What GPU to buy for VR?

[chiliwili69]

New VR devices are requiring more powerful GPUs and I am currently analyzing what are the options for a new GPU.

 

I currently have a Quest3 and a RTX3080. That card is more than enough for 150% SteamVR supersampling ( it means rendering 18Million pixels) at 72Hz. So I am always at 72fps unless my CPU (7800X3D) become the bottleneck in AI dense scenarios.

 

My next VR device to try (I will see if it will replace the Quest3) is the BigScreenBeyond2, which should arrive in July.

That device has a panel of 2560x2560 and at 100% SteamVR it renders 25.4 Million pixels at 75Hz.

 

I have been looking at benchmarks for DirectX11, which is the one used by IL-2.

I have been always with NVIDIA because the past problems with AMD cards, but I am interested to know the performance of the AMD 7900XTX with IL-2 in monitor and VR.

 

My current Mobo is PCIE 4.0.

 

The candidates are in this table:

 

 

Update (1-Sept-2025) : After my fiasco with the RX 7900XTX the alternatives are:

 

Edited September 1 by chiliwili69

[Goldwrench]

I’m not sure about the 7900XT but every other GPU on the your table should have no problem on the BSB2 with Il-2 since it is not a GPU-intensive game. I think you should base your decision on any other applications/games/sims that you use in VR, most of which are probably more challenging for the GPU than Il-2. I run the 7900XTX at 100% super sampling and am locked at 75hz unless in some massive campaign battle where the CPU frametimes can get longer than 13.3ms. Upgrading from the 7950X3D to 9950X3D actually helped here more than I expected.

[chiliwili69]

On 4/21/2025 at 7:43 AM, shipdriver9314 said:

I think you should base your decision on any other applications/games/sims that you use in VR, most of which are probably more challenging for the GPU than Il-2.

Thanks. In my case, my only interest is just for IL-2 in VR, my son also use the PC to play their games in a 4K monitor so he will be plenty with any of those.

 

I saw in the other post the good performance of your 7900XTX with different %supersampling and this is exactly what was making me to consider the AMD cards.

 

I am only a bit confuse about the performance of those cards with the public benchmarks for DX11 shown in the table above. And also the IL-2 4K benchmark we did with a 7900XT giving only 92 fps.

[Goldwrench]

Without having experience with the 7900XT, I would guess that it is borderline for Il-2 in VR at maintaining below 13.3ms frame times on a BSB2 (at least with moderately high settings- I use High preset and mostly high settings like clouds). Il-2 Korea will undoubtably be more GPU demanding than Great Battles so that would be something to consider as well. Also many game titles that your son may play are much more GPU demanding in 4K than Il-2 is in VR (some, like Black Myth Wukong, seem more demanding in 1080p!). The best time to buy AMD 7000-series cards was during the 2024-25 holiday season as you could get them below MSRP (I paid US$815 for my 7900XTX in Dec 2024), but I'm not sure they are as good of a value today, but then again, there really are no good values now for GPUs. 

[Dutch2]

Keep in your mind if you use RaYtracing or RTr is being introduced into IL-Korea and Jasons Pacific flysim, then the cheapest €750,- 9070XT is the much  better deal then the 7900XTX  cheapest €900,-

https://geizhals.de/?cat=gra16_512&xf=9809_05+14+-+RX+7900+XTX~9809_05+15+-+RX+9070+XT&asuch=&bpmin=&bpmax=&v=e&hloc=at&hloc=de&hloc=pl&hloc=uk&hloc=eu&plz=&dist=&mail=&sort=-r&bl1_id=30&togglecountry=set

 

 

 

Edited April 27 by Dutch2

[Goldwrench]

On 4/27/2025 at 11:57 AM, Dutch2 said:

Keep in your mind if you use RaYtracing or RTr is being introduced into IL-Korea and Jasons Pacific flysim, then the cheapest €750,- 9070XT is the much  better deal then the 7900XTX  cheapest €900,

The 9070XT is a great card but it doesn’t have the raw rasterization capability of the 7900XTX/4080 especially for higher -end VR headset. I think there’s a chance that the 9080XT can barely handle 100% supersampling on a BSB2 but if it does, you’d need all the in-game settings to be at the low end. What kind of Raytracing is implemented in future titles makes a big difference too. On a flat screen, MSFS 2024 ray-traced shadows drop fps by only 2 or 3 on my 7900XTX, but those shadows are disabled for VR anyway, whereas something like full path tracing in Black Myth Wukong will bring a 7900XTX to its knees. Future titles that support FSR4 (huge improvement over FSR 3) will benefit from the 9070XT (probably to a bigger degree than raytracing) as well although some of the FSR4 functionality will eventually make its way to the 7000-series cards. I bought a 9070XT for my daughter’s birthday and it was US$85 more than what I paid for my 7900XTX!

