SYN_Vander BENCHMARK v6 to measure IL-2 performance in monitor & VR

May 22

After months of absence I decided to dust off the old flightstick and got me a new GPU for my b-day. RTX 5070 Ti which I just tested a bit flying around in a few quick missions. Since the 4070 Super was quite capable of running the game on the Q3 and at 80 Hz, I had hoped to be able to go to 90Hz with the new card (with similar gfx-settings). And while it ran smoothly 98% of the time with lots of headroom, low level stuff over bigger cities gave me occasional stutters down to 83, 84 FPS... 😕

Went back to 80 Hz, even increased the in-set res by a bit and switched on MSAA x2 and it was smooth sailing all around. 30 to 50 % headroom while flying nap of the earth over a built up area with 12 AI planes and some AI-ground vehicles around on Normandy... funny thing is that the GPU *really* wasn't breaking a sweat - at least not according to the monitoring data.. it boosted pretty high, but never exceeded 60% usage and ~200W power-draw (max is 300W).

Oh well... sorry for going OT there... I'll shut up until I fire up the old benchmark over the weekend to see what my results are in there...

Edited May 23 by 1Sascha

May 23

Started with the CPU Test, but even though I still don't have a 4K-Monitor I will run the GPU-Test next, just to see what will happen. CPU results were done with both 100% stock profile on the GFX-card and then with my max stable OC of +390 core/+2000 VRAM - the results were *literally* identical, so I picked one at random (think this is the OC-result). Which seems to indicate that either my manual OC is pointless (it isn't in 3DMark and other games) or that the CPU-benchmark does a good job of mainly/only targeting the CPU ... I

Result for the CPU bench:

2025-05-23 08:43:11 - Il-2
Frames: 7859 - Time: 60000ms - Avg: 130.983 - Min: 112 - Max: 162

Might be noteworthy that while Vsync was set to off in the game, my first attempt in the GPU-test resulted in "pegged" FPS at 165 (I'm on a 165 Hz monitor). Plus I got the exact same results - down to all three decimal points on the average - for both the stock and OC-run, so I suspected Vsync might have not been off.

Went into NV control-panel where Vsync was set to "use app's setting" (at least I think that's what that means) but it was apparently still active.

BTW: Do I want to keep that switch on the default setting or turn it on or off for day-to-day VR? Or is it irrelevant for VR?

Switched it to off in the CP and got this (on my OC-profile):

Result for the GPU bench (run at 1440p because ... no 4K-monitor... )

2025-05-23 09:16:52 - Il-2
Frames: 11590 - Time: 60000ms - Avg: 193.167 - Min: 162 - Max: 250

VR is going to take a bit more effort but I will run those next or later today.

EDIT: VR Test 1 done and here are the results:

2025-05-23 09:43:52 - Il-2
Frames: 4821 - Time: 60000ms - Avg: 80.350 - Min: 70 - Max: 93

I was too lazy to switch over to SteamVR so this was on OpenXR/OpenComposite. Since the Occulus App has three resolutions it labels "1.0x" at 90 Hz, I chose the lowest of the three and then set the required per-eye resolution within the OpenXR Toolkit.

Sadly, VR-Test2 crashed the game (D3D error, IIRC?) and since I was lazy I tried re-running it with a lower OC on the GPU without re-booting the PC. It ran fine that time, but FRAPS didn't record the data for that second run. Going to have to re-run that one at a later time..

Apart from the upgrade to the 5070 Ti, my system config hasn't changed, but here is the gist of it again:

Mobo: MSI Tomahawk Z690 DDR4 - all chipset and BIOS/ME stuff up-to-date (please don't melt my CPU!!)

CPU: Intel i7-14700 KF

RAM: 2x 16 GB G.Skill TridentZ RGB DDR4 3600/CL16, XMP on

GPU: Palit RTX 5070 Ti GamingPro OC - manual OC to +390 core and +2000 VRAM

NV-driver: 576.52

OS: Windows 11

On a less benchmark-y note (and as I mentioned earlier): I played around with the settings on the new card yesterday, thinking I could get away with 90 Hz with the same in-game settings/res that the 4070 Super *just* managed offline on 80 Hz. Which... seemed to work fine at first, but when I did some low level flying over a bigger city on the Normandy map, I did get some stutters down to 82 - 84 FPS. Nothing dramatic, but *definitely* noticeable and annoying.

Went back to 80 Hz and even cranked up the in-game settings a bit, switching from FXAA to x2 MSAA just to see what would happen... and the game ran totally smoothly at those settings, even with MSAA on. Wouldn't have thought that refresh-rate makes a bigger difference than MSAA, but I guess I was wrong.

