Super Power My Machine

[Noel 4,487]

1 minute ago, Republic3D said:

I'm not in 4K, I'm in 3440x1440. 

He says he's on 1080p but he wants to increase this. If he wants to go to 1440p or 4K he needs more VRAM. That's when the GPU upgrade becomes more relevant again. Which falls into your argument. 

I just politely disagree about CPU not helping in dense cities, it definately helped for me. 

It's always going to be a case of matching all the main components lest you be held back by any one.  This is why it's useful to monitor real time load w/ something like MSI Afterburner/RTSS.  For me, w/ 9900K at 5Ghz all cores, never is my CPU the bottleneck, it's always the 2070 Super.  I really do need at least a 3080, but would rather have one w/ 12Gb of VRAM at least.  A 3090 will be GPU overkill, at least certainly VRAM overkill.

[manageablebits 11]

I'm on a Ryzen 3600 and 2060 SUPER.  What made it work good for me (not tried the big passenger planes though) was realising I had a gsync 144 monitor in another room which works much better than vsync and fps limits.

[The Moose 1,177]

1 hour ago, Alpine Scenery said:

I am guessing your not flying in pure ultra then, you are using a tweaked subset of ultra, technically Ultra means pure Ultra.
The reason I note that is because that is what reviewers and testers go by (unaltered Ultra).

NY is not 30 FPS on a 2060 Super with a Ryzen 5 3600 in Ultra 1080p, not the denser parts of the cities.
It is around 24-28 FPS if you are lucky when flying low, and it can even dip below 20 in short bursts in certain parts.

Not "pure" Ultra. correct.

Vsync on/30fps/TAA/Render scale 100.   The only settings lowered are shadow maps(1024) Terrain shadows(512) Motion Blur (off) Lense flare(off).  Everything else is the highest possible.

     Just flew around Manhattan (PG on) at building height, live weather, live traffic, multiplayer traffic on.   Even switched to the Stormy preset to see if it made a difference.   Locked at 30fps, never dipped under it for a second.   (JustFlight Arrow with 750 mod)

https://www.imagebam.com/view/ME4F8G

Edited May 9, 2021 by The_Moose

[Alpine Scenery 1,609]

That's a relatively high FPS and slow plane, also depends on NVIDIA settings. It's hard to compare Apples to Apples. Yes, it's possible to get 30fps most of the time with certain settings, but that's not technically ultra. Your shadow maps  and terrain shadows are cut in half from Ultra.

[filou 2,741]

1440P / GTX1080 / 24GB DDR3

if i follow the logic of some i should have 78 fps whith the RTX 3080...

[Alpine Scenery 1,609]

16 minutes ago, filou said:

1440P / GTX1080 / 24GB DDR3

if i follow the logic of some i should have 78 fps whith the RTX 3080...

 

Going from a GTX 1080 to an RTX 3080 would be huge!... Probably not 78, probably low 50's I'm guessing. I think the below is nearly true Ultra, so you should easily get +5 or more over that. Again though, people are not comparing FPS the same way, there are too many variables here. There is no real way to just say FPS, have to use same tools and methods.

Edited May 9, 2021 by Alpine Scenery

[filou 2,741]

3 minutes ago, Alpine Scenery said:

Not sure at 1440p, at 1080p the GPU makes more difference, but either way, going from a GTX 1080 to an RTX 3080 would be huge!...

Probably not 78, probably low 50's I'm guessing.

Yes and without speaking with a new CPU on top of that ...If it could be the case 50 fps in the big airports it would be paradise!

[Alpine Scenery 1,609]

If you notice from that chart, the difference between the 2080 and 3080 is only 6 FPS, but the difference between the 2070(2060s) to 3800 is 13 FPS. That's why it all depends on what card you start with to whether or not you should upgrade CPU or GPU, but usually it's GPU is the thing to upgrade.