[Aapje]

All the pricing went up, although it is now going down again.

 

And the 9080 XT doesn't exist and almost certainly will not exist. And they never promised FSR4 support for the 7000-series. Just that they would look into it.

[Dutch2]

23 hours ago, shipdriver9314 said:

The 9070XT is a great card but it doesn’t have the raw rasterization capability of the 7900XTX/4080 especially for higher -end VR headset. I think there’s a chance that the 9080XT can barely handle 100% supersampling on a BSB2 but if it does, you’d need all the in-game settings to be at the low end. What kind of Raytracing is implemented in future titles makes a big difference too. On a flat screen, MSFS 2024 ray-traced shadows drop fps by only 2 or 3 on my 7900XTX, but those shadows are disabled for VR anyway, whereas something like full path tracing in Black Myth Wukong will bring a 7900XTX to its knees. Future titles that support FSR4 (huge improvement over FSR 3) will benefit from the 9070XT (probably to a bigger degree than raytracing) as well although some of the FSR4 functionality will eventually make its way to the 7000-series cards. I bought a 9070XT for my daughter’s birthday and it was US$85 more than what I paid for my 7900XTX!

All depends which country sure, but paying $85,- less or even the same for an 7900XTX than indeed a 9070XT would be foolish. But if it does not matter getting ca €200,- lesser, then why not eventually go for €200,- more to the RTX5080. Watching closely I do see them for around 1100,- 

 if only playing this IL2 in VR, a game thats clearly  is an 2D game transferred into VR mode, (l do own the corrugated ju52 btw,), just ask yourself is it not better to wait for Korea, Jason Williams pacific.  I suppose with the new or modified game engine the clumsy 3D designs would be gone and you will not go back. 

Edited April 30 by Dutch2

[Goldwrench]

20 hours ago, Dutch2 said:

All depends which country sure, but paying $85,- less or even the same for an 7900XTX than indeed a 9070XT would be foolish. But if it does not matter getting ca €200,- lesser, then why not eventually go for €200,- more to the RTX5080. Watching closely I do see them for around 1100,- 

 if only playing this IL2 in VR, a game thats clearly  is an 2D game transferred into VR mode, (l do own the corrugated ju52 btw,), just ask yourself is it not better to wait for Korea, Jason Williams pacific.  I suppose with the new or modified game engine the clumsy 3D designs would be gone and you will not go back. 

Foolish? Not if the 7900XTX beats the 9070XT in every VR benchmark- which it does. I would hesitate to recommend the 9070XT for Il-2 VR for a BSB2 if we don’t know whether it can sustain good frame times at 3560x3560 per eye. One can talk about future games all day but 1) they will be even more challenging for the GPU than an old title like Il-2 GB and 2) the OP may actually want to play in the ETO. Regarding the 5080, the OP already has that on their list (as long as it has all of its ROPS and the drivers actually work). 

[Dutch2]

36 minutes ago, Goldwrench said:

Foolish? Not if the 7900XTX beats the 9070XT in every VR benchmark- which it does. I would hesitate to recommend the 9070XT for Il-2 VR for a BSB2 if we don’t know whether it can sustain good frame times at 3560x3560 per eye. One can talk about future games all day but 1) they will be even more challenging for the GPU than an old title like Il-2 GB and 2) the OP may actually want to play in the ETO. Regarding the 5080, the OP already has that on their list (as long as it has all of its ROPS and the drivers actually work). 

Foolish is when buying the 9070XT for around the same price as the better performer the 7900XTX, clear on that point now I do hope. 

The rops was a failure from the first 5000 series and as being an owner of the 5070ti, I can tell you the drivers do work in IL2-VR, nothing to worry about.  
 

