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

[dburne]

5 hours ago, chiliwili69 said:

 

They run at 3600MhZ but with 14-14-14-34 latencies. So they could be a good upgrade for future. 2x8Gb are enough for IL-2 VR.

 

Only for IL-2 VR, any other flight sims forget it. I would not put less than 32GB in a system today. Also as IL-2 continues to advance with features it will use more as time goes on. I have 32 GB now and am glad I do considering the sims I use, my next build will have 64 GB of ram. Heck I have 24 GB in just my GPU alone.

Edited April 21, 2022 by dburne

[-332FG-Cue-Ball]

14 hours ago, dburne said:

 

Only for IL-2 VR, any other flight sims forget it. I would not put less than 32GB in a system today. Also as IL-2 continues to advance with features it will use more as time goes on. I have 32 GB now and am glad I do considering the sims I use, my next build will have 64 GB of ram. Heck I have 24 GB in just my GPU alone.

Yeah, this is my thought as well. Not sure I'm too keen on spending $150 on more RAM to replace the perfectly good memory I just bought a year ago. I'd love an extra 10 fps, but that's a tough pill to swallow after just spending a grand on a new video card.

[dburne]

5 minutes ago, -332FG-Cue-Ball said:

Yeah, this is my thought as well. Not sure I'm too keen on spending $150 on more RAM to replace the perfectly good memory I just bought a year ago. I'd love an extra 10 fps, but that's a tough pill to swallow after just spending a grand on a new video card.

 

Yeah I have 32 GB of 3200 MHz cl 14 ram, I will likely not replace it on this z390 platform I am running on now, Will wait till I do a new build.

[A_radek]

Game version: 4.704

 Motherboard:   Asus TUF B550 Pro
 CPU:                 Ryzen 5 5600X
 CPU Freq:        4.65 Ghz
 Cores:               6
 Threads:           12
 RAM size:        32Gb (2x16GB)
 RAM Freq:       3600
 NB Freq:          1800 MHz
 RAM timings:  16-16-16-36
 GPU:                3080
 

2022-04-21 20:03:20 - Il-2
Frames: 6648 - Time: 60000ms - Avg: 110.800 - Min: 97 - Max: 158
 

 Motherboard:   Asus TUF B550 Pro
 CPU:                 Ryzen 5 5800X3D
 CPU Freq:        4.45 Ghz
 Cores:               6
 Threads:           12
 RAM size:        32Gb (2x16GB)
 RAM Freq:       3600
 NB Freq:          1800 MHz
 RAM timings:  16-16-16-36
 GPU:                3080

2022-04-21 21:30:38 - Il-2
Frames: 7471 - Time: 60000ms - Avg: 124.517 - Min: 109 - Max: 166
 

Notes:
same bios version in both tests.
Running win 11.
Noticed a cpu temp increase of around 6-7c during the benchmark switching to the 5800x3d.
While the 5600x bench-runs produce very consistent results, the 5800x3d results fluctuate and produced anything from 122 to 127fps averages.

It's an upgrade. Though no dramatic performance-increase as in some other large world titles, at least it didn't perform worse than the old 5600x due to the lower frequency. As it has done in some games.

@shirazjohn, looking forward to your results. x570 might make a difference. Also, noticed on the benchmark list you clock the snot out of your ram. I'm running a b-die F4-3600C16D-32GTZN kit, though only enable dcp on that, with no other tweaks in bios. Very curious how your plenty faster ram affects results.

 

Edited April 21, 2022 by SvAF/F16_radek

[shirazjohn]

Have run some fairly quick tests and even though an improvement over my 5600x its not as good as I had hoped but I've not played with the memory settings to much yet so maybe some improvement there. Hopefully will have some more time to test over the weekend.

 

5600x version 4.704 pbo + 200mhz mem 4000 mhz 16-16-16-16-30

2022-04-19 19:08:58 - Il-2
Frames: 7139 - Time: 60000ms - Avg: 118.983 - Min: 105 - Max: 163

2022-04-19 19:11:09 - Il-2
Frames: 7249 - Time: 60000ms - Avg: 120.817 - Min: 105 - Max: 171

2022-04-19 19:13:11 - Il-2
Frames: 7217 - Time: 60000ms - Avg: 120.283 - Min: 105 - Max: 16

 

After the last hotfix I lost around 5ps average but i believe this is has been the same for others (maybe due to the banding fix or sky update).

 

5800x3d version 4.704 xmp profile 3600 mhz 14-15-15-15-35
2022-04-21 18:16:14 - Il-2
Frames: 7704 - Time: 60000ms - Avg: 128.400 - Min: 111 - Max: 179

2022-04-21 18:18:31 - Il-2
Frames: 7898 - Time: 60000ms - Avg: 131.633 - Min: 114 - Max: 185

2022-04-21 18:20:40 - Il-2
Frames: 7639 - Time: 60000ms - Avg: 127.317 - Min: 110 - Max: 179

 

1 hour ago, SvAF/F16_radek said:

Noticed a cpu temp increase of around 6-7c during the benchmark switching to the 5800x3d.
While the 5600x bench-runs produce very consistent results, the 5800x3d results fluctuate and produced anything from 122 to 127fps averages.