In my own findings, it was slightly more pronounced, so a 2060super to a 2080 was actually around 15fps difference where I was testing, I wasn't able to test a 3080 unfortunately, so assuming in my test a 3080 would have been at least 18-20fps higher than a 2060super. That was using my tweaked (sort of Ultra but not really) settings, I would call it halfway between high and ultra, and yes at that time we were testing 1440p actually. We tested both.

Whereas I only eeked out 3-4 FPS out of a $600 CPU upgrade, so yah, it's not even close, for most people, go with the GPU unless you have at least a 2080s/3070/3080 already, then it becomes a tougher call unless you have a really poor CPU.

 

Edited May 9, 2021 by Alpine Scenery

[Alpine Scenery 1,609]

You can get 10 or so FPS increase on the peak FPS from a CPU upgrade in some cases with unoptimized settings, from say a 3600 to 5800,

BUT...

Where it actually matters when it dips below 30-40, the difference is +3-5 fps even with a 2080 from a 3600 to 5800 AMD upgrade, and it's only +2-3 fps with a 2060 on a 5800. Again though, it depends on setting and drivers so hard to compare, but that is ballpark anyhow. The worst case and LOW FPS is the number we care about, best case doesn't really matter much, as it is the lowest number that you will feel the most, rather than the highest.

Edited May 9, 2021 by Alpine Scenery

[filou 2,741]

I think the new modern processor that will replace my I7 4790k will easily give me 10 fps more.

[Alpine Scenery 1,609]

4790k is pretty old CPU, I am not sure about that one. On the top end, you might get +20 FPS, but where it matters (lowest FPS), probably +5 to +10. Increasing highest FPS numbers doesn't do much, it's the lowest FPS numbers you really want to go up more (usually). People confuse it, plus MSFS isn't like other games, 25 FPS can be fine and 35 FPS might be stuttering at times, so it's hard to compare, depends on overall smoothness as well.

Edited May 9, 2021 by Alpine Scenery

[The Moose 1,177]

28 minutes ago, Alpine Scenery said:

Again though, people are not comparing FPS the same way, there are too many variables here. There is no real way to just say FPS, have to use same tools and methods

Absolutely correct, which is why you making statements like

"NY is not 30 FPS on a 2060 Super with a Ryzen 5 3600 in Ultra 1080p, not the denser parts of the cities.
It is around 24-28 FPS if you are lucky when flying low, and it can even dip below 20 in short bursts in certain parts."     without any context (plane/weather etc)   is absolutely pointless.

Anyway... just re-ran the Manhattan flight with pure Ultra settings (and NV control panel default settings untouched) in the Arrow, The G-91 and the TBM between 1500ft and 600ft All of which resulted in 30fps solid.

I'm certainly not trying to get into a petty argument with you.  The whole point was unless the OP is using a very heavy on the FPS plane he absolutely should be getting better performance than he currently is.  The 2060 Super is more than capable of 30FPS solid with pure Ultra settings in very demanding scenarios @1080p.  Until he posts again with his settings and a specific scenario though we cant even begin to help.

[filou 2,741]

1 hour ago, Alpine Scenery said:

4790k is pretty old CPU, I am not sure about that one. On the top end, you might get +20 FPS, but where it matters (lowest FPS), probably +5 to +10. Increasing highest FPS numbers doesn't do much, it's the lowest FPS numbers you really want to go up more (usually). People confuse it, plus MSFS isn't like other games, 25 FPS can be fine and 35 FPS might be stuttering at times, so it's hard to compare, depends on overall smoothness as well.

I completely agree.

[Noel 4,487]

1 hour ago, Alpine Scenery said:

 and 35 FPS might be stuttering at times...

Not if you vsync to 30Hz , at least not from the hardware's deficiencies at that moment in time.

[Seth2021 291]

Some screenshots of EGLC London city area / ultra settings / 1440p  with 3090 Asus rog OC + 8700k ( delidded) 5.2ghz. 

Just upgraded on 11900k ( adaptive boost )  + 64g ram 3600 CL16 t i have 10 to 15 fps more,