Not sure if I'll keep it at 80 Hz with more eye-candy or start looking for in-game settings that will allow the game to run smoothly at 90 Hz.

Edited May 23 by 1Sascha

May 23

12 hours ago, 1Sascha said:

that the CPU-benchmark does a good job of mainly/only targeting the CPU ... I

Yes, this is the purpose of the CPU test, to bring to the limit the CPU, not the GPU, that´s why it is done at 1080p.

12 hours ago, 1Sascha said:

Or is it irrelevant for VR?

I think it is irrelevant for VR.

12 hours ago, 1Sascha said:

esult for the GPU bench (run at 1440p because ... no 4K-monitor... )

I will not upload this to the table since it is not 4K. If you have a 4K TV you can connect the PC to it and run it.

May 23

12 hours ago, 1Sascha said:

Not sure if I'll keep it at 80 Hz with more eye-candy or start looking for in-game settings that will allow the game to run smoothly at 90 Hz.

Thank you for posting, it is the first 5070Ti card here. That card couple very well with the Quest3, since it doesn´t demand to render a high number of pixels.

Regarding frequencies, with the Quest3 I always use 72Hz, since for the limited bandwidth (200 Mbps for H265) it gives better image quality that at 90Hz.

Do you really notice the differnce between 80Hz and 90Hz? You can ask to a friend or family member to make you a test of let´s say 20 tests (some of them at 80Hz and other at 90Hz) in an easy map and scenario (toreach always either 80fps or 90fps) and see if you are able to guess what frequency is using.

May 24

15 hours ago, chiliwili69 said:

I will not upload this to the table since it is not 4K. If you have a 4K TV you can connect the PC to it and run it.

Well.. yeah...naturally. I've always run the 4K-test at 1440p as well just to see what would happen for myself and while I know you can't use them for the table, I've always posted those results here too because... why not? Plus the results might be of interest to some folks and since I have the data anyway...

Besides: As you can see in my results, the new card did not produce much of an improvement in the CPU-test (as would be expected with the exact same CPU, RAM and board) and I wanted to see *some* uplift in my scores... 😄

Re refresh rates.... I'm not *too* sure after only playing around with the Q3 again for a bit now (after a months-long break), but I do remember seeing a difference between 72 Hz and 80 (or 90?) at the same in-game settings and per-eye-res. Never really used 90 before as the 4070 Super was *just* enough for 80 Hz (with the right in-game settings) and was of course totally out of its depth at 90 Hz.

Of course: 72 Hz isn't nearly as taxing on the GPU so with the new card I might get away with perhaps even 4x MSAA at that refresh-rate. And MSAA, in my experience, makes the game look noticeably prettier already. I remember there was some IL-2 YT-er who upgraded to my CPU and to a 4070 Ti Super last year and who posted his settings for VR... Guess I'll use his values and start from there.

What I still don't get is that this card too (just like the 4070 Super before) isn't reporting anywhere near full load while doing VR. In fact, it reports about the same percentage values for GPU-load as the 4070 S did (~70 - 75%) and pull about the same amount of power (~170W). And while it also doesn't run very hot (monitor games can warm it up more), it does boost pretty high... I've seen over 3200 MHz on the core in VR whereas I'll never see more than 3100, 3150 in monitor gaming or benchmarks - both while on the same OC-profile.

Weird.

It is a worthwhile upgrade though (IMO) if anyone's wondering. I paid a little under my local MSRP - which is the only reason I decided to do the update now. Which is still *a lot* for a GPU of this class, IMO and I would *not* recommend this card at EUR 900 or 1000.

Though those prices seem to be a thing of the past, at least for some models right now and in Germany. MSRP is 879 and I paid 849 for the Palit GamingPro OC-model - cheapest one on the day would've been the MSI Shadow for 797 incl shipping, but I didn't want that one because plastic backplate and because MSI's entry-level coolers are notoriously bad.

May 31

Took me a while but here are the results for VR Test2:

2025-05-31 21:08:31 - Il-2
Frames: 3483 - Time: 60000ms - Avg: 58.050 - Min: 42 - Max: 91

Note: This was done with a manual OC on the GPU ... +340 MHz Core and +2000 on the VRAM in Afterburner.

EDIT: And after that I played around in a busy quick mission over Le Havre and... I think the upgrade did pay off.. 😄

Lowest 1.0x resolution set in the Occulus-App, high in-game preset, high-ish details including MSAA x2 at 80 Hz and with ~2800x3000-something resolution set in the Toolkit and even when flying nap of the earth through the city streets I never got less than 80 FPS and, according to the overlay, never below 23 percent of free GPU-capacity. I don't think I've ever seen the game run this smoothly in VR, even with my old 4070 Super at 72 Hz and with markedly lower in-game settings. Couldn't even dream of activating MSAA on that card..