As you say temps up a little and a bigger fluctuation in results.

Edited April 21, 2022 by shirazjohn

[chiliwili69]

15 hours ago, SvAF/F16_radek said:

Frames: 6648 - Time: 60000ms - Avg: 110.800

 

Thank you for these tests. The first one with the 5600X you obtain exactly the same performance than me with version 4.703, so I assume that there should not be a major difference.

 

Regarding the 5800X3D I am quite impressed with the extra 14 fps you got. I was really expecting much less. Those extra fps should help as well in both VR tests, since VR is sooner bottlenecked by CPU in dense scenarios.

15 hours ago, SvAF/F16_radek said:

CPU:                 Ryzen 5 5800X3D
 CPU Freq:        4.45 Ghz
 Cores:               6
 Threads:           12

 

I suposse you mean 8 cores 16 threads (or did you disabled some cores?)

15 hours ago, shirazjohn said:

After the last hotfix I lost around 5ps average but i believe this is has been the same for others

 

Many thanks for those tests. Exactly you also had that 5fps drop in the CPU test, confirming the drop introduced with version 4.703.

15 hours ago, shirazjohn said:

xmp profile 3600 mhz 14-15-15-15-35

 

Most likely you had less than the other radek test because you changed memory. I believe you will get same gain if you are able to run your new CPU with the same speed and latency than previously.

[A_radek]

18 hours ago, shirazjohn said:

Have run some fairly quick tests and even though an improvement over my 5600x its not as good as I had hoped but I've not played with the memory settings to much yet so maybe some improvement there. Hopefully will have some more time to test over the weekend.

 

5600x version 4.704 pbo + 200mhz mem 4000 mhz 16-16-16-16-30

2022-04-19 19:08:58 - Il-2
Frames: 7139 - Time: 60000ms - Avg: 118.983 - Min: 105 - Max: 163

2022-04-19 19:11:09 - Il-2
Frames: 7249 - Time: 60000ms - Avg: 120.817 - Min: 105 - Max: 171

2022-04-19 19:13:11 - Il-2
Frames: 7217 - Time: 60000ms - Avg: 120.283 - Min: 105 - Max: 16

 

After the last hotfix I lost around 5ps average but i believe this is has been the same for others (maybe due to the banding fix or sky update).

 

5800x3d version 4.704 xmp profile 3600 mhz 14-15-15-15-35
2022-04-21 18:16:14 - Il-2
Frames: 7704 - Time: 60000ms - Avg: 128.400 - Min: 111 - Max: 179

2022-04-21 18:18:31 - Il-2
Frames: 7898 - Time: 60000ms - Avg: 131.633 - Min: 114 - Max: 185

2022-04-21 18:20:40 - Il-2
Frames: 7639 - Time: 60000ms - Avg: 127.317 - Min: 110 - Max: 179

 

As you say temps up a little and a bigger fluctuation in results.

 

Thanks! Seems memoryspeed produces a significant performance increase still. Thought gains would be smaller considering the larger cache. I’ll test some more this weekend. Seems latest ryzen master also supports ”memory boost” (whatever that is) for the 5800x3d. Though unfamilliar with that software.

 

4 hours ago, chiliwili69 said:

 

 

I suposse you mean 8 cores 16 threads (or did you disabled some cores?)

 

 

Yes sorry. Bad copy/paste. 8c 16t. No tweaks in bios other than enabling memory dcp.

[shirazjohn]

17 hours ago, SvAF/F16_radek said:

Thanks! Seems memoryspeed produces a significant performance increase still. Thought gains would be smaller considering the larger cache. I’ll test some more this weekend.

Hi i done a little more testing last night and higher frequencies / tighter timings doesn't seem to have as much of a performance increase as it did with my 5600x , although now I've changed my memory settings to what i ran the at 5600x at (4000 mhz 16 16 16 30) i am getting more consistent results at around 130 fps average.

Edited April 23, 2022 by shirazjohn

[A_radek]

15 minutes ago, shirazjohn said:

Hi i done a little more testing last night and higher frequencies / tighter timings doesn't seem to have as much of a performance increase as it did with my 5600x , although now I've changed my memory settings to what i ran the at 5600x at (4000 mhz 16 16 16 30) i am getting more consistent results at around 130 fps average.

Tried setting memory to 3600 mhz 14-15-15-15-35 as in your first bench and confirmed in cpu-z it was running like that. But, probably doing something wrong as I’m only getting ~1fps increase if at all. Hard to tell with the fluctuating results.

 

Good to hear of your progress!