Edited April 30 by Dutch2

[Goldwrench]

7 hours ago, Dutch2 said:

Foolish is when buying the 9070XT for around the same price as the better performer the 7900XTX, clear on that point now I do hope. 

The rops was a failure from the first 5000 series and as being an owner of the 5070ti, I can tell you the drivers do work in IL2-VR, nothing to worry about.  
 

Feel free to provide info on how the 9070XT is a better performer than a 7900XTX in VR in Il-2.  

[Dutch2]

4 hours ago, Goldwrench said:

Feel free to provide info on how the 9070XT is a better performer than a 7900XTX in VR in Il-2.  

I did not write that down about better performance, only for €200,- lesser it could be also be an very interesting Vcard to consider. I can not watch into his real budget. 
Thats all noting more. 
For the rest I do not know on what you are pushing for. What’s next, a complete analysis of my writing here, I’m only trying without Google translator typing my English here. I hope you do not mind, I stop here as I’m highjacking this topic from the TS. 
 For the TS  based at your list I would go for the better RTX5080/5090 maybe the 4090, as keeping one eye to (I suspect) the very high performance eating VR in Korea and Jason Williams Pacific. 
Have all a nice day.  🫡

Edited May 1 by Dutch2

[=SqSq=SignorMagnifico]

I don’t know if this helps, but I upgraded from an RTX 3080 Ti to an RTX 5080. I made a video that compares how IL-2 runs on both. Maybe this can give you an idea of how much headroom you would have for VR with each setup. 

 

[chiliwili69]

On 5/2/2025 at 6:28 PM, =SqSq=SignorMagnifico said:

I don’t know if this helps

 

Hey!, Thank you for this video! With many data and many scenarios/planes. The gain you got is quite significant overall is around 30-40% with monitor 1440p and all setting maxed-out including MSAAx8 which requires a big extra GPU load.

 

It is interesting to see that the required GPU power of the 5080 is lower than the 3080Ti in all the test you did. Around one third less (Except in the last D-Day where it was even less)

On the other hand, the total UPS power is a bit higher since your CPU is rendering more fps.

 

Also, in the last D-Day test (with more AI planes/objects), the bottleneck is the CPU in the 5080 case, since the GPU is not fully loaded.

 

Yes, the 5080 is a good candidate as well for me.

Edited May 4 by chiliwili69

[=SqSq=SignorMagnifico]

On 5/4/2025 at 1:05 PM, chiliwili69 said:

 

Hey!, Thank you for this video! With many data and many scenarios/planes. The gain you got is quite significant overall is around 30-40% with monitor 1440p and all setting maxed-out including MSAAx8 which requires a big extra GPU load.

 

It is interesting to see that the required GPU power of the 5080 is lower than the 3080Ti in all the test you did. Around one third less (Except in the last D-Day where it was even less)

On the other hand, the total UPS power is a bit higher since your CPU is rendering more fps.

 

Also, in the last D-Day test (with more AI planes/objects), the bottleneck is the CPU in the 5080 case, since the GPU is not fully loaded.

 

Yes, the 5080 is a good candidate as well for me.

Thank you for the view, and your detailed breakdown of what was happening. I tried the CPU heavy scenario on purpose, just to see what would happen, and I wasn’t surprised there was little difference. It performs as I expected it would: more FPS with fewer watts. Now I’m ready for when IL-2 Korea drops.

[1Sascha]

Unless you have money to burn, current GPU lineups are a bit... crappy for higher-res VR-users, aren't they? 

 

I was trying to decide between the 5070Ti and the 5080 and... I'm not too sure about the 5080. Would be an easier sell if it had 24 GB (and a larger die). As it is, its uplift over the 5070Ti seems a bit "meh", considering the difference in price and that it's called a "5080". Techpowerup puts it at only 15 percent higher overall than the 5070Ti, which is pretty disappointing, IMO.

When I bought my 5070Ti last week, I could've had a 5080 for "just" ~1050 EUR ("cyberweek"-sale at a German HW-vendor), but I decided against it after double-checking its relative performance. I also only have a 750W PSU, so I would've needed an upgrade there too (theoretically) with the bigger card.