But I guess an offline career-mission will be the real test as some of those are even more demanding on the system than even the busiest quick missions.

S.

Edited May 31 by 1Sascha

June 24

Motherboard:    Z97 Gaming 5 (MS-7917)
CPU:            Intel Core i7-4790K
CPU Freq:       4.4 Ghz
Cores:          4 (number of active physical Cores)
Threads:        4 (HyperThreading is off)
RAM size:       16Gb (4x4GB) DDR3
RAM Freq:       2134 MHz (this is 2xDRAM freq with Dual channel)
NB Freq:        4000 MHz (or Uncore Frequency or UCLK)
RAM timings:    9-11-10-28
GPU:            Nvidia GeForce RTX 2080 
OS:             Windows 10
IL2 version:    6.001

2025-06-23 22:11:24 - Il-2
Frames: 4358 - Time: 60000ms - Avg: 72.633 - Min: 60 - Max: 97

2025-06-23 22:17:40 - Il-2
Frames: 4287 - Time: 60000ms - Avg: 71.450 - Min: 62 - Max: 97

Edited June 24 by tati

July 4

On 6/24/2025 at 8:20 AM, tati said:

Motherboard:    Z97 Gaming 5 (MS-7917)
CPU:            Intel Core i7-4790K
CPU Freq:       4.4 Ghz
Cores:          4 (number of active physical Cores)
Threads:        4 (HyperThreading is off)
RAM size:       16Gb (4x4GB) DDR3
RAM Freq:       2134 MHz (this is 2xDRAM freq with Dual channel)
NB Freq:        4000 MHz (or Uncore Frequency or UCLK)
RAM timings:    9-11-10-28
GPU:            Nvidia GeForce RTX 2080 
OS:             Windows 10
IL2 version:    6.001

2025-06-23 22:11:24 - Il-2
Frames: 4358 - Time: 60000ms - Avg: 72.633 - Min: 60 - Max: 97

2025-06-23 22:17:40 - Il-2
Frames: 4287 - Time: 60000ms - Avg: 71.450 - Min: 62 - Max: 97

Thanks for posting the first benchmark with version 6.001. You used a legendary CPU. It was my first CPU when I started to do overclocking!

July 9

Motherboard:	ASUS Crosshair VIII Hero (WI-FI)
CPU:			AMD Ryzen 7 5800X3D
CPU Freq:		4.450 Ghz
Cores:			8
Threads:		16
RAM size:		32Gb (2x16GB)
RAM Freq:		DDR4-3600 MHz
NB Freq:		1800 MHz
RAM timings:	16-16-16-36
GPU:			AMD Radeon RX 7900 XTX
OS:				Windows 11 Pro 24H2
IL2 version:	6.001

2x Active Monitor (Primary XG27AQDMG 2560x1440 / Secondary LG 27GN880-B 2560x1440)

GPU Test @1920x1080 Frames: 6090 - Avg: 101.500 - Min: 80 - Max: 134

@2560x1440 Frames: 6121 - Avg: 102.017 - Min: 86 - Max: 140

July 15

On 7/9/2025 at 9:05 PM, FG28_THE_ERASER said:

Motherboard:	ASUS Crosshair VIII Hero (WI-FI)
CPU:			AMD Ryzen 7 5800X3D
CPU Freq:		4.450 Ghz
Cores:			8
Threads:		16
RAM size:		32Gb (2x16GB)
RAM Freq:		DDR4-3600 MHz
NB Freq:		1800 MHz
RAM timings:	16-16-16-36
GPU:			AMD Radeon RX 7900 XTX
OS:				Windows 11 Pro 24H2
IL2 version:	6.001

2x Active Monitor (Primary XG27AQDMG 2560x1440 / Secondary LG 27GN880-B 2560x1440)

GPU Test @1920x1080 Frames: 6090 - Avg: 101.500 - Min: 80 - Max: 134

@2560x1440 Frames: 6121 - Avg: 102.017 - Min: 86 - Max: 140

Thank you for this test with that top AMD GPU 7900XTX, it is an interesting card.

In your test you almost duplicate the number of pixels when going from 1080p to 2.5K but the benchmark score remain the same, this means that the real bottleneck is in the CPU. I will not update your test in the table since for the GPU test only 4K test are valid.

Did you run the CPU test?