Was eyeing a 4000 cl16 kit yesterday and now tempted. Wondering though if you managed to run infinity att 2000mhz when running the 4000 kit? That skatterbencher fellow mentioned he couldn’t get it stable past 1900mhz.

[shirazjohn]

31 minutes ago, SvAF/F16_radek said:

Was eyeing a 4000 cl16 kit yesterday and now tempted. Wondering though if you managed to run infinity att 2000mhz when running the 4000 kit? That skatterbencher fellow mentioned he couldn’t get it stable past 1900mhz.

I've been running with inf fabric at 2000 mhz with no problem at all, i don't know if the chipset makes a difference x570 v b550. I have been running at 4000 mhz for probably a year and its been rock solid.

 

This is the memory I have I'm no expert just bought what i thought was right at the time.

https://www.overclockers.co.uk/team-group-8pack-ripped-edition-16gb-2x8gb-ddr4-pc4-28800c14-3600mhz-dual-channel-kit-black-my-001-8p.html

Edited April 23, 2022 by shirazjohn

[shirazjohn]

On 4/22/2022 at 12:36 PM, chiliwili69 said:

Most likely you had less than the other radek test because you changed memory. I believe you will get same gain if you are able to run your new CPU with the same speed and latency than previously.

Hi Chilli ran the test again with the same mem settings as the 5600x, I had a slight performance improvement with more consistent results but not the gains as with the 5600x. Also ran the Vr test at 100%ss and have about 5 fps improvement, I previously ran the test post version 4.703 with my 5600x and had an average of 63fps which was also down on pre 4.703 v tests (unfortunately deleted the data by accident). 

 

Dram 4000 mhz 16-16-16-16-30

2022-04-22 21:41:19 - Il-2
Frames: 7851 - Time: 60000ms - Avg: 130.850 - Min: 113 - Max: 185

2022-04-22 21:43:12 - Il-2
Frames: 7836 - Time: 60000ms - Avg: 130.600 - Min: 114 - Max: 179

2022-04-22 21:45:01 - Il-2
Frames: 7786 - Time: 60000ms - Avg: 129.767 - Min: 112 - Max: 177

2022-04-22 21:46:54 - Il-2
Frames: 7841 - Time: 60000ms - Avg: 130.683 - Min: 113 - Max: 181

Average 130.475

Vr test reverb G2 100% ss  gtx 1080ti

2022-04-22 21:55:28 - Il-2
Frames: 4082 - Time: 60000ms - Avg: 68.033 - Min: 53 - Max: 79

2022-04-22 21:57:29 - Il-2
Frames: 4052 - Time: 60000ms - Avg: 67.533 - Min: 55 - Max: 79

[chiliwili69]

5 hours ago, shirazjohn said:

Vr test reverb G2 100% ss  gtx 1080ti

2022-04-22 21:55:28 - Il-2
Frames: 4082 - Time: 60000ms - Avg: 68.033 - Min: 53 - Max: 79

2022-04-22 21:57:29 - Il-2
Frames: 4052 - Time: 60000ms - Avg: 67.533 - Min: 55 - Max: 79

 

Thanks again for this extra test!  It seems your previous RAM profile give you more consistency in the results.

 

Regarding the VRtest2 (at 100%) you are clearly GPU limited, specially with latest versions which increase the GPU load.

 

If you want to really see your fps boost in VR you will need to run the VR test1 (50%), where your GPU is more relaxed and we measure CPU performance.

 

You reported here you VR numbers in previous version:  

 

The number are in the table in row 57.  Nothing is fogotten... ?

 

 

 

[Voyager]

Pre-teardown tests:

 

Configuration 1: Ryzen 5800X with 3080 Ti, and DDR4-3600
 Motherboard:    Gigabyte X570 Aorus Xtreme v 1.2
 Bios Version F36a

 CPU:        AMD Ryzen 7 5800X
 CPU Freq:    4450 MHz
 L3 cache:    2x16 MB
 Cores:        8
 Threads:    16

 RAM type:    DDR4
 RAM size:    128 GB (4x32)
 NB Freq:    1799.6 MHz
 RAM Freq:    1799.6
 RAM timings:    16-22-22-42
 
GPU:        EVGA 3080 TI FTW3, Core 1800Mhz, Memory 9501Mhz (Factory overclock)
 Driver Version 512.15

VR Headset 1: Vive Pro 2
VR Headset 2: HP  ReverbG1

Game Version 4.704

 

