Went from a 4080 to a 4090….results

User Support Hub Virtual Reality (VR)
1

This is for those that are thinking about going from a 4080 to a 4090 to get that double performance for a few hundred dollars more. That was me too and I got lucky and thanks to a twitter notification I caught a Best Buy drop and was able to score!

Updated: I originally had this for the short version: So here’s the short version: solely looking at FPS point of view, I got about a 10% bump over the 4080. Yep…that’s it!

BUT…as mentioned elsewhere some people talked about disabling hyper threading. I did (talked about just below this paragraph) and I just did my first real non-testing flight and my results say a good solid 20% boost over the 4080. I flew from Vegas to LAX and there was a ton of cloud cover and full on pouring rain at LAX and in the PMDG 737 @ 26K feet it stayed pretty steady at 55FPS. The 4080 based on my experience would be mid upper 40’s so I say 20% is fair. This is at nearly full ultra as shown below in my settings. Any disappointment I MAY have had is gone. This thing rocks. On the ground parked in the pouring rain at LAX I was hovering around 40FPS. If I looked around real quick it might drop to 27 or so for just a second or two and then back up.

Update a few days later: I disabled hyper threading in the bios and got about ac10% additional bump in FPS. Did this based on multiple threads from people talking about it. It appears to have been effective.

With that said, I should note that I got a whopping 40%-ish bump going from a 3080Ti to a 4080. THAT was a heck yeah kinda experience. I was hoping on the same type of experience going from a 4080 to a 4090 but it just didn’t happen. What I did get was a little more CPU bound notification.

Also, I should note that while the FPS was not dramatic, the smoothness was noticeable. Less stuttering and “catch up” like where you might get some frames that lag and then catch up real quick. And in some cases better than a 20% FPS improvement but just not the same “wow” I got going from the 3080Ti to the 4080.

With the price difference of only $400 between the 4080 and the 4090, the 4090 is still the proper choice as I do believe they will improve the sim to where the 4090 will shine even more and same for when I bump my CPU in the near future.

Lastly I do run higher settings than many might. This is with TLOD and OLOD both at 200 and most of the other settings at Ultra with some at High but I am on a quest of the ultimate VR experience.

Frames at JFK on the ground consistent 30FPS average, same for KLAX. In Paris 35+ which was a pretty good bump in itself. In the air over busy areas at 1000ft upper 30’s to low 40’s. Numbers only get stronger from there as altitude increases peaking at around 60FPS. Most at altitude is around 50FPS so pretty good but I was hoping for 40FPS on the ground at JFK or LAX.

Machines is
Win 11 latest version
I9-11900k
64GB ram
3TB of m.2 hard drive space
HP G2 headset
Nvidia DLSS with Quality setting. Pretty big difference with DLSS on the 4080, less of a difference using it with the 4090 and it’s the DLSS that is taxing the CPU. With no Nvidia DLSS on it’s a few FPS less but no CPU bound messages.

These are the settings I am using:

image

6 Likes
2

What frequency (e.g. CPU@4.5Ghz) are you boosting all cores to? Is it possible that you are CPU limited with the 4090?

2 Likes
3

What is your OXR render scale?

4

The CPU is the limiting factor on most systems.

2 Likes
5

Mainthread limited is probably costing you fps. Other than LOD and traffic you must raise everything to ultra including in NVCP and even then you should start looking at bumping renderscale and resolutions as far as VR lets you. In other words your gpu work needs to work as hard as you can make it.

6

You might find this video interesting:

2 Likes
7

I make no adjustments to my render scaling. It’s at 100.

8

THAT will be a statement many people might not agree with considering the price difference. There are still many people out there for who even a 3080TI is a very significant investment or who are waiting that the 3070TI will fall to a price of 400,-$

Thanks very much for your results. For me it’s clear indication that if I decide on a new GPU when the Ryzen 7800X3d comes out, I would certainly not consider the 4090.

9

Just for comparison with a 5800X3D and 4080, I tried your approx settings (OXR 100, DLSS Quality, VR Ultra Preset) at KLAX in the PMDG 737 with lots of MP traffic to compare with your 30 FPS you say you are getting there and got the following result:

image
image1301×723 137 KB

image
image1847×1851 240 KB

At my normal settings (OXR 180, DLSS Quality, VR mix of ultra/high/medium, OXRTK FFR and turbo mode) in the same aircraft and location, I get the following, which looks significantly sharper, a few FPS less and neglible less detail than all ultra settings.

image
image1920×1881 280 KB

My suggestion to you is to either lower the CPU affected settings (TLOD and to a lesser extent OLOD) or, if you really must have this level of detail, raise your OXR render resolution until you just start losing FPS (which with a 4090 should be around 200) and enjoy the significantly sharper DLSS image that results.