July 17

04.07.2025 в 10:53, chiliwili69 сказал:

Thanks for posting the first benchmark with version 6.001. You used a legendary CPU. It was my first CPU when I started to do overclocking!

This legend served me for many years. Promised myself not to upgrade till there would be a CPU that would double performance at reasonable price. This day has come.

Default BIOS settings + XMP + Intel 200S Boost

Motherboard:    MSI PRO Z890-P WIFI (MS-7E34)
CPU:            Intel Core Ultra 7 265k
CPU Freq:       5.5 Ghz
Cores:          8P + 12E  (number of active physical Cores)
Threads:        20
RAM size:       48Gb (2x24GB) DDR5
RAM Freq:       4000 MHz (8000MT/s)
NB Freq:        3800 MHz (Uncore Frequency in CPU-Z)
RAM timings:    38-52-52-136-671 2T
GPU:            Nvidia GeForce RTX 2080 
OS:             Windows 10
IL2 version:    6.001

2025-07-09 01:13:10 - Il-2
Frames: 8485 - Time: 60000ms - Avg: 141.417 - Min: 120 - Max: 187

Overclocked:

Motherboard:    MSI PRO Z890-P WIFI (MS-7E34)
CPU:            Intel Core Ultra 7 265k
CPU Freq:       5.6 Ghz
Cores:          8P + 12E  (number of active physical Cores)
Threads:        20
RAM size:       32Gb (2x16GB) DDR5
RAM Freq:       3600 MHz (7200 MT/s)
NB Freq:        4000 MHz (Uncore Frequency in CPU-Z)
RAM timings:    34-42-42-84-574 2T
GPU:            Nvidia GeForce RTX 2080 
OS:             Windows 10
IL2 version:    6.001

2025-07-13 17:37:03 - Il-2
Frames: 9490 - Time: 60000ms - Avg: 158.167 - Min: 137 - Max: 207

July 17

I'm looking at upgrading my computer I built way back in 09. Since then I've upgraded my video card twice but it's at the end of it's life given the motherboard and CPU. I've looked at a lot of these posts but since there are 36 pages of results and comments -- I haven't look at all of them and haven't looked at many of the older posts (pre 2022). Because none of the current online video card reviews list IL-2 in their reviews (that I've found), and some of them list up to 45 different games, are any of those games roughly equivalent to what IL-2 might do with a graphics card?

July 18

22 hours ago, tati said:

This legend served me for many years. Promised myself not to upgrade till there would be a CPU that would double performance at reasonable price. This day has come.

Indeed!

Thank you for this nice comparison before and after your upgrade.

In the overcloked test it is dificult to tell from where it is obtained the gain, from the 0.1Ghz overclok or from the RAM timmings.

Enjoy your upgrade!

July 18

9 hours ago, Crockett said:

are any of those games roughly equivalent to what IL-2 might do with a graphics card?

Not to my knowledge.

Summarizing these 22 pages of tests, you can just take a look of the spreadsheet with the results of the tests (CPU, GPU, VR).

Depending of you specific use case (Monitor1080, monitor 2.5K, monitor 4K, or VR, what VR) you will need to decide for the GPU.

Regarding CPU, since IL-2 depends on single-core performance, the best way to know how a CPU will perform is to look at this table https://www.cpubenchmark.net/single-thread/

If you play in monitor, most of the CPUs will be more than fine.

But if you play in VR, depending on budget the best CPU options are AMD 9800X3D or Intel 14900K or Intel Ultra9 285K

July 18

14 hours ago, Crockett said:

I'm looking at upgrading my computer I built way back in 09. Since then I've upgraded my video card twice but it's at the end of it's life given the motherboard and CPU. I've looked at a lot of these posts but since there are 36 pages of results and comments -- I haven't look at all of them and haven't looked at many of the older posts (pre 2022). Because none of the current online video card reviews list IL-2 in their reviews (that I've found), and some of them list up to 45 different games, are any of those games roughly equivalent to what IL-2 might do with a graphics card?

The most similar games are flight or racing sim titles, like MSFS, Asetto Corsa and F1. But it is still inferior to looking at the the tests performed here, as @chiliwili69 correctly notes.

However, keep in mind that a new game engine is on the horizon, so the performance in the current game engine is not going to be the same for that game engine. So instead of aiming for something that works well in IL-2 right now, you can look into a solid card for the future.

I'm personally very much into getting good bang for your buck, but unfortunately the situation now is not optimal. The situation at the lower end is rather poor, with the 9060 XT 16 GB being the only decent option, but it is already so expensive that it makes more sense to get the 9070 XT instead if you can swing it.