	VR Test: Current Settings VP2 (Extreme 2868x2868)
	Min	Max	Avg
1	37	61	50.867
2	38	61	49.983
3	37	61	51.2
	37.33333	61	50.68333
	CPU Test
	Min	Max	Avg
1	97	161	112.767
2	100	165	114.867
3	102	167	117.05
	99.66667	164.3333	114.8947
	Psuedo GPU Test (5120x1440)
	Min	Max	Avg
1	115	196	155.283
2	119	197	156.917
3	118	194	156.433
	117.3333	195.6667	156.211
	GPU Test
	Min	Max	Avg
1	104	175	139.067
2	106	162	140.7
3	106	171	140.117
	105.3333	169.3333	139.9613
	VR Test 1 Vive Pro 2: (Ultra: 2188x2188)
	Min	Max	Avg
1	42	91	59.467
2	42	91	56.583
3	42	91	59.95
	42	91	58.66667
	VR Test 2 Vive Pro 2: (Ultra: 3128x3128)
	Min	Max	Avg
1	42	46	44.933
2	42	46	44.933
3	42	46	44.95
	42	46	44.93867
	VR Test 1 Reverb G1 (2208x2165)
	Min	Max	Avg
1	56	91	75.233
2	57	91	75.8
3	57	91	76.617
	56.66667	91	75.88333
	VR Test 2 Reverb G1 (3124x3056)
	Min	Max	Avg
1	54	91	72.7
2	54	91	72.667
3	56	91	74.633
	54.66667	91	73.33333

 

 

Configuration 1: Ryzen 5800X with 3080 Ti, and DDR4-2666

DDR4-2666

Timings: 20-19-19-43

 

	5800X DDR 2133
	VR Test: Current Settings VP2 (Extreme 2868x2868)
	Min	Max	Avg
1	37	61	44.95
2	37	61	45.217
3	38	61	44.917
	37.33333	61	45.028
	CPU Test
	Min	Max	Avg
1	93	151	107.1
2	94	159	109.7
3	94	158	109.117
	93.66667	156	108.639
	GPU Test
	Min	Max	Avg
1	113	198	151.217
2	116	196	152.9
3	119	192	153.633
	116	195.3333	152.5833
	GPU Test
	Min	Max	Avg
1	102	167	139.733
2	108	181	142.267
3	105	167	142.5
	105	171.6667	141.5
	VR Test 1 Vive Pro 2: (Ultra: 2188x2188)
	Min	Max	Avg
1	42	86	47.75
2	42	89	49.711
3	42	91	49.5
	42	88.66667	48.987
	VR Test 2 Vive Pro 2: (Ultra: 3128x3128)
	Min	Max	Avg
1	43	46	44.95
2	39	46	44.9
3	42	46	44.95
	41.33333	46	44.93333
	VR Test 1 Reverb G1 (2208x2160)
	Min	Max	Avg
1	52	92	64.55
2	52	91	66.017
3	53	91	64.9
	52.33333	91.33333	65.15567
	VR Test 2  Reverb G1 3124x3056
	Min	Max	Avg
1	50	91	61.367
2	50	90	59.783
3	51	91	62.783
	50.33333	90.66667	61.311

 

 

After I've got the full set, I'll attached the spreadsheet as a *.txt csv so it's less messy to extract

[chiliwili69]

11 hours ago, Voyager said:

Pre-teardown tests

Many thanks for this detailed tests. Are you going to upgrade?

 

Everything as expected, nothing weird at all.

 

I suposse the reason of having lower values in the VP2 are due to being only at 45 or 90, but nothing in between. right?  This will be seen in the frametimes csv produced by fraps.

[thermoregulator]

What frequencies do you guys get with 5800x3d? I get maximum 4450 Mhz on one core during gameplay and only 4300 Mhz on all core load. I also have considerable lower scores in benchmarks than they had in most reviews. Temps are quite high too... I am considering reverting back to 5 Ghz 5900x, even though 5800x3d "feels" little faster in this game...

[A_radek]

1 hour ago, thermoregulator said:

What frequencies do you guys get with 5800x3d? I get maximum 4450 Mhz on one core during gameplay and only 4300 Mhz on all core load. I also have considerable lower scores in benchmarks than they had in most reviews. Temps are quite high too... I am considering reverting back to 5 Ghz 5900x, even though 5800x3d "feels" little faster in this game...

Also getting 4450 single core frequency. Haven't checked all core.

Understandable. Seems this isn't one of those games that thrives out of the massive cache. But yes, what I don't get if talking about "feelings" is that it just seems to perform so much better in heavy scenarios than the 14fps cpu-test increase I got switching from the 5600x would imply. Testing last night, my regular settings bumped from medium to ultra shadows and on a very populated airquake server, I was surprised not to dip below 80 hz. And staying well below the 12.5ms required. Even with a squad taking of simultaneously and furballs above.

I'll have to vr benchmark and compare to chilis 5600x/b550/valve index result. 

Edited April 25, 2022 by SvAF/F16_radek

[thermoregulator]

3 minutes ago, SvAF/F16_radek said:

Also getting 4450 single core frequency. Haven't checked all core.

