Hi all, as I read so many posts about the performance gains in v5, I of course was looking forward to see how it performs on my rig. It was not an disappointment, but my personal experience was not so euphoric although I am very happy with the performance as well. Basically because I ignored FPS numbers until I got most of my addons installed again, I was surprised that obiously some people did (again) forget the fact that a brand new installation usually performs better and that at the beginning, most early adopters just installed one or two addons.

As a matter of fact, I did take two screenshots from a very heavy scenario, EGLL, as shown below. I know, I could have used the slew function to put the R22 in the exact same position, but it anyway shows what I observe on my rig: yes, the performance of v5 is better, but no, the FPS increase in the fully stuffed simulator is far not that huge as some of the reports here were suggesting (at least not for my build). 25FPS in v5 vs. 21FPS in v4.5, or translated, v5 provides me with 20% more FPS. Don't get me wrong, I take those 20% in any case and I am very happy with it, but it is far from "doubling" as some reported.

Of note, below screenshots show a comparison as close as possible, means: I used exactly the same settings in v4.5 as I did in v5, except the "enhanced atmospherics". And in both shots the sim is equipped with the AS EGLL scenery, ORBX OpenLC Europe and my heavily modified UT2 traffic. I will for sure also repeat the test with the classic sky depiction in v5, might provide another 2-3 FPS.

Now what I really wonder is: could this not so prominent performance increase be related to the CPUs we use? Does v5 profit even more from more cores than v4.5? Do RTX cards profit more from DX12 than earlier generations? It might be, because most of the reports about high performance gains come from users with CPUs providing 8 cores or even more. Or the opposite, from users having still their 4 core CPU in duty (a known fact that DX12 provides more performance "gain" on heavily CPU limited systems). I do not know, my 6 core CPU and my 1080Ti seem to profit not that much from neither DX12 or better core usage, although I am super happy already with this 20% increase. As I currently use the same AM as I did in v4.5 (AM=1365), I will run for sure also some tests without AM to see if my 6 Core CPU will profit from HT in regard of pure FPS numbers. I already know that using HT results in more stutters (again) in my case.

I will for sure run some additional tests around EGLL, as the performance in v5 is still somehow puzzling to me. In EHAM from FT with TE Netherlands active for example, I get almost constant 30FPS with some dips down to 25FPS, something I never observed in v4.5, where FPS were mostly around 20. It seems to me that the performance increase observed in v5 is greatly depending on the addons used (not really surprising) and that some addons are equally taxiing as they were in v4.5, some less.

What are your observations? Do you see such huge performance increase as others or are you closer to what I observe?

BTW: I did run EGLL once with FPS set to unlimited and yes, the FPS loss by limiting them is still rather big.

to me its not the number of FPS you see mentionned on the screen, I don't look at that anymore , its the the feel smoothness of the V5 sim that is astonishing.

I did a very simple test :

plain vanilla P3D V4.5 install (no addons) with a default aircraft and with very high settings , takeoff from LSGG in VFR conditions and did some agressive turns over the city (not realistic I concur). Although I had very high FPS , there were always some slight stutters when looking at different views.

Now i did exactly the same test in P3D v5 (same settings) and it's 100% smooth, no stutters at all .

So based on the above I can say that I have better performance with P3D v5 and it will certainly be maintained while installing further add ons

It's at simple as that really.

If it was this simple. I managed to have my v4.5 set up that it is extremely smooth. So far, I do not see any difference in the perception of smoothness in v5. In contrast, while I obviously managed to have my v4.5 set up so perfect for my rig (finely balanced tweaks and settings), I am still not there in v5. Means: there are considerably more slight stutters in v5 than I have in v4.5.

But generally you are absolutely right, the aim is smoothness, not FPS numbers, although they are partially connected.

I can confirm, that the increase of performance in case of FPS is not that huge. But for my personal feeling the overall appearance is smoother. Of this depends on soooo many different things for every individual rig, that you can´t say in general this or that setting performing better.

In case of hardware there are some general aspects. Yes, a RTX 20something should get more advantage out of DX12, because the GPU architecture is developed for it. The GTX 10something cards have the former pascal architecture and where developed when DX11 was coming or already used. On the CPU side i can see that, especially with HT off, it´s an advantage if you have a CPU which has just more real core like mine I9 which has 8 physical cores whose all overclocked to 5 gig´s. But even this doesn´t mean you get such a huge performance. I. e. my former CPU was an i7 with 6 cores and since i have the i9 there is a little bit more performance but it´s in an area of may 4,5 to 10 FPS. 

FPS is not a problem here for me. The problem is stutters over large cities. Only happens over large cities too. (Especially London). 
Also popping scenery. It’s as if the sim can’t keep up with my system! It’s trying to load things too slowly. 

Definitely saw a performance increase. However, I didn't test for peak numbers, as I have no use for 60FPS+, I just tested to see how well it felt and maintained my externally locked 30FPS in different scenarios. Was able to fly over a lot of high density cities, with sliders turned up an additional notch or two and see no performance decrease, no autogen load stutters and no camera stutter caused by cores overloading causing various camera applications to get bogged down too. That to me is huge.

However, my go-to stress test of using OrbX's SoCal, leaving KLAX in a fast plane to clog up the cores even faster, was usually where I'd benchmark my sim. Done it so many times I always knew where I'd get the first stutter as the autogen loaded, where to look to tank the FPS (hello cranes), where the FPS would creep down below 20 etc. first run was fairly unimpressive, with all cores wildly flailing and getting pinned, usual stutters and sub-20 FPS. Then I decided to mask off the logical cores on my 8700k, create an AM based HT off scenario and suddenly the performance just sky-rocketed. Was now able to take off, maintain 30FPS during this stress run and saw a single autogen load stutter that was fairly short. 

That, having used the ESP platform for many years now, was a huge step forward. If I'm able to maintain 30 there, I've now received a huge overhead going forward throughout the rest of the sim.

Edited April 30, 20206 yr by Sethos

This is a good v5-v4.5 fps test

Conclusion 20%-30% gain

 

I think DX12 makes NV Link GPU`s more of an option than SLI was.

https://www.youtube.com/watch?v=84OkcOYmXOk&t=521s

Check out the DX12 snipper elite bench.

I'm not surprised that 20% is all the extra gains you see at the expense of an unstable DX12 sim. This happened to me when I tested Vulkan in XP11. Both are still a work in progress. With only a few add-ons and plugins fps were rather good, but when I tested a fully loaded XP11 Vulkan I got about a 20% increase but at the expense of longer loading times and blurry textures.

The best way to get better FPS is to upgrade hardware. I saw performance increases when I went from vRaptor to SSD &  gtx970 to a 1070. I have 16gig of RAM and I intent to upgrade to 32 gigs and if I see a 1080ti or 208ti at a good price I'll upgrade my video card.

Those two changes will increase fps and get me ready for MSFS 2020.

I am amused.  Dismissal of 20% gain.  Honestly... very amusing.

"Do you see such huge performance increase as others"

 

yes.

I think the biggest thing for me is finally being able to fly into larger cities. This is where I'm seeing the biggest performance gain. In the past any decent sized city would just crush my framerates and make it a stuttering mess. I have sliders slightly higher than v4 and was just able to shoot a dusk approach into KLAX with ENVTEX/ENVSHADE/ORBX and hold a very smooth 40-50FPS. This is a night and day difference from any experience I've ever had. For me this opens up so many new areas that I've always avoided flying to.