However, given that your motherboard and CPU are the reason to upgrade, you can also choose to only do a platform upgrade now, and do a GPU upgrade later. We should get Super cards with increased VRAM from Nvidia at the end of the year or early next year, and then I expect there to be many more decent GPU options across various price points.

4 hours ago, chiliwili69 said:

But if you play in VR, depending on budget the best CPU options are AMD 9800X3D or Intel 14900K or Intel Ultra9 285K

I would not advise the 14900K unless you also use the PC for productivity software that can use those cores. It is way too expensive and too hot for just gaming. The 285k option can be decent if you can get a very good deal, for example from Microcenter for Americans, but it is a dead-end platform.

I would not sleep on the 7800X3D, which in many places is much cheaper than the 9800X3D and not that much slower.

Edited July 18 by Aapje

July 20

On 7/18/2025 at 4:22 AM, Aapje said:

But if you play in VR, depending on budget the best CPU options are AMD 9800X3D or Intel 14900K or Intel Ultra9 285K

I don't do VR -- only Track IR which I'm happy with and don't plan to go VR anytime in the future. I've always been an Intel fanboy but I'm retired and mainly use my computer to fly IL-2 so don't need to consider productivity! I've been looking at the AMD 9800X3D and won't be doing any upgrade before Christmas, so will look at video cards then.

Thanks for your suggestions

July 20

On 7/15/2025 at 8:44 AM, chiliwili69 said:

Thank you for this test with that top AMD GPU 7900XTX, it is an interesting card.

In your test you almost duplicate the number of pixels when going from 1080p to 2.5K but the benchmark score remain the same, this means that the real bottleneck is in the CPU. I will not update your test in the table since for the GPU test only 4K test are valid.

Did you run the CPU test?

Oh, sorry, disregard my previous Scores. I think I had the wrong settings applied each time, and I'm unsure why. I did the Benchmark again.

Here are the correct Ones for the CPU Test:

@1920x1080 Frames: 6162 - Avg: 102.700 - Min: 84 - Max: 139

And for anyone in the Forum who likes to see (as Fluff) at 1440p and 1440p all maxed settings:

@2560x1440 Frames: 6175 - Avg: 102.917 - Min: 85 - Max: 143 / Same Settings as the CPU Test.

@2560x1440 Frames: 6106 - Avg: 101.767 - Min: 85 - Max: 135 / All Maxed out with HDR etc.

I wonder if we should do five runs and then calculate the average of those five runs. I had all the programs open that I usually use for flying (TrackIR Joystick connected, etc.), and since it's summer in Switzerland right now, I have an average room temperature of 26 °C in the Room. Therefore, the manual Fan Control is active for the Case Fans and GPU at 100% speed.

August 29

I have just run the CPU Test ang GPU Test with my current setup (CPU:7800X3D, GPU:3080) with the new IL-2 v6.002.

CPU Test: (monitor 1080p)
Frames: 8984 - Time: 60000ms - Avg: 149.733 - Min: 125 - Max: 187
Frames: 9142 - Time: 60000ms - Avg: 152.367 - Min: 123 - Max: 202
Frames: 9081 - Time: 60000ms - Avg: 151.350 - Min: 126 - Max: 192

It looks that there is a +6fps with respect to my previous test in v5.201. 🙂

GPU test: (monitor 4K)

Frames: 7424 - Time: 60000ms - Avg: 123.733 - Min: 101 - Max: 161
Frames: 7399 - Time: 60000ms - Avg: 123.317 - Min: 101 - Max: 159
Frames: 7419 - Time: 60000ms - Avg: 123.650 - Min: 102 - Max: 162

It looks that there is a -5fps with respect to my previous test in v5.201. 😞

August 29

So you net +1 FPS! Don't worry, be happy! 😁

S!Blade<><

August 31

Well, I did some CPU test and GPU 4K with my recently arrived second hand Sapphire Nitro Vapor-X 7900XTX and the results are even worse than with the 3080!!:

Just with default setting, no overclok netither any FSR upscaling or frame generation techniques.

CPU Test (1080p): (about 20 fps less than with the 3080)

Frames: 8026 - Time: 60000ms - Avg: 133.767 - Min: 121 - Max: 164
Frames: 7905 - Time: 60000ms - Avg: 131.750 - Min: 122 - Max: 164
Frames: 7871 - Time: 60000ms - Avg: 131.183 - Min: 122 - Max: 159

GPU Test (4K Monitor): (about 14 fps less than with the 3080)

Frames: 6557 - Time: 60000ms - Avg: 109.283 - Min: 100 - Max: 121
Frames: 6550 - Time: 60000ms - Avg: 109.167 - Min: 100 - Max: 121
Frames: 6552 - Time: 60000ms - Avg: 109.200 - Min: 99 - Max: 121

I don´t know if I am doing something wrong but at least I was expecting to perform better than my previous 3080.