Understandable. Seems this isn't one of those games that thrives out of the massive cache. But yes, what I don't get if talking about "feelings" is that it just seems to perform so much better in heavy scenarios than the 14fps cpu-test increase I got switching from the 5600x would imply. Testing last night, my regular settings bumped bumped from medium to ultra shadows and on a very populated airquake server, I was surprised not to dip below 80 hz. And staying well below the 12.5ms required. Even with a squad taking of simultaneously and furballs above.

I'll have to vr benchmark and compare to chilis 5600x result. 

Yes, the game is much smoother on 5800x3d. Even on dense career missions there are just few dips bellow 90 Hz. Thats something I wasn't able to achieve since migration to Win 11 (which reduced performance significantly on my system). Sadly, I can't enjoy it much, because my new 3090 ti is not compatible with Reverbs G2 V2 cable, and my old v1 cable is in terrible shape causing random occasional black screens... I can't even count all the cable issues I had with those Reverbs...

[shirazjohn]

3 hours ago, thermoregulator said:

What frequencies do you guys get with 5800x3d? I get maximum 4450 Mhz on one core during gameplay and only 4300 Mhz on all core load. I also have considerable lower scores in benchmarks than they had in most reviews. Temps are quite high too... I am considering reverting back to 5 Ghz 5900x, even though 5800x3d "feels" little faster in this game...

I am the same max 4450mhz but haven't checked all core. I also feel gameplay is much smoother have bumped frontline activity up to dense and still seems smoother than before even though I'm seriously bottlenecked by my 1080ti. Thought it may have been a placebo until you guys mentioned  it.

[R3animate]

I think the "feeling" smoother is a result of better 0.1% lows and 1% lows -- it's probably helping with some stutter here and there, even if the game isn't seeing huge gains. I don't think I can justify an upgrade from my 5800X :).

 

HOWEVER, this bodes well for zen4 assuming they do a 3D-cache version there too - it might push Intel to do the same, and then everybody wins.

[A_radek]

52 minutes ago, R3animate said:

I think the "feeling" smoother is a result of better 0.1% lows and 1% lows -- it's probably helping with some stutter here and there, even if the game isn't seeing huge gains. I don't think I can justify an upgrade from my 5800X :).

 

HOWEVER, this bodes well for zen4 assuming they do a 3D-cache version there too - it might push Intel to do the same, and then everybody wins.

Yes that could be it. Could also be the latest patch smoothing things out coinciding with my cpu change. Not sure if anyone not having changed cpu "feels" the same. I'd also stick with the vanilla 5800x for now. But had to try, and it's doing remarkably in other games.

[shirazjohn]

4 hours ago, thermoregulator said:

Temps are quite high too... I am considering reverting back to 5 Ghz 5900x, even though 5800x3d "feels" little faster in this game...

Temps for me are not much more than 5600x , I'm getting around 55 degrees during gameplay. 

[Voyager]

6 hours ago, chiliwili69 said:

Many thanks for this detailed tests. Are you going to upgrade?

 

Everything as expected, nothing weird at all.

 

I suposse the reason of having lower values in the VP2 are due to being only at 45 or 90, but nothing in between. right?  This will be seen in the frametimes csv produced by fraps.

Yes. Got a 5800X3D and am doing a before/after on my different flight sims. Haven't gotten through all the tests yet, so may be later this week before I can actually get the new chip in.

[RAAF492SQNOz_Steve]

On 4/25/2022 at 11:56 PM, Voyager said:

Yes. Got a 5800X3D and am doing a before/after on my different flight sims. Haven't gotten through all the tests yet, so may be later this week before I can actually get the new chip in.

 

Any update on how the 5800x3D is going compared to the 5800X?

[shirazjohn]

Hi guys i have been running with Open Xr the last few days and have been quite impressed, spotting and id-ing seem improved along with clarity and a reduction in shimmering  so decided to run the benchmark to compare open XR with steam Vr and it appears there is a slight improvement with Open Xr.

 

Processor 5800X3D, Gpu 1080ti

 

Vr test reverb G2 100% ss (Steam Vr) 

Frames: 4082 - Time: 60000ms - Avg: 68.033 - Min: 53 - Max: 79

Frames: 4052 - Time: 60000ms - Avg: 67.533 - Min: 55 - Max: 79

 

Vr Test G2 50% ss (Steam Vr)

Frames: 5297 - Time: 60000ms - Avg: 88.283 - Min: 77 - Max: 91

Frames: 5228 - Time: 60000ms - Avg: 87.133 - Min: 72 - Max: 91

 

 

Vr test Reverb G2 100% ss (Open XR)
Frames: 4231 - Time: 60000ms - Avg: 70.517 - Min: 57 - Max: 80

Frames: 4215 - Time: 60000ms - Avg: 70.250 - Min: 58 - Max: 80

 

Vr Test Reverb G2 50% ss (Open XR)

Frames: 5294 - Time: 60000ms - Avg: 88.233 - Min: 76 - Max: 91

Frames: 5300 - Time: 60000ms - Avg: 88.333 - Min: 77 - Max: 91
 

[-332FG-Cue-Ball]

Re-ran the two non-VR tests again after upgrading my motherboard from an older BIOS (v2006, released last year) to the newest stable release (v2604). The only thing I changed was to upgrade the BIOS, then go back and reset the built-in overclocking settings. Unfortunately, this isn't quite an apples-to-apples test since there was a new version of IL-2 released between my runs. However, I saw a minor improvement in both tests. I suspect this is due to the BIOS and not the minor changes in the game.