 

Anyway: Just a 15% uplift over the next smaller NV-card while still costing over €1000 is a bit laughable for a xx80-class card, IMO. And to top it all off it, it doesn't even come with added VRAM - though if we can believe the leakers, they already have a 5080 Super with 24GB lined up for the refresh. Typical. 😕

 

 

I think my advice would be to stay put for now and if that's not an option, either go big or go one class smaller (meaning 40/5090 or 5070 Ti/9070 XT) and to avoid the 5080.

 

Squaddies of mine swear by the 7900 XTX because of its added VRAM (they fly mostly DCS), and one of them just bought a brand-new one a day before I got my 5070 Ti. But that thing's a 2.5 year old card at this point and still costs well over EUR 800, so... I'm not too sure I would get one in 2025. Shame that AMD decided to not produce any bigger cards than the 9070 XT this gen...

 

S.

 

 

 

 

 

 

Edited May 25 by 1Sascha

[giftgruen]

Hmm - my upgrade path was 3070=>4080=>4090 and now 5090 since some days

 

Without having concrete benchmark numbers, I'd say that with my Quest Pro headset, the 4090 was the 'biggest jump'.

But even the 5090 now ( which I bought because I found a good SecondHand-Deal and also plan to upgrade my HeadSet )

gives me the better visual experience, though a Quest Pro is not HighRes-VR nowadays.

 

I'd maybe add that I love MSAAx4 in IL2 and also high settings ( shadows, clouds and so on ) plus SuperSampling

because this really gives some eyecandy that can't be expressed in pure resolution.

 

In the end, my personal conclusion is simply: The more GPU, the better. Even if pure numbers tell that a GPU generally is able

to drive a Headset ( e.g. with low ingame settings ) the visual experience will be somewhat flawed.

 

The 4080 is not that huge upgrade from my experience, VR really benefits from the brute force that *90 cards bring in.

 

True: Expensive.  Maybe try to get a used one ( I always do so with GPUs ).

 

 

 

 

Edited August 28 by giftgruen

[chiliwili69]

IL-2 is a DX11 rasterization game (the only game I play), therefore I am interested in benchmarks using DX11.

 

Since DX11 is old, most of the current benchs are using DX12, Ray-tracing, DLSS, etc. So it is dificult to evaluate GPUs.

The only public benchmark I have found is this: https://www.videocardbenchmark.net/directx-11.html

 

 

And here it is surprising that the 7900XTX is above the 5090. I can not believe that.

 

Is there anyone with a 7900XTX which can run the IL-2 SYN_VANDER benchmark in the GPU 4K test?

 

In my city I saw a second hand Sapphire Vapor-X 7900XTX for 780€, I am considering to grab it to replace my current 3080. So, I will be able to test it myself for shake of really know it!

[LuftManu]

2 hours ago, chiliwili69 said:

IL-2 is a DX11 rasterization game (the only game I play), therefore I am interested in benchmarks using DX11.

 

Since DX11 is old, most of the current benchs are using DX12, Ray-tracing, DLSS, etc. So it is dificult to evaluate GPUs.

The only public benchmark I have found is this: https://www.videocardbenchmark.net/directx-11.html

 

 

And here it is surprising that the 7900XTX is above the 5090. I can not believe that.

 

Is there anyone with a 7900XTX which can run the IL-2 SYN_VANDER benchmark in the GPU 4K test?

 

In my city I saw a second hand Sapphire Vapor-X 7900XTX for 780€, I am considering to grab it to replace my current 3080. So, I will be able to test it myself for shake of really know it!

 

I had a 7900 XTX.

No way.

[Voxman]

As a 3080ti owner, I'm curious as well.

 

[chiliwili69]

21 hours ago, chiliwili69 said:

In my city I saw a second hand Sapphire Vapor-X 7900XTX for 780€, I am considering to grab it to replace my current 3080. So, I will be able to test it myself for shake of really know it!

Well, I decided to go for it. Tomorrow afternoon I will test it.

I don´t expect to be at the level or a 4090 or 5090 in IL-2, but I think I will get a good gain with respect to my 3080. Let´s see.

[chiliwili69]

I just got today the AMD Raedon RX 7900XTX ( Sapphire Nitro Vapor-X) and quickly run some test in 4K with the SYN_VANDER benchmark and oops, 

 

results here

Edited August 31 by chiliwili69

[Panzerlang]

So that DX11 PassMark, wtf! Lol.

