4K over 1920*1080 - what is the penalty

[Ray Proudfoot 8,960]

Mike,

I agree that a test using a saved flight is the only way to get consistency. And regarding DL it’s not an issue as I won’t be flying at night except in rare cases.

You didn’t mention cloud cover. That surely is a big consumer of FPS. My 1080 copes very well but in very heavy cloud with multiple layers it can drop below 25 on occasions.

A test with the aircraft slewed at 1000ft just west of T5 at Heathrow Xtended facing east with clear and overcast skies and 150 Ai would be a good test for me. 1920x1080 and again at 3840x2160.

[AnkH 2,267]

Any specific reason why you do not follow the tip and test this yourself via DSR? Doubtful that the FPS numbers of someone elses rig can be compared to what you will get, this never worked since early FS days.

[Ray Proudfoot 8,960]

3 minutes ago, AnkH said:

Any specific reason why you do not follow the tip and test this yourself via DSR? Doubtful that the FPS numbers of someone elses rig can be compared to what you will get, this never worked since early FS days.

I’ve never heard of DSR until this thread. And as my TV is only FullHD is it a meaningful test?

[sultanofswing 57]

2 hours ago, Ray Proudfoot said:

But my primary interest is in fps, not how AA looks. 90% of my flying time is spent inside a cockpit and I'm sure the extra clarity of 4K would outweigh any shimmering or jaggies. That's quite subjective too whereas fps before and after is objective and that was the purpose of my question.

Things are a bit confusing because @slait reports virtually no impact on fps whereas @swiesma says his 1080 couldn't give him satisfactory performance.

@swiesma, that's a big % difference. Was 20% in clear skies and 50% in cloudy ones? That would make sense since multiple cloud layers does make the biggest impact on mine.

The simple answer appears to be a 1080 will be fine in southern California but for Europe and the UK in particular only a 1080i will do.

That is why I opened my post with "AA settings and what a person can tolerate are relative to the person looking at the sim".  If you don't mind jaggies than you can turn down AA and get the same performance you get at 1080p with a lot of AA. Me personally I cannot deal with jaggies and shimmering as it hurts my eyes.

[tutmeister 710]

All this is a bit subjective as everyone is using different settings not to mention different versions of the sim. I've been using 4k for a few years and started on a 970GTX and cpu @ 4.5 which was perfectly fine before dynamic lights came along. Currently using a 1080Ti and a kaby lake build running at 4.5GHz.

Honestly a 970 was fine if you just knock the sliders back a bit. Similar to a 780 I'd say. In fact I didn't see massive changes in fps due to resolution back then as I was cpu limited.

Ray if you post your settings and add ons I can replicate them and try at various resolutions if you want.

Chris

[Pete Dowson 558]

16 hours ago, Ray Proudfoot said:

Can anyone temporarily switch their display to 1920*1080 and run a test then do the same at native 4K? Honestly, that would be far more helpful than subjective posts.

You can a 4K test by simply using the nVidia facility to multiple the resolution. This will make P3D send than many more pixels. The work to bring in back down to your screen resolution is done by the video card, but ther'e certainly enough ooomf in your 1080 to handle that!

nVidia Control Panel - Manage 3D Settings - DSR Factors

You'll then find the higher resolution(s) selectable in the P3D menu.

When I was testing here with a view to having 2x 65" 4K TV screens to replace the projection screen, I measured something like a 20% drop in fps for 1 x 4K, which I thought was pretty good for 4x more pixels being pushed out by P3D.

Pete

 

[Ray Proudfoot 8,960]

I did some googling and found how to enable 4x DSR. Running my standard FPS test with the Cub positioned over T5 at EGLL facing east I get 24fps in 1920*1080.

The same with 3840*2160 set (so P3D window occupies a quarter of the screen) shows 25fps. So in that situation there's no penalty.

No whether that is a genuine test is debatable because looking at the image it looks no sharper than 1920*1080.

