Much better video card = same frame rate...

[Sub2K]

I agree with everything you mentioned TJ, now I just need to figure out how to 'tune' my system to take advantage of the new card as much as possible.

 

BTW, I just ran the Single Threaded CPU MARK and got 2391 (91th Percentile) - They sure don't make things like they used to... :)

 

Thanks a lot for all the feedback/help!

[Tj1376]

I agree with everything you mentioned TJ, now I just need to figure out how to 'tune' my system to take advantage of the new card as much as possible.

 

BTW, I just ran the Single Threaded CPU MARK and got 2391 (91th Percentile) - They sure don't make things like they used to... :)

 

Thanks a lot for all the feedback/help!

 

 

No problem. Holler if you have any questions.

 

Also if you get a chance, let me know how much memory you’re using and your graphics setup in DCS. I’m trying to decide if I want to add more memory (at a ridiculously stupid price) or wait another year and rebuild. I’m not much better than you- 4770k with a 780ti. My biggest issues are the lack of GB on the gpu and in the system. I constantly run at 15.8gb. :(

 

TJ

 

 

Sent from my iPhone using Tapatalk

[erniedaoage]

Hi Sub2k, my suggestion is try running dcs in window mode so uncheck fullscreen and see if you gain something on my side it doubles fps and i still don't know why.

[Sub2K]

@erniedaoage,

 

Good suggestion, but I have tried both Windowed and Fullscreen with no change in FPS.

[Sub2K]

@TJ,

 

I just checked, and my RAM usage holds steady at 57% (about 13.7 GB/24 GB).

 

As soon as I can upload a screenshot, I'll post my settings.

[TexasWarbird]

You should upgrade to a 3770k CPU, it's the same socket (1155) Also 1920x1080 try that res and check performance.

[BoneDust]

@BoneDust,

 

At this point, I would settle for a 15 FPS gain... :(

 

I hear man, but for me the card cost over $1,100 in Canada, so that much to gain 15fps is not great. I get a steady 45fps; never below but never above and I've tried all the tricks people say.

[Headwarp]

We're not even sure a 2080Ti can handle high settings + msaa and get over 45 fps in DCS VR yet, which is basically ASW kicking in because you can't get a stable 90FPS.

 

But - there's more to it than simply swapping GPU.. check your DCS profile in NV control panel, if it doesn't exist add it..make sure power management option is set to "maximum performance"

Sometimes it's best to completely uninstall everything nvidia with DDU and install new when swapping cards as well. Also - be aware of steam's super sampling settings, personally I recommend setting it to "manual" and 100% because it will default to higher resolutions, which impact performance. From there start at 1.0 PD in DCS, and if it allows for it, turn the pixel density up if you can until you can no longer maintain 45 fps and then drop back down a notch to where you can. This is DCS's method of supersampling for VR, and turning it up will make your GPU work harder while providing a clearer image. This also should have the effect of lowering the ceiling for cpu clockspeed bottlenecks, if any are present to begin with and you manage to pin your 1080Ti.

 

I some times get 60-80 FPS @ altitude in my odyssey with my 980Ti.

 

one of my mates is also in ASW with his 1080Ti on a 7700k (turbos to 4.5ghz out of the box, I've at least heard mentioned more than once capable of being oc'd to like 4.9ghz)

 

Pretty much no cpu/gpu upgrade is going to make DCS run without ASW with current hardware, unless the RTX series turns out to be amazing and ED manages to allow for Deep-learning AA/SS on the tensor core, which I'm not holding my breath for considering the time development takes with this team. Although it would be nice because MSAA currently is a huge FPS hit in DCS and it's too shimmery without it at least 2x, even if turning it off allows for much higher framerates.

 

That being said- 45 fps in VR in DCS is absolutely playable, its a feature of your VR headset kicking in to keep you from getting woozy by using rendering techniques to keep what you see looking smooth. You can probably still achieve it with high textures + medium shadows and pixel density increases where I have to compromise with my 980Ti with Medium textures and shadows off or flat.

 

sandy bridge CPU's OC'd to 4.5ghz or more are still excellent gaming CPU's, especially given that VR and high resolutions/supersampling put more workload on the GPU, and that also because a single GPU still won't saturate pci-e 2.0 x16. However moving up to a 1080Ti or better I hear some people claiming they can actually utilize 5ghz cpu. But then again I'm unsure at what resolutions people play at. Anything GTX 970 or better will likely be CPU bottlenecked @ 1080p allowing for FPS gains by increases in CPU clockspeed or faster architectures, but the higher the resolution, or in VR's case rendering each eye, the more likely for your gpu to be the bottleneck, and the less clock speed matters on the CPU.