[LuftManu]

13 hours ago, chiliwili69 said:

I just got today the AMD Raedon RX 7900XTX ( Sapphire Nitro Vapor-X) and quickly run some test in 4K with the SYN_VANDER benchmark and oops, 

 

results here

 

Hello:

RDNA2-3 performs worse than Nvidia on Il-2 as we commented many times! 

For Korea and with DLSS a possibility (commented in Discord) I would get an Nvidia. Even more for VR.

 

Kind regards,

[chiliwili69]

Yes, I know, I know.... NOW.

You know, sometimes I prefer to double check things myself. I have run the CPU test and have seen that they are even worse than with the 3080.

Something weird with AMD GPUs for IL-2. Lesson learnt. I will sell the 7900XTX.

 

Now looking into what card the choices are:

 

4090&5090: Out of radar really, I don´t think will needed for B2B (for now) and that prices are really high.

 

I was even looking at a 3090Ti second hand or a new 5080.

 

This is what chatGPT5 told me:

 

Your case is with IL-2 in VR + Bigscreen Beyond 2 and the choice between RTX 3090 Ti (used) or RTX 5080 (new) on a PCIe 4.0 system.

---

The key points first

• PCIe 4.0 will not bottleneck you with either a 3090 Ti or a 5080 (not even a 5090). Benchmarks show only ~0–2% difference between PCIe Gen5 x16 and Gen4 x16 in games; Gen3 x16 is where performance starts dropping more.

• Bigscreen Beyond 2: resolution is 2560×2560 per eye. It runs at 75 Hz natively and 90 Hz with internal upscaling (1920×1920→2560×2560), so in IL-2 you’ll usually target 75 fps native or 45 fps with stable reprojection.

• IL-2 in VR is often CPU-limited, especially in missions with lots of AI. CPUs with large cache (like AMD’s X3D series) help a lot.

---

RTX 3090 Ti vs RTX 5080 (what you gain/lose)

Raw gaming performance

• The 5080 is roughly 15% faster than the 4080 and below the 4090. Compared to the 3090 Ti, it’s usually a solid uplift (~20–35% depending on the title).

VRAM and bandwidth

• 3090 Ti: 24 GB GDDR6X, 384-bit bus, ~1008 GB/s bandwidth.

• 5080: 16 GB GDDR7, 256-bit bus, ~960 GB/s bandwidth.
  → For IL-2 in VR, 16 GB is enough. The 24 GB on the 3090 Ti only really helps with heavy mods or ultra-high-res textures.

Power and features

• 3090 Ti: ~450 W, runs hot, needs a strong PSU.

• 5080: ~360 W, plus DLSS 4 / Frame Generation, DisplayPort 2.1b, and Blackwell-era driver optimizations (also relevant for VR).

---

Recommendation

• Unless the 3090 Ti is dirt cheap, go for the 5080: better performance per watt, warranty, future-proof features (DP 2.1b, DLSS 4, Blackwell VR optimizations), and PCIe 4.0 will not hold it back.

• Pick the 3090 Ti only if you can get it very cheap (say under ~€400–450 in good condition) and if the 24 GB VRAM is valuable to you. For gaming/VR alone, the 5080 is the better choice

[DBCOOPER011]

8 hours ago, chiliwili69 said:

Recommendation

• Unless the 3090 Ti is dirt cheap, go for the 5080: better performance per watt, warranty, future-proof features (DP 2.1b, DLSS 4, Blackwell VR optimizations), and PCIe 4.0 will not hold it back.

• Pick the 3090 Ti only if you can get it very cheap (say under ~€400–450 in good condition) and if the 24 GB VRAM is valuable to you. For gaming/VR alone, the 5080 is the better choice

 

I would recommend the 5080 or wait for the 5080 super. The 5080 is available now at MSRP ($999.00) prices, atleast in the US. It appears you can get almost 4090 performance with an overclock on them.  

[Aapje]

I would recommend waiting for the Super-cards if you have a somewhat decent card right now. There are a bunch of indications that the VRAM requirements will shoot up in the future for at least some AAA games, so having 50% more VRAM is probably going to improve the resale value quite a bit. And many of the flight sims are quite VRAM hungry.

 

Rumor is that the Super-cards will be released this year.

[Dagwoodyt]

18 hours ago, chiliwili69 said:

• 3090 Ti: 24 GB GDDR6X, 384-bit bus, ~1008 GB/s bandwidth.

