD

    Joined:
    Jul 13, 2014
    Location:
    At the intersection of Conch and Coral.
    #9
    Cool. Also, I just tried Grid Map at lowest settings... as much as 250 FPS, and still stuttering. Same while capped.

    BTW, do you mind trying a re-installation of the game? I might try it soon, myself.
     
  10. BillyBobBongo, May 21, 2017
    Last edited: May 23, 2017

    BillyBobBongo macrumors 68020

    BillyBobBongo

    Joined:
    Jun 21, 2007
    Location:
    On The Interweb Thingy!
    #10
    Right, @PowerMac G4 MDD, here's what I've got so far.

    System: Mac Pro 5.1 with dual X5680 (3,33 GHz 6-Core), 48 GB of RAM and an EVGA GTX 1060 SC 6GB graphics card.

    Games and monitor were both set to the following resolutions during testing:

    1080p: 1920 x 1080
    1440p: 2560 x 1440
    Ultra-Wide Quad HD (UWQHD): 3440 x 1440

    Tomb Raider

    Ultimate @ 1080p: Min 58.1, Max 60.0, Average 59.9
    Ultimate @ 1440p: Min 44, Max 60.0, Average 57.9
    Ultimate @ UWQHD: Min 30, Max 60.0, Average 50.2

    Rise of the Tomb Raider (DirectX 12 Off)

    Very High @ 1080p: Overall Score: 57.91
    Mountain Peak: Min 35.41, Max 83.44, Average 83.44
    Syria: Min 29.50, Max 88.88, Average 58.66
    Geothermal Valley Min 27.67, Max 71.32, Average 55.31

    Very High @ 1440p: Overall Score: 46.94
    Mountain Peak: Min 29.20, Max 87.07, Average 53.32
    Syria: Min 17.27, Max 77.94, Average 43.54
    Geothermal Valley Min 15.05, Max 62.22, Average 43.45

    Very High @ UWQHD: Overall Score: 37.61
    Mountain Peak: Min 22.17, Max 86,87, Average 42,15
    Syria: Min 18.10, Max 49.93, Average 35.76
    Geothermal Valley Min 19.40, Max 49.96, Average 34.64

    Assetto Corsa

    Ultra @ 1080p: Min 10, Max 131, Average 80
    Ultra @ 1440p: Min 14, Max 105, Average 64
    Ultra @ UWQHD: Min 13, Max 86, Average 56

    Unigine Heaven (Direct3D11)

    1080p: Min 9.2, Max 213.7, Average 105.6, Score 2660
    1440p: Min 11.9, Max 134.8, Average 61.3, Score 1543
    UWQHD: Min 12.3, Max 92.9, Average 44.2, Score 1114

    Unigine Valley (Direct3D11)

    1080p: Min 16, Max 128.1, Average 67, Score 2805
    1440p: Min 10, Max 109.8, Average 58, Score 2428
    UWQHD: Min 19.2, Max 94.4, Average 49.1, Score 2053

    Unigine Superposition (DirectX)

    720p Low: Min 59.03, Max 159.16, Average 100.31, Score 13411
    1080p Medium: Min 50.73, Max 86,06, Average 66.86, Score 8938
    1080p High: Min 38.77, Max 59.24, Average 47.75, Score 6384
    1080p Extreme: Min 14.26, Max 19.97, Average 17.45, Score 2333
     
  11. PowerMac G4 MDD thread starter macrumors 68000

    PowerMac G4 MDD

    Joined:
    Jul 13, 2014
    Location:
    At the intersection of Conch and Coral.
    #11
    Overall, not bad, actually. This card seems to be fine in a Xeon machine. My own card and machine do well with GTA V and other games I play... but BeamNG is the pesky one. I'm curious to see what performance you see with that title.
     
  12. Monyx macrumors regular

    Joined:
    Oct 24, 2005
    Location:
    Australia
    #12
    no need...i will know soon enough just ordered evga 1060 6Gb from Amazon. I'm timing tests on my gtx660 exporting HVEC 4k from Premiere to Media Encoder, timeline renders with lumetri filters, unigine heaven and H265 in Davinci Resolve Studio then will do same in the same system/osx with the new card and post here. I also suggested to Barefeats to add the gtx1060 to their pascal comparison as I it would seem for applications like those above the gtx1060 might be the best price/performance ratio.
     
  13. Beagl3 macrumors newbie

    Joined:
    Jan 2, 2016
    Location:
    Germany
    #13
    Maybe your stuttering is caused by the amount of CPU threads in your system. If a PC has more threads than the game can utilize, it often results in a performance hit. Try assigning only CPU core #0 to a stuttering game in the task manager. The frame rate will of course drop significantly, but perhaps the problem can be at least identified. You mentioned that you hadn't experienced such performance issues with your GTX 660, this could be a result of 1060's Pascal architecture. AFAIK, NVIDIA removed the hardware scheduler from the GPU and replaced it with driver based solution in Maxwell. Eventually this also plays in as a factor.
     
  14. PowerMac G4 MDD thread starter macrumors 68000

    PowerMac G4 MDD

    Joined:
    Jul 13, 2014
    Location:
    At the intersection of Conch and Coral.
    #14

    Yeah, I'll try assigning CPU 0 and see what happens. I think BeamNG.Drive is an especially intensive game and may just show itself more than other games do, in regards to stuttering. GTA V plays just fine, at the highest settings possible; of course, I'm sure I would still get better frames if I had an i7. It's just that the Xeons are simply good enough to still handle GTA V fine, since it's not as intense and is likely optimized way better.

    BeamNG.Drive is more intense, and it stutters even while I receive decent FPS. In fact, I put them game on the lowest settings possible, while on a less GPU-intensive map, and I was still receiving stuttering (albeit on/off) at nearly 250FPS. Capping it at 60 FPS didn't help, BTW.

    Regarding the 660, it actually couldn't run the game as well as this 1060 can. The 1060 does MUCH better; however, with low settings on, the 660 seems to do better than the 1060 - not in terms of FPS, but in terms of lack of stutter. There was noticeable stuttering with the 660, but not as much. (Of course, the 1060 can handle the heavier maps way better, and it can do so at the highest settings.)
     

Share This Page