 

ASW might make your gpu work less hard because it only needs to achieve 45FPS to function which flat out pins my poor 980Ti in DCS World even at 1.0 PD, although the Odyssey's native resolution is on par with the Vive Pro, so it's comparable to a pixel density increase in the rift/vive.

 

Basically - moral of my story - If you're getting and maintaining 45 fps in DCS world in VR - rejoice - framerate isn't a reason for upgrading CPU/MOBO/RAM. If you absolutely must hit 90 FPS in VR, we're simply still waiting on a GPU that can achieve it in DCS with MSAA on. That's going to be a few years. I say that coming from an i5 2500k @4.5ghz rig to my current 8700k 4.7ghz and getting almost identical performance in DCS at least on my 3440x1440 monitor which allows for higher settings than VR @ odyssey/vive pro resolutions. I didn't have my odyssey when I was still using that system. Same GPU between both systems, that = gpu bottleneck. It's just against my principals to upgrade my GPU without skipping at least one generation, no matter how much my fancy monitor was begging for more gpu power (mostly in flight sims go figure). This is also why I buy the best version I can when I do upgrade, looking to get 3-5 years use out of it. Getting my money's worth.

 

If you do upgrade your core components, do it because of new features and drivers that will get updated for at least a couple years, unlike 2nd gen intel boards that are using drivers from 2010 and 2013. PCI-E x4 NVME drives, faster ram, pci-e 3.0/4.0, RGB lighting (cuz it's shiney/seksai) and better multitasking/production performance, or because the latest intel CPU's hit 5ghz pretty easily and you just like being a tech nerd because I doubt you'll see much, if any, difference in gaming performance based on personal experience. Heck, intel's 9th gen will feature mainstream 8 core processors in response to ryzen that should be able to achieve clockspeeds similar to the 7th and 8th gen intel cpus, although that won't impact gaming much, as we're mostly still dependent on single core performance, but - for multitasking/production work on a rig that also mains as a gaming PC that sounds pretty awesome considering they will have the single core performance to maintain the position of King for gaming cpus.

 

*Final edit of this lengthy informative post* - you can also disable ASW/ASR in steamVR (maybe occulus home too i've never used a rift to know) altogether and get an idea of what framerates your setup is actually capable of with your current settings, but if you're prone to motion sickness.. I wouldn't wear that headset for very long. And ensure only "Allow asynchronus reprojection" is checked in SteamVR settings, leave "Allow interleaved reprojection" unchecked. When you're ready to live with 45FPS and smooth gameplay and turn the feature back on that is. :)

 

Also forgive me if DCS runs natively out of Occulus home, my WMR headset uses steamVR to launch DCS, but I'd imagine Occulus home would have similar settings. It's like not being able to help AMD users with specifics because nVidia likes to take my money, but I can at least give a general guideline.

Edited September 2, 2018 by Headwarp

[Sub2K]

@Headwarp,

 

Thanks for some good insight into your experience and for providing some recommendations.

 

I use a TrackIR 5 and I would be thrilled if I could get a steady 45 FPS. Unfortunately, as I mentioned in my previous posts, I seem to be stuck in low 20s ad vitam aeternam.

 

I also get up to 60 FPS when I'm basically looking at... nothing.

 

I'll keep my trials and errors testing going, and hope that Vulkan may help us all someday. :)

[Tj1376]

Basically - moral of my story - If you're getting and maintaining 45 fps in DCS world in VR - rejoice - framerate isn't a reason for upgrading CPU/MOBO/RAM. If you absolutely must hit 90 FPS in VR, we're simply still waiting on a GPU that can achieve it in DCS with MSAA on. That's going to be a few years. I say that coming from an i5 2500k @4.5ghz rig to my current 8700k 4.7ghz and getting almost identical performance in DCS at least on my 3440x1440 monitor which allows for higher settings than VR @ odyssey/vive pro resolutions. I didn't have my odyssey when I was still using that system. Same GPU between both systems, that = gpu bottleneck. It's just against my principals to upgrade my GPU without skipping at least one generation, no matter how much my fancy monitor was begging for more gpu power (mostly in flight sims go figure). This is also why I buy the best version I can when I do upgrade, looking to get 3-5 years use out of it. Getting my money's worth.

 

 

Great info- bummed to hear that the upgrade didn’t net much benefits. I was hoping the 10-15% better single thread performance would contribute to a noticeably better frame experience. Thanks for sharing your experience so other can learn!