• 5080: 16 GB GDDR7, 256-bit bus, ~960 GB/s bandwidth.
  → For IL-2 in VR, 16 GB is enough. The 24 GB on the 3090 Ti only really helps with heavy mods or ultra-high-res textures.

Power and features

• 3090 Ti: ~450 W, runs hot, needs a strong PSU.

• 5080: ~360 W, plus DLSS 4 / Frame Generation, DisplayPort 2.1b, and Blackwell-era driver optimizations (also relevant for VR).

 

 

Based on those specs I don't understand why a 5080 would perform better than a 3090Ti in Il-2 GB VR. The "runs hot" assertion neglects those EVGA cards that use a vapor chamber. 

[chiliwili69]

On 9/1/2025 at 2:13 AM, Panzerlang said:

So that DX11 PassMark, wtf! Lol.

Yeah, those numbers from DX11 Passmark for AMD cards contain an evil inside.

I also run the GPU passmark app (PerformanceTest) without any upscaling, framegeneration or OC, and I obtained these numbers:

 

For the DX11 bench I obtained 243, which is quite below the avg of the DX11 passmark show in the public table (356).

In the general 3D Graphics Mark I had 24797.

 

My theory to explain this big difference is that some users of that AMD card are selecting the AMD Adrenalin GPU software the Highperformance (HYPR-RX) or Quality modes which automatically activates the upscaling techiniques and synthetic frame generation:

 

 

In fact, if you compare the different baselines of that GPU you will see high numbers around 34000-38000 and others around 24000-26000, meaning the people who activate those modes or not.

 

[chiliwili69]

16 hours ago, Dagwoodyt said:

Based on those specs I don't understand why a 5080 would perform better than a 3090Ti in Il-2 GB VR.

 

Yes, it is not easy to trust on public benchmarks to consider one card or other if our interest is IL-2 GB (a DX11 game) in VR.

 

If you look at the table in the first post of this thread you will see there is another public banchmark, FireStrike Ultra(4K) which is based in DX11:

 

https://www.3dmark.com/search

 

 

If we compare the Firestrike(avg) with the tests we did with the GPU test of the IL-2 SYN_VANDER benchmark we obtain a nice straight line:

 

 

So far, it reproduces the performance we are obtaining in IL-2. But of course, not many samples are in IL-2 SYN_VANDER test.

Based in that linear aproximation a 3090Ti should deliver around 143 fps in the SYN_VANDER GPU test and the 5080 shoud deliver around 180 fps.

 

But this is just theory! You never know.

 

If I compare the specs of the 3090Ti and the 5080, they have similar bandwidth but I like the bigger 384-bit bus of the 3090Ti.

They have the same number of processors, but base/boost clock of the 5080 run higher, so it should perform better (in theory)

[LuftManu]

53 minutes ago, chiliwili69 said:

 

Yes, it is not easy to trust on public benchmarks to consider one card or other if our interest is IL-2 GB (a DX11 game) in VR.

 

If you look at the table in the first post of this thread you will see there is another public banchmark, FireStrike Ultra(4K) which is based in DX11:

 

https://www.3dmark.com/search

 

 

If we compare the Firestrike(avg) with the tests we did with the GPU test of the IL-2 SYN_VANDER benchmark we obtain a nice straight line:

 

 

So far, it reproduces the performance we are obtaining in IL-2. But of course, not many samples are in IL-2 SYN_VANDER test.

Based in that linear aproximation a 3090Ti should deliver around 143 fps in the SYN_VANDER GPU test and the 5080 shoud deliver around 180 fps.

 

But this is just theory! You never know.

 

If I compare the specs of the 3090Ti and the 5080, they have similar bandwidth but I like the bigger 384-bit bus of the 3090Ti.

They have the same number of processors, but base/boost clock of the 5080 run higher, so it should perform better (in theory)

 

Hi! Ill conduct a 5080 test later today on the 4k Benchmark on the game so we can see real difference while in game.  The 5080 with GDDR7 has almost the same bandiwth as the 4090 on GDDRR6X. The bus is the main difference.  The GDDR7 has the clocks at 32gbps but ships with 30gbps speed (5080s have a big OC margin, so we can expect the 5080 Super to have this by default) so we could also try to underclock and overcock to see how this changes the result on the engine (Thats a fun test for today!)