 

CPU Test in 1080p:
BIOS v.2006: Frames: 6049 - Time: 60000ms - Avg: 100.817 - Min: 87 - Max: 139

BIOS v.2604: Frames: 6349 - Time: 60000ms - Avg: 105.817 - Min: 94 - Max: 152

 

GPU Test in 4K:
BIOS v.2006: Frames: 7601 - Time: 60000ms - Avg: 126.683 - Min: 90 - Max: 148

BIOS v.2604: Frames: 7603 - Time: 60000ms - Avg: 126.717 - Min: 94 - Max: 159

[chiliwili69]

17 hours ago, shirazjohn said:

Vr Test G2 50% ss (Steam Vr)

Frames: 5297 - Time: 60000ms - Avg: 88.283 - Min: 77 - Max: 91

Frames: 5228 - Time: 60000ms - Avg: 87.133 - Min: 72 - Max: 91

 

Good to see OpenXR doesn´t impact performance (even improve it a bit).

 

On the VR test1 (50% SS, so not constrained by GPU) you obtain same values than your previous 5800X, which they were already very good (87fps).

Being so close to 90fps limit, this VR test is not suitable to see improvements. I am thinking to perhaps raise the bar of the settings for the VR test.

[Voyager]

Since there were a bunch of driver updates between when I first ran the benchmarks and now, I re-reran the CPU and VP2 VR Test 1

 

Configuration 1: Ryzen 5800X with 3080 Ti, and DDR4-3600
 Motherboard:    Gigabyte X570 Aorus Xtreme v 1.2
 Bios Version F36a

 CPU:        AMD Ryzen 7 5800X
 CPU Freq:    4450 MHz
 L3 cache:    2x16 MB
 Cores:        8
 Threads:    16

 RAM type:    DDR4
 RAM size:    128 GB (4x32)
 NB Freq:    1799.6 MHz
 RAM Freq:    1799.6
 RAM timings:    16-22-22-42
 
GPU:        EVGA 3080 TI FTW3, Core 1800Mhz, Memory 9501Mhz (Factory overclock)
 Driver Version 512.59

VR Headset 1: Vive Pro 2
Game Version 4.704

 

	CPU Test
	Min	Max	Avg
1	98	161	112.567
2	94	155	110.33
3	95	158	109.633
	95.66666667	158	110.8433333

 

	VR Test 1 Vive Pro 2: (Ultra: 2200x2200)
	Min	Max	Avg
1	43	90	53.467
2	42	90	58.583
3	42	91	60.533
	42.33333333	90.33333333	57.52766667

 

Configuration 2: Ryzen 5800X3D with 3080 Ti, and DDR4-2666

DDR4-2666

Timings: 20-19-19-43

 

	CPU Test
	Min	Max	Avg
1	106	167	120.5
2	107	165	121.933
3	108	165	122.183
	107	165.6667	121.5386667
	Psuedo GPU Test (5120x1440)
	Min	Max	Avg
1	117	197	156.833
2	116	189	157.233
3	121	189	157.55
	118	191.6667	157.2053333
	GPU Test
	Min	Max	Avg
1	105	176	143.917
2	109	180	144.3
3	106	179	144
	106.6667	178.3333	144.0723333
	VR Test 1 Vive Pro 2: (Ultra: 2200x2200)
	Min	Max	Avg
1	69	91	85.3
2	65	91	82.467
3	62	91	82.667
	65.33333	91	83.478
	VR Test 2 Vive Pro 2: (Ultra: 3148x3148)
	Min	Max	Avg
1	42	89	48.55
2	42	82	46
3	42	90	47.667
	42	87	47.40566667

 

 

Configuration 2: Ryzen 5800X3D with 3080 Ti, and DDR4-3600

DDR4-3600

Timings: 16-22-22-42

 

	CPU Test
	Min	Max	Avg
1	109	173	125.2
2	106	166	123.133
3	109	168	124.55
	108	169	124.2943333
	Psuedo GPU Test (5120x1440)
	Min	Max	Avg
1	118	196	156.95
2	118	179	150.067
3	114	177	149.333
	116.6667	184	152.1166667
	GPU Test
	Min	Max	Avg
1	106	183	143.9
2	109	180	144.2
3	110	169	143.9
	108.3333	177.3333	144
	VR Test 1 Vive Pro 2: (Ultra: 2188x2188)
	Min	Max	Avg
1	65	91	84.8
2	64	91	83.767
3	63	91	83.483
	64	91	84.01666667
	VR Test 2 Vive Pro 2: (Ultra: 3128x3128)
	Min	Max	Avg
1	42	88	48.3
2	42	91	52.333
3	42	91	50.883
	42	90	50.50533333

 

I've attached the spreadsheet as a csv file. Just rename the extension to .csv and you should be able to open it directly in any comma separated file.

