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

[Guest]

Here are the results of some testing I did in P3D V4.3 comparing 2X 2080Ti (SLI) vs. 1X 2080Ti vs. 1X TitanXP.

CPU 7900X 5Ghz, 32GB DDR, Win10 1809, M.2 SSDs

I tested with 3 Graphic settings scenarios using a single recorded flight so that playback was always consistent.  In all cases AI Traffic and Road/Boat Traffic was disabled.  Weather was clear.

The graphics settings (cfg files) I used are (4K and TEXTURE_SIZE_EXP=10 in all tests)

Medium (All Terrain settings at Max with exception of Building AG = Very Dense, DL enabled)

https://drive.google.com/open?id=1c2mIEHf3HvNKQoDNNYeRNsvFOiimUljn

High (Medium settings + DR = Medium, 4XSSAA, +Shadows SimObject & Buildings, + Reflections Terrain & Buildings)

https://drive.google.com/open?id=1Vd_9Cc8zSlWk6ut0BfrdD9jA6heMnZjo

Very High (High settings + DR = Max, 8XSSAA, +Reflections All)

https://drive.google.com/open?id=16Tj-rqlbFTWdkYUvr8kYGNAQUiciMJEw

Add-Ons active:

• Orbx ESNQ (with Dynamic Lighting)
• Orbx FTX Global
• Orbx Norway
• Orbx OpenLC EU
• Orbx HD Trees
• PILOT's FS Global Ultimate 2018 Mesh
• TOGA Projects ENVTEX and ENVSHADE (set to Quality mode)
• FSFX Packages (OldProp)

Saved Flight File(s):

https://drive.google.com/open?id=1HAmL-T6DtilKGy0_3iMMwF2wvXdDFyaO

https://drive.google.com/open?id=1tTc2npHyaoZyBWkhCq_RI_IhPfM4rXvd

Data Logging was done with MSI AfterBurner and the resulting HML files (these are CSV files) were imported into MS Excel and converted to XLS.  Beginning and ending rows of data were removed (they had 0 FPS because I was showing graphics settings and/or loading the recording) so as to be a consistent active set of data.

Excel data for:

Medium TitanXP

https://drive.google.com/open?id=1vtdUhyObZtZAx8mIrcGLYrm8enD8dsEY

High TitanXP

https://drive.google.com/open?id=1lOLEywAxRusnf6Z8EWv_raSL2L04IJts

Very High TitanXP

https://drive.google.com/open?id=1VO9rS-E4FD6yKWpTFXZ4nAhYfhtZv926

Medium 1X 2080Ti

https://drive.google.com/open?id=1ogvnDZrIO_w5zh_CR4sGGfvUd1V2hORi

High 1X 2080Ti

https://drive.google.com/open?id=1XIN6UnNgDGpeAsj6wgeyuIcZmUbQ_jmM

Very High 1X 2080Ti

https://drive.google.com/open?id=1084SgYlIUsfil6h6IpI8MhJfQDNi9G3q

Medium 2X 2080Ti

https://drive.google.com/open?id=171CUmF2-QtL-EExaTi6Snce1xD4Av_Pu

High 2X 2080Ti

https://drive.google.com/open?id=1DyWN2kRMIyZ9OAJnDQDlenA4RXIrMeSs

Very High 2X 2080Ti

https://drive.google.com/open?id=1aEQU_B08t6u6mNpDp8gG_UmEzDcuPvVM

Below is the sample flight playback used for testing (original flight was recorded for consistency and used for playback for each test iteration).  This video contains the graphics settings I will ultimately "settle" with for my "normal day to day flying".  I do have a video of every single test scenario but didn't have time to upload them all, so this video represents the test flight combined with the final settings I'll be using for "day to day" flights (switching between Unlimited and TFR=31 as needed).

Cheers, Rob.

[ppo 67]

Tks for this Rob.

Impressive to say the least:

Negative side:

±4 fps more from a Titan to a single 2080ti in very high settings......

Positive side:

2 2080ti however show a substantial increase in very high settings....

Conclusion:

I will not buy a 2080ti in the future. There is no point.

Cheers,

Pierre

[Jetman67 506]

Hi rob

Is the 2080 rtx ti much quicker than the 1080 gtx ti in games and p3d

 

Wayne 

[Ray Proudfoot 8,919]

Many thanks for that extensive testing Rob. Two 2010Ti cards is a nonstarter for me.

I don’t suppose you ran the same test with a 1080Ti did you? The comparison would be helpful.

[Ted Striker 338]

Similar question to Wayne's. How does the TitanXP compare to the 1080ti in P3Dv4? Just rough approximation is fine.

Ted

[LB777 100]

Many thanks for running through all these tests!

The numbers really are depressing though...

