3080 reviews are out

[SharpeXB]

I'm more interested to know benchmarks for X-plane Vulkan since hopefuly, DCS will get it sometime in the next 10 years. Soon™

I didn’t see any improvement with Vulkan in XP11

[Svsmokey]

DCS NEEDS Massive optimization in it's basic code, Vulcan is sorely needed.

 

All indications are that Vulkan is yet years (plural) away . Kate and Chizh have both made statements to that

effect , though without that precise language .

 

I further fear the implementations of features such as dynamic weather , new effects , higher res models etc , etc will lead to further performance degradation in VR well before the engine changes arrive .

 

Perhaps we can change ED's mind regarding DLSS 2.0 to help keep DCS playable in the interim , at least on the graphics side .

[GunSlingerAUS]

So, just to be clear - will a 3090 deliver better FPS for me in the Reverb compared to my 2080 Ti? I thought it would, as I was under the impression that my GPU is the bottleneck in VR?

[GunSlingerAUS]

You have to put this in context. In 4K DCS I have no scenario that goes below my frame limited 60fps on my 2080Ti at max settings with no MSAA (not needed in 4K), and that includes SC.

 

Any GPU upgrade would not show any improvement on that.

 

The previous poster was seeing a ‘slide show’ in Normandy on his old 960, so of course he will see an improvement.

 

The value of a GPU upgrade depends on your starting point.

 

So you're running all settings maxed, including mirrors? I have a very similar system, but I do run either 2xMSAA or 4xMSAA as I get jaggies on the 55" screen at 4K, and I'm running the Tomcat with 512 mirrors. I get drops to around 45fps in some scenarios. I can't imagine your CPU is giving you a 30% boost over mine?

[GunSlingerAUS]

Perhaps we can change ED's mind regarding DLSS 2.0 to help keep DCS playable in the interim , at least on the graphics side .

 

I honestly don't think we'll see Vulkan in DCS - not unless they do a v3.0.

 

As for DLSS 2.0, I'm not sure if the engine is suited for it - it's drawing a lot of long distance scenes, and from what I've seen in games like Control, that's where DLSS doesn't do so well.

[SharpeXB]

So you're running all settings maxed, including mirrors? I have a very similar system, but I do run either 2xMSAA or 4xMSAA as I get jaggies on the 55" screen at 4K, and I'm running the Tomcat with 512 mirrors. I get drops to around 45fps in some scenarios. I can't imagine your CPU is giving you a 30% boost over mine?

I’m running everything maxed and AA 2x on a 4K monitor. Certain scenarios I which can tell are CPU limited never see improved frame rates from a new graphics card. Otherwise I’ll keep a constant 60 FPS. The obvious CPU culprits are when you’re on the ground with lots of close by aircraft etc. To test, just drop your resolution to 1080x1920 and see if your frame rate jump up. If not, it’s the CPU that’s limiting you.

Edited September 19, 2020 by SharpeXB

[Taz1004]

So, just to be clear - will a 3090 deliver better FPS for me in the Reverb compared to my 2080 Ti? I thought it would, as I was under the impression that my GPU is the bottleneck in VR?

 

3090 in theory should push higher fps in VR. As GPU is limiting factor in VR... other than few cases with lots of AI in MP where CPU will be limiting. Question is can 3090 push to 90. I would love for you to buy it and let us know. ;)

[Waxer]

As for DLSS 2.0, I'm not sure if the engine is suited for it - it's drawing a lot of long distance scenes

 

I think that is the issue. DLSS fills in the blanks and makes guesses. The nature of DCS - outside the cockpit at least is that you are scanning the sky for a bandit, who is a couple of fast moving pixels. Maybe the software would work if you just wrote script instructing the algo to leave bandit dots alone and render them fully.

 

RTX ray tracing and DLSS would certainly be a nice in cockpits however. But I'd rather see better multicore CPU optimisation as a priority than GPU bells and whistles like those. CPU generational performance is moving much slower than GPU performance.

[Geskes]

You have to put this in context. In 4K DCS I have no scenario that goes below my frame limited 60fps on my 2080Ti at max settings with no MSAA (not needed in 4K), and that includes SC.

 

Any GPU upgrade would not show any improvement on that.

 

The previous poster was seeing a ‘slide show’ in Normandy on his old 960, so of course he will see an improvement.

 

The value of a GPU upgrade depends on your starting point.

 

I think you are right in saying that for dcs 4k currently a 3080 does not make sense when you already in the 2080 range unless you have a high refresh rate monitor.

 

To be clear, I came from a 1080 gtx and on that card 4k dcs is unplayable. A 2080Ti was out of budget for me. I skipped the 20 series. And coming from there the 3080 is a no-brainer. It was cheaper then a 2080 super was this Monday. Going to test it with the rift now, wonder how the psu holds out when it also has to power the rift now...

[stormrider]

I didn’t see any improvement with Vulkan in XP11

I'm not talking about Vulkan vs not-Vulkan. I'm talking 2080ti Vulkan vs 3080 Vulkan in XP11. But in the end, all the signs indicate that Vulkan in DCS might actually not happen at all.