1 Like
10

I added a graphic of my exact settings in my original post.

Also, for comparison sake, there was a guy on YouTube talking about massive frame rates. Fortunately he put his settings in the vid too. With the settings below there’s about a 20 FPS gain even in the busiest of areas. At JFK with these settings I had about 50FPS. I buzzed Manhattan and had 15+ FPS gain from my ultra settings in post 1 at very low altitude. In addition, flying at about 8000ft just north of KCLT with clear sky’s it knocked off a whopping 80FPS peak with these settings. So if someone is ok with the medium level settings, these numbers were impressive. I did NOT do these same tests with my 4080 though for a comparison.

image

2 Likes
11

Just adding in here that based on what I read in a few posts, I disabled hyper threading and all indications are that I gained at least another 10% FPS. Only problem is I didn’t do that when I had the 4080 for comparison. So going from runway 31 at KJFK clear skies at noon, between the airport and downtown was doing 45+ FPS at about 1500ft in a single engine caravan with the mostly ultra settings in post 1. I then went into downtown at building level and swept through Central Park at about 300ft. Downtown and buzzing Central Park was about 30-35 FPS and when I turned around and came back through was 35-40 and the trek back to JFK was 45-50 FPS. I wish I would have done the no hyper threading before I got rid of the 4080 but it is what it is. KLAX with the PMD 737 improved to about 35-40 FPS sitting on the ground near a terminal looking around. Consistent with the claimed 10% bump with hyper threading turned off.

2 Likes
12

Just for your future shopping… with Nvidia, the first two digits are the ones that indicates you’re moving between GPU generations, so that massive bump from 30 to 40 is expected.

the last two digits are kind of ‘sub-variants’ of the generation and generally the gap between them will be much less pronounced (as you discovered). Anything from a xx70 and up is going to be a pretty solid gaming card, with the xx90 being the king. A 5 to 10% bump is pretty much par for the course.

Anything xx60 and below is only really suitable for low-end gaming, or general desktop computing. In fact going from say a 2080 to a 3050 might actually be the same or worse even though its on the newer generation (haven’t tested, just pointing out that the numbers below 70 go down pretty quickly in performance).

You can only really compare apples to apples if you’re comparing for instance a 2080 to a 3080, etc… Comparing a 2080 to a 3070 gets a little foggy… same as you go up in the the 40xx series.

13

Thanks for sharing your results and updates on the switch.
I have a 12700K with a 4080 oc, Dx12, hags on, have my Pcore and Ecores dialed in and now I am wondering if I should turn off hyperthreading. hmm… tweaking continues lol.

1 Like
14

I gotta say I just had to update my first thread. Since I disabled hyper threading I did a real flight between Vegas and LAX (vs test scenarios) with tons of cloud cover and pouring rain at LAX and this thing flat out rocks. 55FPS at 26k feet. I’m not totally sure of how much difference disabling hyper threading did but there’s no denying how phenomenal this flight was. I was not this impressed with it prior to disabling hyper threading but I am now. AND…I didn’t get any CPU bound nags that I can think of. Your market may vary but I’m pretty happy.

15

60 isn’t for low end gaming at least for a recent generation. It’s aimed for 1080p Ultra settings and 1440p entry level performance.
70 1440p gaming/ ultrawide
80 Super ultrawide/ 4k

The 50series are the low end cards aimed for small form factor builds and lower power consumption.

Anyway I’m well off topic. Op will probably get a decent boost going to a better cpu.

16

:wink:
Its getting a little nerdy in here…

70 or better for gaming.
60 or less for other stuff.

The rest is getting a little more granular than I think the average person really thinks about.

1 Like
17

Add to that list - 90 series high res VR

1 Like
18

I am undecided to buy either 4080 or 4090 for my 2 headsets (Quest 2 and Pico 4).
If I understand you correctly the RTX gives you a good 10% increase only?
Could it be that your 11900K is a bottleneck for the 4090 and the reason why it is not much more than 10%?
Could you have a look at how high the usage of the 4090 is in percent? Thank you.
A CPU like 12700K/12900K or 13700K/13900K could find some more performance.
Otherwise I tend to buy the 4080 because 400$ more for 10% is not enough for me personnally.

19

Tested Hyper Threading on and off on 12700K, 4080, Win 11, Hags on
NYC discovery flight, not much difference. If I was hard press to define fps it would be 1-2 fps max with it off.
Perhaps AMD folks are seeing better it being off.

20

I have a 5800X3D and can’t tell the difference with a 4090 with SMT on or off, in fact with DX12 I seem to get a better spread of workload with it on. So I leave it on for when I need to use focus stacking software for my macro photography apps.

2 Likes