Welcome

Please register for Total War Access to use the forums. If you're an existing user, your forum details will be merged with Total War Access if you register with the same email or username. For more information please read our FAQ’s here.

Categories

Performance on a 2080Ti

lukemurtagh1995lukemurtagh1995 Junior MemberPosts: 27Registered Users
Hello everyone,

I was thinking about buying an MSI 2080 Ti Gaming X Trio GPU however I noticed some issues people had with Warhammer 2 and the new Turing cards a few months ago regarding a memory leak. Can anyone confirm if these issues have been solved? Since this is a game I play a lot I may not get the card if it still has issues with this game. I would also plan on playing Three Kingdoms with it as well.

Thanks!
Tagged:

Comments

  • psychoakpsychoak Posts: 2,359Registered Users
    I saw mentions that the drivers had been corrected.

    This isn't likely to be a relevant issue long term even if it still has issues. My 1080 Ti only runs at half load for full ultra settings at 2560x1440. It is now running with an i9 9980XE, and the CPU still bottlenecks first. The 2080 Ti is major overkill for TW.
  • lukemurtagh1995lukemurtagh1995 Junior Member Posts: 27Registered Users
    psychoak said:

    I saw mentions that the drivers had been corrected.

    This isn't likely to be a relevant issue long term even if it still has issues. My 1080 Ti only runs at half load for full ultra settings at 2560x1440. It is now running with an i9 9980XE, and the CPU still bottlenecks first. The 2080 Ti is major overkill for TW.

    Wow really? My current 1080 can only handle most settings on Ultra at 1080p and between 60-100 FPS . I know the 1080 Ti is more powerful than a 1080 but I didn’t expect there to be such a difference.
  • psychoakpsychoak Posts: 2,359Registered Users
    It's probably not the 1080.

    Check your core usage while the game is running. Odds are you'll have individual cores pegging 100%.

    DX11 is ****, and Warscape is only relatively well threaded. When I run a 4v4 custom battle, it taps out around 29% CPU utilization on my new system, which is 50% load for those of us who understand the way hyper-threading works. It doesn't spread this evenly over all 18 cores though, it's not even using several of them.

    If I load up Ashes of the Singularity, I can run that sucker maxed out, on the largest map, and never peg a core even with tens of thousands of units running around. During the benchmark, 17 out of 18 cores are close to evenly utilized, at around 30% load on one logical core each, with just a single core lightly utilized. It achieves 96% plus GPU utilization while it works this distribution magic.

    Switch it to DX11 mode, and it does the same thing, but the two cores handling the feed are maxed out near 100% usage. DX11 has more threading than previous versions, but it still has one active state, and the active state isn't threaded, so you create two states, and switch them back and forth to load into the GPU. If I had a 2080 Ti, or I was running SLI, DX11 would solidly bottleneck my ass at about the same performance level I can achieve with just the 1080 Ti.

    Nitrous is one of the first, and one of the very few, possibly the only engine, to achieve such stellar levels of parallelism. It's designed from the ground up for multi core systems, Warscape isn't.
  • lukemurtagh1995lukemurtagh1995 Junior Member Posts: 27Registered Users
    So DX11 is the issue? Whenever I try to run DX12 I lose quite a few frames per second. I’ve noticed my GPU is running at 100% during the battle and campaign benchmarks as well. The only time my GPU isn’t running at 100% is during the Skaven benchmark, where halfway through it drops to around 60%.

    My CPU by the way is an i7-9700K overclocked to 5.0GHz.
  • lukemurtagh1995lukemurtagh1995 Junior Member Posts: 27Registered Users
    edited May 17
    psychoak said:

    It's probably not the 1080.

    Check your core usage while the game is running. Odds are you'll have individual cores pegging 100%.

    Switch it to DX11 mode, and it does the same thing, but the two cores handling the feed are maxed out near 100% usage. DX11 has more threading than previous versions, but it still has one active state, and the active state isn't threaded, so you create two states, and switch them back and forth to load into the GPU. If I had a 2080 Ti, or I was running SLI, DX11 would solidly bottleneck my ass at about the same performance level I can achieve with just the 1080 Ti.

    Nitrous is one of the first, and one of the very few, possibly the only engine, to achieve such stellar levels of parallelism. It's designed from the ground up for multi core systems, Warscape isn't.

    I’ve just checked the benchmarks with DX12 on. I lost around 20 FPS and there were significant stutters. My CPU core usage remained the same as well. I don’t have any cores reaching 100%. The closest I get to that is core 3 hitting between 90 and 95%. In normal usage in a regular 20vs20 unit battle my CPU Core 3 is anywhere between 50% to 90%.
    Post edited by lukemurtagh1995 on
  • psychoakpsychoak Posts: 2,359Registered Users
    That 60% drop on the GPU is one of the cores being pegged, but you're definitely not CPU bound in a typical situation, not substantially at least.

    As I said before, CA's DX12 implementation is buggy **** and so it wont help you.
  • lukemurtagh1995lukemurtagh1995 Junior Member Posts: 27Registered Users
    edited May 17
    psychoak said:

    That 60% drop on the GPU is one of the cores being pegged, but you're definitely not CPU bound in a typical situation, not substantially at least.

    As I said before, CA's DX12 implementation is buggy **** and so it wont help you.

    So I am bottlenecked by my GPU? I know in the Skaven benchmark my cores are occasionally at 99-100%, but to be fair the majority of battles, for me at least, don’t have near as many units as are in the Skaven benchmark, so I’m very very rarely seeing the same amount of CPU core usage in actual gameplay as I am in the Skaven benchmark. Or at least I presume this to be the case. From what I gather the Skaven benchmark is designed to give your system a hard time.

    Edit: during normal play I see my GPU usage at 99-100% far more often than any of my CPU cores, which makes me think I am GPU bottlenecked.

  • psychoakpsychoak Posts: 2,359Registered Users
    Yeah, with it overclocked to 5Ghz, you're avoiding a frequent bottleneck there, so you can squeeze out some more performance. Probably not a whole lot though, like I said, I'm CPU bound.

    Get a 2080 Ti, and you'll be CPU bound more often than not I expect. There is far less difference in your 5Ghz overclock versus the two core turbo of 4.5Ghz on the 9980XE, than there is between the 2080 Ti and the 1080 Ti, and it will be a fairly linear relationship.
Sign In or Register to comment.