Chris, appreciate the offer but it may be simpler to provide a saved flt. Give me a few minutes.

@Pete Dowson, Hi Pete, see above.

[AnkH 2,267]

1 hour ago, Ray Proudfoot said:

I’ve never heard of DSR until this thread. And as my TV is only FullHD is it a meaningful test?

As written by the other user: yes, it is a meaningful test. And it is irrespective of the display you use. You can activate DSR in your nVidia display driver menue. Simply select all potential DSR factors. As soon as you did this, you will be able to select resolutions higher than FullHD in your graphics menue inside Prepar3d. Select 4K (3840 x 2160) and your nVidia GPU will render the image in 4K, then it will downsample it to FullHD. Like that, you can already test now, how your FPS will change using 4K. Of course, the downsampling process needs a little bit performance, but this is so little, that the results you will see will still be representative.

[Ray Proudfoot 8,960]

I modified my test flight and added 3 cloud layers - 1 x stratus and 2 x cumulus. FPS at 3840*2160 are still identical to 1920*1080 which is encouraging.

I'm attaching the saved flight if anyone wants to try it.

https://dl.dropboxusercontent.com/s/fo1npvlll1xiz37/FPS_Test_EGLL_3CloudLayers.zip?dl=0

[Pete Dowson 558]

16 minutes ago, Ray Proudfoot said:

The same with 3840*2160 set (so P3D window occupies a quarter of the screen) shows 25fps. So in that situation there's no penalty.

Why a quarter of the screen? You should be in full screen mode! (Alt+Enter).

16 minutes ago, Ray Proudfoot said:

No whether that is a genuine test is debatable because looking at the image it looks no sharper than 1920*1080.

I can't be sharper!  It isn't a genuine 3840*2160 as the video card is then downgrading it back to match your screen. Also the other DSR option is a smoothing setting, which i think defaults to 30. That's far too high and will make it even less sharp!

Pete

 

[Ray Proudfoot 8,960]

Just now, Pete Dowson said:

Why a quarter of the screen? You should be in full screen mode! (Alt+Enter).

I can't be sharper!  It isn't a genuine 3840*2160 as the video card is then downgrading it back to match your screen. Also the other DSR option is a smoothing setting, which i think defaults to 30. That's far too high and will make it even less sharp!

Pete

 

I am in full-screen mode. Switching to windowed mode reverts to my native 1920*1080 and shows P3D on the full screen. Or do I need to set that to 3840*2160 as well?

I'll play around with the smoothing setting.

[Ray Proudfoot 8,960]

I've setup 3 cloud layers with the top one over 30,000ft thick and as thunderstorm. Still no real change in fps.

I'll change scenery density settings and maybe that will show a difference..

[Kyle Schurb 89]

I hook my PC up to my 4K TV in the living room on occasion. I generally notice less than 10% performance decrease going from 1080p monitor to the 4K TV. In other words, negligible for me (~4-5fps). 

I disagree with those saying you're processing 4x the data. 4x the pixels, sure, but that's not a direct correlation to everything else going on in the background.

I have very similar specs, except a GTX 1070.

Another consideration is if you're going to v4, there's a noticeable performance increase with similar settings (not direct slider position as that's changed). So if you factor in the performance penalty going to 4k, I think you'll come out about even, but with a better looking sim.

[tutmeister 710]

I'll try your test flight later tonight.

You will only see a difference if you were already maxing out your gpu at 1080p. If you have spare overhead in your gpu then changing to 4k will not drop your fps.

Downloading and running gpu-z is good for things like this, go to the sensor  tab and you can see the percentage load on your gpu as well as lots of other useful stuff like video memory use and temperatures. You can then try resolutions and settings and watch it change.

Chris

[slait 169]

Agreed with the last couple of posters. A 1080ti is going to rob your wallet of $700-800 so unless you have money to burn you may not want to jump into that stylish card just because it's available. You may find the more economical 1070 or 1080 is plenty.