 

TJ

 

 

Sent from my iPhone using Tapatalk

[Headwarp]

@Headwarp,

 

Thanks for some good insight into your experience and for providing some recommendations.

 

I use a TrackIR 5 and I would be thrilled if I could get a steady 45 FPS. Unfortunately, as I mentioned in my previous posts, I seem to be stuck in low 20s ad vitam aeternam.

 

I also get up to 60 FPS when I'm basically looking at... nothing.

 

I'll keep my trials and errors testing going, and hope that Vulkan may help us all someday. :)

 

Just a guess here.. you aren't running with civillian traffic on are you?

 

Turn that off. It's a resource hog. Also - might be worth disabling MSAA in game and enabling FXAA in NV control panel. You'll get shimmers...but framerate will go up quite a bit. Also worth running "dcs_updater.exe repair" as DCS can be weird about graphics driver updates sometimes. Without msaa on my 3440x1440 monitor i'd easily be pushing 80-100 fps @ 3440x1440, turning it on dropped me to about 50-60 at low altitudes. Which is also similar to my experiences on my 2500k build. On both builds - 980Ti would be @ 99% utilization in just about any game I play at that resolution.

 

The only other thing I can think of is open task manager and make sure your CPU is indeed running at 4.4ghz and not being throttled. If it isn't running at 4.4ghz, clean or replace your CPU heatsink + apply some new thermal paste because if you don't dust them regularly they will get caked up in dust. It was a flight sim that made me investigate performance issues I was having, when I realized my CPU wasn't even running @ stock clocks. It was when I replaced my cpu cooler that I overclocked it for the first time. My 2500k OC'd to 4.5ghz on air. I can only imagine I could've gotten it faster with an AIO liquid cooler. I'm doubting with your benchmarking scores that heat could be the issue, but worth looking at.

 

And Im sorry I seemed to hone in on Bonedust's VR experience in this thread when I wrote that post and somehow managed to think you were using VR too. I mean you're running at a pretty high resolution with 3 screens... but I imagine there's a setting being overlooked somewhere causing you to run at such low fps. You say you get the same framerate you had with a 1060, and I just don't see that relating to a CPU bottleneck, unless the CPU was throttling due to heat. A 1060 would surely be pegged trying to run 6,220,800 pixels which is a bit over 2 million more pixels than i was running on my 980Ti (which is closer to a 1070 in performance). I.E. - if my 980Ti was doing all the work @ 3440x1440, your 1060 should definitely have been doing all the work @ 5760x1080, and you should be seeing higher framerates with a faster gpu. Then again - if you've even got one core of your cpu pegged to 100% (which would =12.5% cpu usage total with 8 threads) and no heat issues when flying then I'm just completely wrong and my 2500k was simply amazing.

 

P.S. - Click the Go advanced button at the bottom of any post you're trying to edit or make, at the top will be an icon that looks like a paperclip, it opens a new window that allows you to upload files like screenshots and what not. Once you've uploaded them, you use the dropdown menu next to the paperclip icon, click your file and it will attach that screenshot to your post based on where your cursor is. Use PRNT SCRN to take a screenshot of your system settings in DCS. They get saved to C:\users\xxxx\saved games\DCS.openbeta\screenshots\ (if on the beta version) You can also use windows snipping tool to show us what you've done within NV control panel.

Edited September 3, 2018 by Headwarp

[Aluminum Donkey]

@Headwarp,

 

Thanks for some good insight into your experience and for providing some recommendations.

 

 

Sub2K, I forgot to ask you something!

When you overclocked your CPU, did you OC only the cores, or did you also OC the cache?

 

Reason I ask, is that there are separate multipliers for the cores and the cache memory on a CPU. If you aren't sure, go into your BIOS settings and make sure you're also running the cache as fast as you can get it to go while being stable.

 

If your cores are running at 4.4 GHz (which is excellent) but the cache multiplier is only, say, 28 or 30, it won't improve matters much. It might take more voltage than you're normally used to, but your CPU can probably handle 1.35V to maybe 1.4V without getting too hot.

 

Let us know what happens :)

AD

[Sub2K]

@ Headwarp

@ Aluminum Donkey

 

Finally! I've attached two screencaps taken in Free Flight over Dubai:

 

- My current settings

- My Task Manager resources

 

See anything weird/unusual??

 

P.S. I've since turned off Smoke Density.

Edited September 3, 2018 by Sub2K

[derodo]

Much better video card = same frame rate...