With this results I really don´t need to do any benchmark test in VR.

My son also complained about how badly it was performing in Cyberpunk compared with the 3080, but that´s logicall since that game uses raytracing.

I know that in the past AMD GPUs performed worse but I naively thought that it was something from past...

You should know that a second hand market a 3080 is now for about 350€ and the sapphire 7900XTX is around 750€

so conclusion is easy:

DON´T GO for AMD GPUs for IL-2

Edited September 1 by chiliwili69

September 1

added CPU test (with my 7800X3D) in the post above with the 7900XTX

September 3

Hi! Adding my scores on Core Ultra 9 after a few months of fixes and good ram timings.
Doing now 4K.

CPU TEST - 2025-09-03 15:02:23 - Il-2
Frames: 10051 - Time: 60000ms - Avg: 167.517 - Min: 138 - Max: 203

Motherboard: Asus ROG Z890 Apex
CPU: Core Ultra 285K (SP 88 - Pcores 88 - Ecores 98 MC68)
CPU Freq: 5.7 Ghz Pcores (5.6 All core OC) / 5.0 Ghz Ecores (All core OC)
Cores: 24 (number of active physical Cores)
Threads: 24
RAM size: 48Gb (2x24GB)
RAM Freq: 8200 MHz (OC) 67.8 ns Latency
NB Freq: 4100 MHz (or Uncore Frequency or UCLK) (OC)
RAM timings: 38-48-48-84
GPU: RTX 5080 Asus Astral OC version (No OC)

CPU with D2D 34 / NGU 34 / Cache 41

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)

Seems like General quality presets affect CPU a lot for the Core Ultra, cause I'm getting better FPS on the 4K GPU test?

September 4

18 hours ago, LuftManu said:

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)

Seems like General quality presets affect CPU a lot for the Core Ultra, cause I'm getting better FPS on the 4K GPU test?

Many thanks for these first GPU tests for the 5080, it confirms the expected performance of that card for IL-2. I added both tests to the table.

The CPU test (using 1080p) is just focusing to load CPU and unload GPU. So, only CPU is bottlenecked.

The GPU test (using 4K) is just focusing to load GPU and unload CPU. So, only GPU is bottlenecked.

So, we can evaluate individually the CPU and GPUs. So, it is normal to obtain different numbers (higher or lower) in CPU test and GPU test.

Obviuosly there are interactions in each model of CPU and GPU (ie a bad GPU can influence the CPU test and viceversa)

Edited September 4 by chiliwili69

September 4

After my low number with the 7900XTX in IL-2, I resolved to grab a new 5080 card, in particular MSI 5080 Inspire 3X OC for 1240€. It has a small form factor, it is a bit smaller than my previous 3080!

Just for this first tests I didn´t do any OC over the 5080, just installed the NVIDIA drivers and nothing of MSI OC software.

I run the CPU 1080 tests (same procesor than before:7800X3D) and GPU 4K tests:

CPU Test (monitor 1080p):
Frames: 8685 - Time: 60000ms - Avg: 144.750 - Min: 124 - Max: 188
Frames: 8804 - Time: 60000ms - Avg: 146.733 - Min: 126 - Max: 189
Frames: 8724 - Time: 60000ms - Avg: 145.400 - Min: 124 - Max: 189

GPU Test (monitor 4K):

Frames: 11359 - Time: 60000ms - Avg: 189.317 - Min: 160 - Max: 244
Frames: 11361 - Time: 60000ms - Avg: 189.350 - Min: 160 - Max: 247
Frames: 11281 - Time: 60000ms - Avg: 188.017 - Min: 160 - Max: 239

With respect to my previous 3080 card, I am greatly surprised with this +66fps in the GPU test results with no OC. I don´t know why in the CPU test I was -5fps.

I think it would be a good GPU for my next VR headset (B2B). Of course, If I decide to keep it over my current Quest3.

September 4

2 hours ago, chiliwili69 said:

Many thanks for these first GPU tests for the 5080, it confirms the expected performance of that card for IL-2. I added both tests to the table.

The CPU test (using 1080p) is just focusing to load CPU and unload GPU. So, only CPU is bottlenecked.

The GPU test (using 4K) is just focusing to load GPU and unload CPU. So, only GPU is bottlenecked.