[adyfoot 19]

19 minutes ago, Ray Proudfoot said:

Many thanks for that extensive testing Rob. Two 2010Ti cards is a nonstarter for me.

I don’t suppose you ran the same test with a 1080Ti did you? The comparison would be helpful.

Rob's Titan XP is pretty much equivalent to a 1080Ti - it just an extra 1GB of RAM.

[Guest]

I do have a 1080Ti, but haven't tested against my 2080Ti's ... I did test a 1080Ti some time ago against my TitanXP, the TitanXP was about 5% higher FPS on average.  Keep in mind my TitanXP is water cooled, the 2080Ti's are not (I have two cooling blocks coming this week that I'll install to get those 2080Ti temps down).

You folks will come to your own conclusions.  Just providing data for how these cards are working for P3D V4.3 as of "now".  As most are aware, to really see these cards shine on the FPS front requires support for DLSS and ray tracing extensions to replace dynamic lights.  The single card result is a little lower than I expected for existing app not setup to use the GPU's new features ... and you can clearly see if one doesn't increase graphics settings/options (load) then 2X 2080Ti actually perform worse than a single TitanXP (most likely due to overhead and NVLink restrictions).

I was impressed with how 2X 2080Ti worked with my very high graphics settings, everything checked and maxed out along with TEXTURE_SIZE_EXP=10 ... and look at those memory numbers, went above 11GB in a few cases.  I settled on graphics that were somewhere between HIGH and VERY HIGH for my day to day flights allowing FPS headroom for the addition of complex aircraft and HiFi weather and AI traffic.  I'm happy with both cards, but I honestly can't suggest there is "value" here ... but "value" is really unique to individuals and isn't something that can be "subscribed" to.  However, with that said, if DLSS and/or ray tracing does get implemented in a sim in the near future that could change my mind on 2080Ti "value".

Drivers I'm using are 416.16.

I'll test AF2 and XP11 next, but I don't have a "formal" baseline setup for those sims and neither AF2 or XP11 support NVLink/SLI so I can't compare that aspect.

Cheers, Rob.

EDIT: when I have more time I'll post screenshots of the actual graphics settings pages from P3D for each "scenario" used (Medium, High, Very High).

[VeryBumpy 272]

Such a horribly inefficient running game. It takes two 2080Ti's to gain a meaningful improvement.

[YoYo 374]

1x 1080Ti vs 1x 2080Ti pls :).

 

[Guest]

It would be interesting to see what the number would be like for "Very HIgh" setting for a 1080ti v's 2 x 1080ti in SLI given that % increase of FPS for 2080ti v's 2 x 2080ti in SLI. I'd be inclined to add another 1080ti if I had one(LOL) rather than get a 2080ti. It would be interesting to see that Comparison again with NVLink on the 2 2080ti's.

Great apples to apples comparison Rob.

It's nice to have a hardware enthusiast in the community with deep pockets who can give the rest of us an unbiased assessment of the latest relevant hardware.

Personally, I don;t think these cards are worth it for P3D right now. They might come into there own after the release of V5. who knows.

My opinion is that its smart to stick with your 1080ti's and if you are on a 980ti take advantage of the used 1080ti's that will be on the market over the next year. By the time the 2080ti is worth having for P3D or X-Plane the 2180ti will be on the market and I'll pickup a used 2080ti or 2.

Edited October 9, 2018 by Guest

[Guest]

I guess I should include some pictures of the cards.  I did lose one PCIe slot that I needed for a 2nd USB 2.0 card (for my GoFlight gear) which was a bummer.  I have a couple of EK 2080Ti full coverage water blocks that will arrive tomorrow so I'm hoping that will free up some PCIe slots so I can put my USB 2.0 card back in (I have two). 

As you can see from the pic, one of my USB 2.0 PCIe cards is partially blocking the fan.

 

Cheers, Rob.

[HiFlyer 9,343]

I was a bit concerned about some of your temps, especially considering you are on water cooling. I would be worried about Kentucky fried CPU!

[YMMB 234]

Thanks Rob for taking the time and financial outlay to provide us great feedback and some things to ponder on our own future upgrades.

Your results probably suggest in my own case that I'll stick with my current water-cooled I7700K (stock) and 1080Ti and maybe overclock  the cpu conservatively just for some interest.

I think I'm comfortable waiting for a major software update of P3D before I venture into a PC accessories store. 

I guess my real wish is for a future iteration of P3D to take on the latest and greatest graphics engine such that smoothness and fps isn't something I even bother to check anymore.

 

 

[Ted Striker 338]

Thanks for taking the time to collect and post your test findings Rob. I am convinced now that a single 2080ti will provide little advantage currently over my 1080ti. Things may change when P3D is upgraded to use the new features of the 2080 and/or I upgrade my overclocked 3770k. Being a worthy candidate for a CPU upgrade I am looking forward to seeing your 9900k results.

Ted