 

Exactly like my setup.

 

One logical core maxed out while the other 7 are not doing much...and the GPU at 60%.

 

I’d say you’re CPU-bound, just like me.

 

We don’t have too many options but cranking down CPU-heavy settings (trees, civ traffic, etc) and hope the Vulkan revamp to kicks ass and allows for better multi-core performance.

 

Edited September 3, 2018 by derodo

[David OC]

@ Headwarp

@ Aluminum Donkey

 

Finally! I've attached two screencaps taken in Free Flight over Dubai:

 

- My current settings

- My Task Manager resources

 

See anything weird/unusual??

 

P.S. I've since turned off Smoke Density.

 

You are maxed out on the last core, once this one core is max out it basically holds the others to ransom and makes them wait. We are hoping Vulkan will help spread this out, we just don't know how long it will take? Vulkan would possibly allow you to keep the CPU and easily fed the 1080ti by utilizing the cores you have now better.

 

The first graph is exactly what's happening to you on your last core. With the "possible DCS Vulkan??" graph below that.

 

 

This pic below is my PC on one monitor @ 1440p trying to find my max fps output on max game settings. I have set process lasso to single thread performance here, so DCS is only using the 4 main cores. Note that the GPU graph holding 100% utilization when running sim, this is holding my cores at 5Ghz. I general have it set to 4.8 most of the time.

 

 

I would like to see you hold your card at 100%, if you cannot, it's generally not getting fed enough. Like I said above tho, Vulkan could change all this perhaps....

Edited September 3, 2018 by David OC

[Tj1376]

@ Headwarp

@ Aluminum Donkey

 

Finally! I've attached two screencaps taken in Free Flight over Dubai:

 

- My current settings

- My Task Manager resources

 

See anything weird/unusual??

 

P.S. I've since turned off Smoke Density.

 

 

Something is odd here- you are only using 13gig of memory on Caucasus with those settings? I had terrain at low and with much worse preload radius and would use 14gig. I recently threw in 8 gig more (had two 4 gig sticks) just to see if it would boot and am now using 20-21 gig of ram on Caucasus.

 

Why would he have so little ram usage?

 

Does Nvidia offer a way to see what the onboard ddr5 is doing? That 1080ti has 11gb, I wonder if DCS is using that?

 

TJ

 

 

Sent from my iPhone using Tapatalk

[Tj1376]

Exactly like my setup.

 

One logical core maxed out while the other 7 are not doing much...and the GPU at 60%.

 

I’d say you’re CPU-bound, just like me.

 

We don’t have too many options but cranking down CPU-heavy settings (trees, civ traffic, etc) and hope the Vulkan revamp to kicks ass and allows for better multi-core performance.

 

 

 

 

Yes- the screenshots clearly show a cpu bound scenario. The primary core is maxed and the gpu is only 60%. About all you can do from here is increase resolution (which taxes the gpu and not the cpu.) however, 1440 monitors aren’t cheap and plane spotting in this resolution isn’t exactly easy.

 

TJ

 

 

Sent from my iPhone using Tapatalk

[Sub2K]

@derodo,

 

Alright then. I guess we wait for Vulkan... :(

[Sub2K]

@TJ,

 

Unfortunately, the native resolution of 2 of my 3 monitors is 1080.

[Sub2K]

@ TJ,

Actually, that screencap was taken over Dubai.

 

Also, I have my pagefile set to 16 GB. Could that be the difference?

Edited September 3, 2018 by Sub2K

[Tj1376]

@ TJ,

Actually, that screencap was taken over Dubai.

 

Also, I have my pagefile set to 16 GB. Could that be the difference?

 

 

No- you’ll only hit the page file if you run out of memory. And you’ll feel it as a hard studder.

 

My guess is that DCS is loading most of the terrain into the 11gb of on board memory that GPU has and that’s why your seeing low DDR3 memory usage. This is a great thing- ddr5 is much faster.

 

It’s easy to test if you still have that 1060 laying around. Just swap them and see what your physical memory does (I’d expect it to go up quite a few GB.) If Newegg and Amazon keep teasing me with super low 1080 prices, I might try this test in the next day or two. :)

 

TJ

 

 

Sent from my iPhone using Tapatalk

[Headwarp]

Personally, windows 10 handles page files rather well. I'd just let it do it's auto management thing.

 

You are definitely maxing out one of your cores, and you're @ 4.3ghz. I still feel like you should be able to get more FPS than I did on my 2500k rig which i had at 4.3ghz using a 980Ti. What kind of frames do you get over Caucasus?