So, we can evaluate individually the CPU and GPUs. So, it is normal to obtain different numbers (higher or lower) in CPU test and GPU test.

Obviuosly there are interactions in each model of CPU and GPU (ie a bad GPU can influence the CPU test and viceversa)

Thanks!

Yes, but it surpirsed me to see that much difference as the GPU on my normal settings 4K/MAX usually sits almost idle unless at ground with tons of action. The CPU really is the most heavy factor on Il-2 unless you are running a few year old card.

I bet Korea will better use this and other CPUs on this part! and if we have DLSS then things will be even more easier on the GPU side, having the CPU the most deciding factor IMO.

Also glad you found that 5080! That's a cool model and also, the performance is there. The 5080s are REALLY easy to OC.
If you have MSi afterburner, push +250 on the clock and +2000 on the memory (max) and you will gain even a 5% in some cases.

PS: I just analyzed your data (Thanks again for this compilation!) and my current config matches my old one, with a more cheaper setup.

This 285K / Apex / Fast ram (used the same kit with OC) and a 5080, gives me the same performance as a 14900KS / 4090. While using a lot less W and costing less money per frame.

285K / Apex -> Amazon Sale 910€ 70W gaming
5080 -> Amazon Sale 1510€ 200W gaming

Total: 2420€ / 270W

14900KS / Board -> 1460€ 150W gaming
4090 -> 2100€ 300W gaming

Total: 3560€ / 450W total

(We can also save some money here with a Ultra 7, as it has LESS latenc than the Ultra 9, as the gentleman above has shown)

265K / High mobo -> 650€ 50W gaming

5070ti -> 950€ on Sale 170W gaming

220W total

For 1600€ you gould get the 85/90% performance of a 4090 and a 14900KS in Il-2 and using half the power. Nice!

Kind regards,

September 16

I have a 5070ti coming in soon and noticed that I hadn't run the benchmark with my latest build, so here's the info with my current build.

Game version 6.002

Windows 11

Motherboard: Gigabyte X670 Gaming X AX V2
CPU: Ryzen 7 7800X3D
CPU Freq: 4.8 Ghz
Cores: 8 (number of active physical Cores)
Threads: 16
RAM size: 32Gb (2x16GB)
RAM Freq: 3000 MHz
NB Freq: 3000 MHz
RAM timings: 36-36-36-96
GPU: 3080 ti

CPU Test 1080

Frames: 8337 - Time: 60000ms - Avg: 138.950 - Min: 117 - Max: 176

VR Test#1

VR Headset: Varjo Aero

Frames: 4683 - Time: 60000ms - Avg: 78.050 - Min: 68 - Max: 90

VR Test#2

Frames: 4772 - Time: 60000ms - Avg: 79.533 - Min: 68 - Max: 90

Will run with the 5070ti probably in a day or two.

Edited September 16 by Voxman

September 16

4 hours ago, Voxman said:

VR Headset: Varjo Aero

Frames: 4683 - Time: 60000ms - Avg: 78.050 - Min: 68 - Max: 90

VR Test#2

Frames: 4772 - Time: 60000ms - Avg: 79.533 - Min: 68 - Max: 90

Thank you for these tests with the Aero.

From there we can observe the following:

- Both VR tests (1&2) obtain the same result, this means that in both cases the bottleneck is the CPU and the GPU still has margin to be fully loaded.

- When you will run the tests again with your new 5070Ti, you will obtain very similar numbers in all (CPU test and VR1 and 2), but having more margin in the GPU load.

- You new 5070Ti will for sure allow you to do more supersampling in your Aero (or MSAA usage if needed).

- You RAM frequency is 3000Mhz which is a bit low for baing a DDR5. Did you enabled AMD Expo in the BIOS for faster profiles? This could improve all your tests (CPU and VR1&2)

September 17

19 hours ago, chiliwili69 said:

You RAM frequency is 3000Mhz which is a bit low for baing a DDR5. Did you enabled AMD Expo in the BIOS for faster profiles? This could improve all your tests (CPU and VR1&2)

Thanks Chili. To answer your question, here's screen shot of CPUZ memory section. I'm running 32G (16G x 2) modules, and have set EXPO settings in the BIOS.

Edited September 17 by Voxman

September 17

6 minutes ago, Voxman said:

here's screen shot of CPUZ memory section.

ahh! that explain well the numbers.

Normally I put the frequency (3000 MHz) multiplied by 2 since it runs in dual channel. So, I will correct the table and will put 6000MHz. Tha´s why it is called DDR5-6000 by RAM manufacturers.