 

Basic findings; this is about 7% faster than the 5800X with DDR 3600 ram, and 12% faster if you are only running stock DDR4, and 2D mode. VR, however, seems to get a very big lift, and seems to show almost no performance difference between DDR4-2666 and DDR4-3600. Its a big enough difference that I'm even seeing a lift in the VR Test 2.

 

And this does seem to be holding in the other flight sims I've checked so far. DCS in the F-14B cold start, CPU Frame times went from 9.2 median to 7.8ms with stock DDR4 to 7.2ms with DDR4-3600, so a nrea27% improvement in the CPU frametime. Now, I'm still GPU limited, but some big honking GPUs are coming out later this year, so that's a problem that may solve itself.

 

There are a couple of areas I'm getting some unexpected results. For the settings I've been using, I'm actually get a reduced average FPS, from about 50 fps to 44.9. However, it looks like I'm getting capped at 46fps, whereas previously I was getting 61 fps, but I'm also getting about 12% better minimum frame rates too. I suspect the update to the VP2 software may have done something with the framerate locking.

 

Now, some of my other CPU benchmarks have dropped, but I'm think for VR, the extra cache seems to really be doing to job, especially if you only have basic DDR4 ram. Overall, I'm actually very impressed.

Il-2 5800X3D Testing.txt

[chiliwili69]

On 5/9/2022 at 2:44 AM, Voyager said:

VR, however, seems to get a very big lift,

Thank you for these detailed tests. Very clear all!. This really take some time to do it.

 

On the CPU test you got exatly the same than Radek (+14fps).

But in the VRtest1 (where the CPU is really evaluated) you get +26fps, in VR this is a huge difference! Only changing the 5800X by a 5800X3D.

This means you will be less bottlenecked by CPU in dense object scenarios.

 

Unfortunately it seems that the VP2, for the same amount of pixels, doesn´t perform as well as the G2. So you are then constrained by GPU (and you have a 3080Ti!!).

It is surprising the low performance of the HTC middle software, comparing it to G2 and Index.

 

What do you use for IL-2, G2 or VP2?

 

I see in the csv that the test for G2 are empty, are you going to run the G2 test as well?, I suposse that with the 5800X3D and the G2 you will be closer to 90 in the VR tests.

[Voyager]

@chiliwili69 I use the VP2 for gaming. The Reverb G1 is simply to small and has to narrow an IPD to fit my head, but I recalled that the VP2 software caps frame rates so hooked up the G1 for comparative testing. 

 

I can go run the G1 tests with the new CPU tonight. 

 

Also need to dig around in the VP2 settings. The middleware does have the frame rate locks in them. I think I'd disabled some of them in my original settings, but they get re-enabled every time their software updates. It's helping! 

 

But, yeah, for VR, this seems to be a huge improvement. Definitely curious how this will stack up again Raptor Lake and Zen 4 next year, but I almost suspect it will still be competitive until either the base DirectX engine changes, or Vcache Zen 4 hits. 

[Voyager]

12 hours ago, chiliwili69 said:

Thank you for these detailed tests. Very clear all!. This really take some time to do it.

 

On the CPU test you got exatly the same than Radek (+14fps).

But in the VRtest1 (where the CPU is really evaluated) you get +26fps, in VR this is a huge difference! Only changing the 5800X by a 5800X3D.

This means you will be less bottlenecked by CPU in dense object scenarios.

 

Unfortunately it seems that the VP2, for the same amount of pixels, doesn´t perform as well as the G2. So you are then constrained by GPU (and you have a 3080Ti!!).

It is surprising the low performance of the HTC middle software, comparing it to G2 and Index.

 

What do you use for IL-2, G2 or VP2?

 

I see in the csv that the test for G2 are empty, are you going to run the G2 test as well?, I suposse that with the 5800X3D and the G2 you will be closer to 90 in the VR tests.

 

 

G1 Testing: 5800X3D DDR 3600 config

 

	VR Test 1 Reverb G1 (2205x2165)
	Min	Max	Avg
1	69	91	85.317
2	66	91	85
3	69	91	86.25
	68	91	85.52233333
	VR Test 2  Reverb G1 3168x3100
	Min	Max	Avg
1	64	91	83.733
2	64	91	82.667
3	61	91	82.733
	63	91	83.04433333

 

 

From the prior testing with the 5800X and DDR4 3600 was:

	VR Test 1 Reverb G1 (2208x2165)
	Min	Max	Avg
1	56	91	75.233
2	57	91	75.8
3	57	91	76.617
	56.66667	91	75.88333
	VR Test 2 Reverb G1 (3124x3056)
	Min	Max	Avg
1	54	91	72.7
2	54	91	72.667
3	56	91	74.633
	54.66667	91	73.33333

 

 

So we're looking at about a 15-20% increase in the minimum frame rates and a 13% increase in the average frame rates for both settings. Both are capped at 91 FPS max frame rates.