 

Cheaper solution than building an entirely new computer - throw a corsair h115I or something in your rig as a CPU cooling solution rather than Air, I have an h110I and my 8700k doesn't even get above 70C when stress testing @ 4.7ghz on all cores. I'd wager you'd be able to get that 2500k up to at least 4.7ghz, which should offer you a framerate increase. 4.5ghz was about the max i could push on air with my sandy bridge, and I backed that down to 4.3 because I was reaching 85C during stress tests. Disabling hyperthreading might allow for higher clockspeeds with lower temps as well. But frankly I'm pretty sure sandy bridge heatspreader is soldered to the cpu unlike newer intel chips, meaning easier to cool.

 

I wouldn't worry about the cache frequency, all I ever did on the 2500k was adjust the clockspeed multiplyer and the voltage, and like 2 other settings. I have adjusted the cache frequnecy on my newer rig, but, my reading and experience leads me to believe upping the cache frequency has minimal effect on gaming performance vs clockspeed adjustments.

 

But I do have to wonder how you could've been cpu bottlenecked with a 1060 when my gpu was and still is a bottleneck at lower resolutions with a more powerful card.

 

Check your CPU usage when no games are in the background, and pretty much idle on the desktop, it should only be between like 1-5%. If it's much higher I'd be getting the windows development kit for Poolmon to try and track down faulty drivers. I had to hunt down intel drivers for a couple components rather than from EVGA's product page because of faulty drivers eating up cpu cycles and/or memory. I think specifically, the NIC drivers had to be a specific version, and perhaps the sata controller and chipset drivers even. 2nd gen boards haven't received proper driver updates for almost a decade now. If it does turn out to be something like faulty onboard nic drivers - you could even fix that with a pci-e nic card. Also check memory usage, faulty drivers may not always directly affect cpu usage @ idle.

 

The only other thing I can think of, that explains why at a higher resolution you were cpu bound witha weaker gpu than my 980Ti is that I built my 8700K before the fixes for meltdown/spectre came and older intel chips took the hardest hit.. but everything I've read says that you won't really see the effects of that when it comes to gaming. Which is also a little disheartening, because I kind of had plans for old sandy since I'd have a 980Ti to throw in it afterI upgrade my GPU.

 

That being said - my experience ends at the 980Ti - I might learn some things when I do ugprade, hopefully this month. For the life of me - the fact that with more pixels, your 1080Ti is giving the exact same framerate as your 1060 as if you were CPU bound on both cards where my 980Ti has been at 99% ever since I got a 3440x1440 monitor on the i5 counterpart of your exact build has me baffled.

 

*edit* Googled "i5 2500k after meltdown" and found some reddit threads regarding 2500k and 2600k saying they didn't notice much of an impact as far as gaming benchmarks. - I personally think it's either some kind of system driver issue or just a matter of more efficient cooling and upping your cpu clockspeeds, unless maybe your ssd has no free space, because it likes free space. Your 1060 should have been pinned at 99% usage. The 1080Ti might be able to use higher clockspeeds at that resolution, idk, but the SAME framerate between the two indicates you were also bottelnecked on the 1060 and that just doesn't make sense to me without something causing higher cpu usage than normal.

Edited September 3, 2018 by Headwarp

[Tj1376]

Much better video card = same frame rate...

 

But I do have to wonder how you could've been cpu bottlenecked with a 1060 when my gpu was and still is a bottleneck at lower resolutions with a more powerful card.

 

That being said - my experience ends at the 980Ti - I might learn some things when I do ugprade, hopefully this month. For the life of me - the fact that with more pixels, your 1080Ti is giving the exact same framerate as your 1060 as if you were CPU bound on both cards where my 980Ti has been at 99% ever since I got a 3440x1440 monitor on the i5 counterpart of your exact build has me baffled.

 

 

Resolution is the reason. You went from 1080p to 1440p- original poster did not (he is still 1080p.) Increasing resolution puts relatively little strain on CPU but tortures the GPU.

 

This is also why most folks don’t recommend a 1080ti for 1080p gaming- you just aren’t going to see any major improvements because at 1080p there still isn’t enough CPU power to push that card.

 

TJ

 

 

Sent from my iPhone using Tapatalk

[ElementLT]

also make sure you have your power management mode set to "Perfer Max Performance" in your Nvidia control panel. So you can take full advantage of your new card.

 

[Sub2K]

@TJ,

 

Actually, since I'm using 3 x 1080p monitors, I'm at 5760.