1X 2080Ti vs. 2X 2080Ti vs. 1X TitanXP - P3D V4.3

[bbain1187 118]

14 minutes ago, RamonB said:

We simmers and gamers are just a drop in a bucket.
IMO nowadays, what drives the video card market is the CryptoMining industry. Some farms have hundreds if not thousands of cards working/hashing  full power 24/7.

Looks like I was misquoted. That was someone else.

I will add though, after looking at Rob's benchmarks again, that gain in the 'Low FPS' category was very impressive between the Titan and even 1 2080 Ti. This could make a big difference for those of us running 4k 30hz/vsync, as it is the dip below 30fps in graphically demanding situations that causes the bad stutters in this setup. The 2080 Ti keeping the low end high could ensure a smooth flying experience on approach to a detailed addon airport in rain and low clouds. I'm just hoping my i7 6700k @ 4.6 can keep up. 

Ben

Edited October 11, 2018 by bbain1187

[RamonB 449]

10 minutes ago, bbain1187 said:

Looks like I was misquoted. That was someone else.

Sorry, selected the quote from your post.
I corrected my previous post.

 

Edited October 11, 2018 by RamonB

[TravelRunner404 596]

22 hours ago, Rob Ainscough said:

I don't know what "optimized" means in the context I see it used by members on AVSIM and elsewhere.

I think it’s the general feeling that something seems off about the P3D performance. I have only been simming for about 18 months and that’s the general feeling I get when I sim. Especially since the buildings look like the same ones from 20 years ago. It has this dated early 2000’s look to its buildings. That combined with generally poor performance when rendering a large city gives the feeling that it could and should be a lot better. I know it’s complicated but why can’t we have the Aerofly New York in P3D and have other cities look like it’s 2018? More of a genuine question than a complaint when I ask that. 

[macwino 14]

Ben, I can confirm your expectations. I've had my 2080ti for a week now, and it makes a marked difference from my 1080ti when approaching a detailed airport in a complex airplane when running at 4k 30hz/vsync. The improved control on approach alone, to me, has been worth the cost of the upgrade. To me, the percentage improvement over the 1080ti is irrelevant; what matters is the smoothness and improved immersion that the card provides. And I think my  experience will get even better when my 9900k arrives in a week or two, to replace my 7700k, which is overclocked to 5.0. 

Unfortunately, Rob is right when he says that there's something wrong with the way this card and XP11 interact. So far I too am seeing little or no difference at all compared to my 1080ti.

Robert

[mpw8679 428]

So basically if u already own a 1080ti there is not much value upgrading to 2080ti.  So how bout 2x1080ti's?

[bbain1187 118]

1 hour ago, RamonB said:

Sorry, selected the quote from your post.
I corrected my previous post.

 

No worries!

39 minutes ago, macwino said:

Ben, I can confirm your expectations. I've had my 2080ti for a week now, and it makes a marked difference from my 1080ti when approaching a detailed airport in a complex airplane when running at 4k 30hz/vsync. The improved control on approach alone, to me, has been worth the cost of the upgrade. To me, the percentage improvement over the 1080ti is irrelevant; what matters is the smoothness and improved immersion that the card provides. And I think my  experience will get even better when my 9900k arrives in a week or two, to replace my 7700k, which is overclocked to 5.0. 

Unfortunately, Rob is right when he says that there's something wrong with the way this card and XP11 interact. So far I too am seeing little or no difference at all compared to my 1080ti.

Robert

Thank you for that, Robert. I just spoke with Nvidia customer support and I can still send it back for up to 30 days from when it was shipped so I'm gonna take the plunge. Will report back my results. 

Ben 

[adyfoot 19]

44 minutes ago, macwino said:

Unfortunately, Rob is right when he says that there's something wrong with the way this card and XP11 interact. So far I too am seeing little or no difference at all compared to my 1080ti.

This is surprising to hear - have you bumped up your AA settings with the RTX card?  I'd expect you to be able to do that and maintain the same FPS you had, with the 1080ti, at a lower AA setting.  That'd be logical.  But computers and logic don't always go hand in hand 🙂

[Guest]

9 minutes ago, macwino said:

there's something wrong with the way this card and XP11 interact. So far I too am seeing little or no difference at all compared to my 1080ti.

Something is up, if I run just one 2080Ti, I get a scrambled intro screen where I can select new flight or resume.  If I run two 2080Ti with NVLink/SLI enabled then the Intro screen is fine … the default for NCP profile is single GPU.  I used DD's Seattle for XP11 and was in the 14-19 FPS range on the 2080Ti's that's about 3 FPS lower than my old TitanXP same location, same aircraft, no weather.

Tried the usual tricks in XP to disable Thread Optimization, enable, and auto … results were all the same, low FPS.  Well, I'm going to be building my i9 9900K next week or following week so that will require a complete OS wipe and re-install of all my sims and add-ons so I'll see if that helps with XP11.

AF2 responded well, but when one is usually over 100 FPS with everything maxed out, that's not really testing the GPU's load.

27 minutes ago, TravelRunner404 said:

Especially since the buildings look like the same ones from 20 years ago. It has this dated early 2000’s look to its buildings.