Edited September 19, 2020 by stormridersp

[Geskes]

Going to test it with the rift now, wonder how the psu holds out when it also has to power the rift now...

 

OK that totally sucked. Oculus software is saying now my computer doesn't meet minimum requirements. I don't think it is working properly. GPU use is only 30% but it is stuck at 40 fps. Even when looking up to empty sky it stays on 40 but even on the 1080 it would reach 80 fps when doing that. So for now, total joke but I wouldn't know why.

[Mud]

OK that totally sucked. Oculus software is saying now my computer doesn't meet minimum requirements. I don't think it is working properly. GPU use is only 30% but it is stuck at 40 fps. Even when looking up to empty sky it stays on 40 but even on the 1080 it would reach 80 fps when doing that. So for now, total joke but I wouldn't know why.

 

Dang, that sucks.

Could be that the Oculus software needs to be updated by them to properly support the cards.

 

In either case 40fps at 30% load does sound promising of what the card can achieve I think.

[imacken]

So you're running all settings maxed, including mirrors? I have a very similar system, but I do run either 2xMSAA or 4xMSAA as I get jaggies on the 55" screen at 4K, and I'm running the Tomcat with 512 mirrors. I get drops to around 45fps in some scenarios. I can't imagine your CPU is giving you a 30% boost over mine?

 

As I said, I don't use MSAA. I find it makes no difference on my 31" monitor and I don't like the effect it has on cockpit dials.

 

That's where your frames have gone.

[imacken]

OK that totally sucked. Oculus software is saying now my computer doesn't meet minimum requirements. I don't think it is working properly. GPU use is only 30% but it is stuck at 40 fps. Even when looking up to empty sky it stays on 40 but even on the 1080 it would reach 80 fps when doing that. So for now, total joke but I wouldn't know why.

 

Sure you haven't got ASW (or whatever Oculus calls reprojection) turned on by accident?

[coldViPer]

OK that totally sucked. Oculus software is saying now my computer doesn't meet minimum requirements. I don't think it is working properly. GPU use is only 30% but it is stuck at 40 fps. Even when looking up to empty sky it stays on 40 but even on the 1080 it would reach 80 fps when doing that. So for now, total joke but I wouldn't know why.

 

 

Have you checked that your PCI-e is using full speed (ie. via GPU-Z)? I ask this because one guy from finnish forum said that he installed 3080 with support bracket and wondered why this card is so slow! Until he checked the PCI-e speed, it was the slowest possible speed (PCIe 4.0 x1 or something)! And the reason was supporting bracket, it lifted the card a little so the contact was bad. He installed the card again without the supporting bracket and now PCIe speed was full 4.0 x16 and the performance like it should be.

[Geskes]

So for now, total joke but I wouldn't know why.

 

Forgot to disable vsync lol. So now the situation is much better. I tried some short flights on high preset with 2x msaa. The helmet switches between 40 and 80 fps a lot. It is very smooth though. But it seems very conservative on switching back to 40 so the gpu never gets under high load. So probably a 3080 over a 2080 in dcs rift s will only get you a little bit more periods in 80 fps.

[Geskes]

Sure you haven't got ASW (or whatever Oculus calls reprojection) turned on by accident?

 

So how do I disable all that jazz anyhow?

[Mud]

CTRL + 1 on the num pad to turn ASW off

[Geskes]

CTRL + 1 on the num pad to turn ASW off

 

I think I got it.

 

Did the huey free flight. Last part over the city. High preset 2x msaa, 100% trees. Now it uses the gpu fully and I get around 60-75 fps flying over that city area there. But fps vary wildly, probably that is just dcs.

 

Also tried christen eagle over caucasus. There framerates were little higher and often perfect 80.

 

I tried to record the performance but apparently afterburner does not record fps but only card metrics??

[Mr_sukebe]

How does it feel and look when compared to your old GPU?

[Geskes]

How does it feel and look when compared to your old GPU?

 

The first thing that is obvious is that it runs much smoother. On old gpu without asw it would be unplayable, now it is smooth, almost no stutters. Also because I can run with msaa now, it has less jaggies. So I am happy already. ASW is terrible for dcs especially in dogfights and such because the opponent will get blurry.

[SharpeXB]

I'm not talking about Vulkan vs not-Vulkan. I'm talking 2080ti Vulkan vs 3080 Vulkan in XP11.

Well XP11 performance is kinda off topic. But generally I understand Vulkan helps offload CPU tasks to the GPU which is why it would be a big help here.

[imacken]

Well XP11 performance is kinda off topic. But generally I understand Vulkan helps offload CPU tasks to the GPU which is why it would be a big help here.

 

Not file those who are GPU bound most of the time, it wouldn’t!

[SharpeXB]

Not file those who are GPU bound most of the time, it wouldn’t!

Then Vulkan probably won’t make any difference in that situation.

[imacken]

Then Vulkan probably won’t make any difference in that situation.

 

Exactly! It’s always a chasing-the-tail exercise!