I assume you use slots #4 and #2 for the RAM in Dual channel (that is DDR5_A2 and DDR5_B2):

September 17

22 minutes ago, Voxman said:

here's screen shot of CPUZ memory section

but could you show also the Memory Tab of the CPU-Z?

September 17

3 hours ago, chiliwili69 said:

but could you show also the Memory Tab of the CPU-Z?

They're in slots #2 and #4.

I appreciate any suggestions. I've tried to OC memory in the past, but really never got a good handle, or guide...plus time.

September 18

21 hours ago, chiliwili69 said:

Normally I put the frequency (3000 MHz) multiplied by 2 since it runs in dual channel.

Dual channel has nothing to do with why the data rate of DDR is double that of the frequency. Most motherboards have two separate connections (= channels) to the RAM, where the RAM sticks in slot 1 and 2 are on one channel, and slot 3 and 4 are on the other channel. That is why, if you use two RAM sticks, you should usually put them in slots 2 and 4, so they are on different channels.

The RAM can then be used more efficiently, where when something (big) gets stored in RAM, it is send to two different RAM sticks at the same time, on different channels. The same for reading from the RAM.

However, that doesn't cause the data to be transferred faster than the frequency, it means that you can use two RAM sticks at the same time.

The reason why the data rate is twice the frequency for DDR, is that data is sent both when the clock signal goes up and when it goes down again. The clock signal is what computers use to make sure that they agree on when the data is available to be read. Sending a bit requires the sender to raise, lower or keep constant, the signal on the data line. This takes time. The key to having fast memory is to tune it so the correct signal is on the line just at the moment that the signal is being read, and then right away, the sender can change the signal again for the next bit. This requires careful coordination with the side that is reading the signal, as they need to read it at exactly the right time. The clock signal tells the sender when they need to make the signal available and tells the reader when they need to read the signal.

However, the clock signal itself is a signal that changes from 0 to 1 and from 1 to 0. Traditionally, the clock signal would only trigger both parties on a switch from 0 to 1, but for DDR, they figured out a way to trigger them also on the transition from 1 to zero:

@Voxman

Overclocking RAM is generally not worth it anymore, unless you do it as a hobby. The gain is usually extremely limited for how much effort it is and the vague instability issues you risk.

This is true in particular for X3D-CPUs that are extremely insensitive to RAM speeds, since they have such a big cache. So it is very rare that they don't have a crucial bit of data in the cache and are restrained by the RAM speed.

September 18

53 minutes ago, Aapje said:

This is true in particular for X3D-CPUs that are extremely insensitive to RAM speeds, since they have such a big cache. So it is very rare that they don't have a crucial bit of data in the cache and are restrained by the RAM speed.

Thanks, I didn´t thought on that. In fact, I have a 7800X3D and my RAM is TridentZ5 Neo (https://www.gskill.com/product/165/390/1662622003/F5-6000J3238F16GX2-TZ5NR) which has an AMD Expo profile of DDR5-6000 CL32-38-38-96 1.35V but this gives me crashes when activated. So I run it with a deafult profile 48000 CL-40-40-40-77. So, it is good to know that those CPUs are less sensible for speeds.

September 18

7 hours ago, Aapje said:

Overclocking RAM is generally not worth it anymore, unless you do it as a hobby. The gain is usually extremely limited for how much effort it is and the vague instability issues you risk.

This is true in particular for X3D-CPUs that are extremely insensitive to RAM speeds, since they have such a big cache. So it is very rare that they don't have a crucial bit of data in the cache and are restrained by the RAM speed.

I didn't know that, but you are totally correct. Did an IL2 CPU test with stock 4800MT and OC 6200MT timings and the results are basically the same..

September 19

12 hours ago, DBCOOPER011 said:

Did an IL2 CPU test with stock 4800MT and OC 6200MT timings and the results are basically the same..

Very good test! thanks for sharing!

SYN_Vander BENCHMARK v6 to measure IL-2 performance in monitor & VR

Recommended Posts

1Sascha

1Sascha

chiliwili69

chiliwili69

1Sascha

1Sascha

tati

chiliwili69

FG28_THE_ERASER

chiliwili69

tati

Crockett

chiliwili69

chiliwili69

Aapje

Crockett

FG28_THE_ERASER

chiliwili69

BladeMeister

chiliwili69

chiliwili69

LuftManu

chiliwili69

chiliwili69

LuftManu

Voxman

chiliwili69

Voxman

chiliwili69

chiliwili69

Voxman

Aapje

chiliwili69

DBCOOPER011

chiliwili69

Create an account or sign in to comment

Create an account

Sign in

Browse All

Activity