Agree, but LM have always avoided working that area since 3rd party has solutions for existing buildings that do look much better.  LM have primarily focused on the engine, leaving the "content" to 3rd party … the only exception here is perhaps "SpeedTrees" which sadly most disable and use the old X trees (myself included). 

If you look at "SpeedTrees" they are definitely better than any "Trees" I've seen implemented in any other flight sim, but they do seem some "tweaking" also … their density needs to be higher and some of the trees have odd colors … there is a SpeedTrees SDK but it's about $6000 per dev license so no one has really attempted to do anything with SpeedTrees yet.  Shame, because SpeedTrees animate to wind and even rotor wash and they are seasonal and volumetric.

Another problem is the shifting of terrain … most noticeable in mountains areas … anyway, all these are items that can be and/or are being addressed over time without re-writing the engine from scratch.

The loading and unloading of airport textures is another issue.

So yes, there are optimizations that can be done and over time are being done … breaking backwards compatibility is the "controlling" factor … trying to accomplish some of these changes without requiring 3rd party content providers to adjust their existing work is a big challenge.  There may eventually be a version of P3D Vx.x where FSX content will simply no longer be tolerated … if this ever does happen (and I hope it does because FSX is still holding us all back) there will be a "dual path" approach so that content providers can build/convert over time provided they feel they can recover their costs.

Anyway, this thread has definitely strayed.

Cheers, Rob.

[z06z33 763]

Rob would it be a worthwhile upgrade from a GTX1080 for p3d? Im running an 8700k at 4.8ghz

Edited October 11, 2018 by z06z33

[macwino 14]

42 minutes ago, adyfoot said:

This is surprising to hear - have you bumped up your AA settings with the RTX card?  I'd expect you to be able to do that and maintain the same FPS you had, with the 1080ti, at a lower AA setting. 

I didn't increase my AA settings because I was ok with their being one notch below max. But I did fiddle with some of the other graphics settings and found that I was soon getting the message that XP was slowing down and that I needed to reduce my settings. Not very analytical for sure, and more testing obviously needs to be done. But I'm concentrating on P3D at the moment.

I didn't experience the graphical issues that Rob has mentioned. And I forgot to try altering the Thread Optimization setting as Rob did but will do so when I next give XP11 a try.

Robert

[westman 257]

Hi rob 

tested with your high settings , not same addons 

CPU I9 7920X 10 cores all synced @5.1ghz ( disabled 2 cores in bios) Gpu 2 x 1080Ti Galaxy little overclocked.

Its very close to your result, the playback file ar not exactly the same but tryed to do it as close as possible.

https://www.dropbox.com/s/64hw4kyipu08yjq/ESNQ-test.png?dl=0

In dx11 the diff to a 1080Ti is not as big as in DX12 if you lock at FireStrike extreme tests now thats that is not P3D.

Add-Ons active:

• Orbx ESNQ (with Dynamic Lighting)
• Orbx FTX Global
• ORBX VECTOR ,Have not//   Orbx Norway
• Orbx OpenLC EU
• Have not//     Orbx HD Trees
• PILOT's FS Global Ultimate 2018 Mesh
• PTA RD preset TOGA Projects ENVTEX and / not /ENVSHADE (set to Quality mode)
• ECZA     //FSFX Packages (OldProp)

 

 

Edited October 12, 2018 by westman

[bbain1187 118]

To repeat what I just posted in the other thread, I just loaded up the FSL A319 in the most demanding situation I could imagine, 3 cloud layers and rain at DD Seattle/Orbx PNW at dusk with moderate-high settings with dynamic lighting on in 4k/30hz and it didn't stutter or dip below 30fps until I increased the autogen to high levels. I'm now Cpu-bound with my i7 6600k @4.6. Happy camper here. 

Ben

[Ray Proudfoot 8,954]

4 hours ago, bbain1187 said:

To repeat what I just posted in the other thread, I just loaded up the FSL A319 in the most demanding situation I could imagine, 3 cloud layers and rain at DD Seattle/Orbx PNW at dusk with moderate-high settings with dynamic lighting on in 4k/30hz and it didn't stutter or dip below 30fps until I increased the autogen to high levels. I'm now Cpu-bound with my i7 6600k @4.6. Happy camper here. 

Ben

You don't say what graphics card you're using. PLEASE put your spec in your sig otherwise users will have to keep scrolling back in messages to see your setup.

[threexgreen 3,139]

16 hours ago, bbain1187 said:

To repeat what I just posted in the other thread, I just loaded up the FSL A319 in the most demanding situation I could imagine, 3 cloud layers and rain at DD Seattle/Orbx PNW at dusk with moderate-high settings with dynamic lighting on in 4k/30hz and it didn't stutter or dip below 30fps until I increased the autogen to high levels. I'm now Cpu-bound with my i7 6600k @4.6. Happy camper here. 

Ben

Is that with the 2080Ti? If so, how does that compare to your former GPU?

[HiFlyer 9,359]

On 10/11/2018 at 2:05 PM, Rob Ainscough said:

"SpeedTrees" they are definitely better than any "Trees" I've seen implemented in any other flight sim

Speedtrees seem much better in DCS, though. I keep wondering why the P3D version is so weird, especially since the Microsoft FLIGHT versions were also pretty nice. This is one of the reasons I fly so many sims, because no single sim has it all, and there is a lot they can learn from one another. Also a lot we can learn from seeing how other sims implement things.