 

Running an Asus 5080 currently

 

https://www.3dmark.com/fs/33640204

https://www.3dmark.com/spy/57797888

 

 

Regarding FireStrike and these tests, I would start measuring using Steel Nomad DX12 or TimeSpy Ultra for having a real sense on whats coming with Korea and expected performance. But again, ingame could perform the opposite. Game engines are funny! 

Kind regards,

[Dagwoodyt]

Just a guess but I suspect that anyone who buys a 5080 to replace a 3090Ti to run Il-2 GB in VR is going to end up feeling deflated. 

Edited September 3 by Dagwoodyt

[dgiatr]

4 hours ago, Dagwoodyt said:

Just a guess but I suspect that anyone who buys a 5080 to replace a 3090Ti to run Il-2 GB in VR is going to end up feeling deflated. 

 

7 hours ago, chiliwili69 said:

 

Yes, it is not easy to trust on public benchmarks to consider one card or other if our interest is IL-2 GB (a DX11 game) in VR.

 

If you look at the table in the first post of this thread you will see there is another public banchmark, FireStrike Ultra(4K) which is based in DX11:

 

https://www.3dmark.com/search

 

 

If we compare the Firestrike(avg) with the tests we did with the GPU test of the IL-2 SYN_VANDER benchmark we obtain a nice straight line:

 

 

So far, it reproduces the performance we are obtaining in IL-2. But of course, not many samples are in IL-2 SYN_VANDER test.

Based in that linear aproximation a 3090Ti should deliver around 143 fps in the SYN_VANDER GPU test and the 5080 shoud deliver around 180 fps.

 

But this is just theory! You never know.

 

If I compare the specs of the 3090Ti and the 5080, they have similar bandwidth but I like the bigger 384-bit bus of the 3090Ti.

They have the same number of processors, but base/boost clock of the 5080 run higher, so it should perform better (in theory)

My 3090 ti results are in the il2 Synvander table with hp G2....

 

For now, I am very happy with my gpu running Quest Q3 at about 3000 resolution , 73 hz, opencomposite and openxr toolkit...Running my gpu a little OC to cope with Q3 compression extra gpu load...

Edited September 3 by dgiatr

[LuftManu]

Just did some GPU test with the 5080.

 

4K GPU TEST - 2025-09-03 15:19:20 - Il-2 
Frames: 10689 - Time: 60000ms - Avg: 178.150 - Min: 151 - Max: 222 (Stock 2917 Mhz)

 

4K GPU TEST 2025-09-03 15:23:06 - Il-2
Frames: 11004 - Time: 60000ms - Avg: 183.400 - Min: 155 - Max: 232 (OC +270 Core / +2000 Memory / 112% TDP)

 

 

 

Kind regards!

 

[chiliwili69]

On 9/3/2025 at 3:27 PM, LuftManu said:

Just did some GPU test with the 5080.

LOL!, your average fps for the GPU test is exactly 180.7!  as predicted by the FireStrike Ultra becnhmark:

[chiliwili69]

Since the B2B is expected to arrive within this month of september I wanted to have something better than the 3080 to properly evaluate the B2B with some supersampling.

So, I decided to go for a 5080 card which I tested with the SYN_VANDER and I am quite happy with the GPU results (here).

 

 

With no OC I obtain 189 fps!, very close to the performance of a 4090 (~192fps). @DBCOOPER011, as you commented above!

So price-wise, for this DX11 game it is worth to go for a new 5080 instead of a second hand 4090.

Thank you for the advices and comments.

[chiliwili69]

Just some extra data about this MSI Inspire 5080 card.

I have just logged some trends with Afterburner (but without doing overclocking in the GPU) with my Quest3 at 200%SteamVR supersampling (about 19 million pixels) at 72Hz mode and no MSAA and the GPU was really relaxed in a 4x4 QMB in Kuban map:

GPU usage around 50%:

 

Power around 160W

 

VRAM close to 12GB (5080 has 16.4 GB)

 

 

GPU temp were around 62C and fans at 30%, very silent.

 

With these numbers I think I would be able to run the BSB2 around 150% SS (38million pixels) at 75Hz, and the GPU will not need to be loaded by the compression work needed for the streaming of the Quest3.