 

Addendum: I found it. I needed to set "PerformanceMode" : true, in the default.vrsettings. That enables the 120hz mode and lets the system run at 60fps. With that, I can now use the attached graphics settings *and* Map scenery distance set to unlimited and get the following performance:

	5800X3D DDR 3600
	VR Test: Current Settings VP2 (Extreme 2880x2880, Perfromance Mode)
	Min	Max	Avg
1	56	61	59.817
2	53	61	59.817
3	55	61	59.867
	54.66667	61	59.83367

 

Minimum FPS for 54.5 fps. Average FPS of 59.867 hz. 

 

We have achieved 60 fps with maximum view distance folks.

 

Victory!

Edited May 11, 2022 by Voyager

[RAAF492SQNOz_Steve]

6 hours ago, Voyager said:

 

 

 

<snip>

Addendum: I found it. I needed to set "PerformanceMode" : true, in the default.vrsettings. That enables the 120hz mode and lets the system run at 60fps. With that, I can now use the attached graphics settings *and* Map scenery distance set to unlimited and get the following performance:

	5800X3D DDR 3600
	VR Test: Current Settings VP2 (Extreme 2880x2880, Perfromance Mode)
	Min	Max	Avg
1	56	61	59.817
2	53	61	59.817
3	55	61	59.867
	54.66667	61	59.83367

 

Minimum FPS for 54.5 fps. Average FPS of 59.867 hz. 

 

We have achieved 60 fps with maximum view distance folks.

 

Victory!

 

Thanks for your efforts! Running the Benchmark so many times is true dedication.

 

I have a 5800x3D on back order and was debating on cancelling it.
Based on your comprehensive testing and useful gains, will leave the order in place.

When I ordered it, delivery was 2 -3 days, then got an email update  for 2 - 3 weeks delivery so suspect I have been the victim of false advertising.

At least it is not at the inflated price that some other retailers are asking (and no stock as well).  

Edited May 11, 2022 by RAAF492SQNOz_Steve
Typo

[BeastyBaiter]

Motherboard: Gigabyte Z490 Vision G
CPU:                 i5 10600k
CPU Freq:        4.9 Ghz (overclocked)
Cores:               6
Threads:           12
RAM size:        32Gb (4x8GB)
RAM Freq:       3200 MHz
NB Freq:          4500 MHz
RAM timings:  14-14-14-34
GPU:                RTX 2080 Super

OS:                  Windows 10 Home

VR headset:    Reverb G2

 

                                       Frames               Min              Max              Avg

CPU Test in 1080p:         5693                 81                133               94

GPU Test VR1:                 3809                 49                91                 63

GPU Test VR2:                3465                 46                 71                 57

 

Edited May 12, 2022 by BeastyBaiter

[chiliwili69]

On 5/11/2022 at 7:49 AM, Voyager said:

So we're looking at about a 15-20% increase in the minimum frame rates and a 13% increase in the average frame rates for both settings. Both are capped at 91 FPS max frame rates.

This +10fps increase with G1 is also amazing. Just by simply going from 5800X to 5800X3D, just cache size.

We need to take into account that the closer you are to 90 the smaller will be the performance increase.

 

Your tests are really enlighting to me. I have now realized what important rol plays the CPU cache size in IL-2 VR. 

 

It would be interesting if someone with a 12900K or 12900KS can run the CPU test and VR tests with latest version of IL-2, just to see if Intel is still the king.

On 5/12/2022 at 3:18 AM, BeastyBaiter said:

GPU Test VR1:                 3809                 49                91                 63

GPU Test VR2:                3465                 46                 71                 57

 

Thank you for those test.

 

In the VR1 test, the CPU is the constraint.  In the VR2 test, the GPU constrained it a bit more, the max went from 91 to 71. Your tests are consistent with other G2 and Intel sub-12th gen.

[BeastyBaiter]

My shiny new RX 6950XT showed up last night, so I reran the test with out of the box settings for it. It is giving worse results but I also noticed it was only at around 40-50% usage in the VR high test while my 2080S was mostly tapped out. Clearly it's a CPU bottleneck most of the time and it appears this card has a bigger CPU bottleneck than the old 2080S does. I noticed similar results in DCS, peak performance suffers due to more CPU load but the card itself is at least twice as powerful.

 

Motherboard: Gigabyte Z490 Vision G
CPU:                 i5 10600k
CPU Freq:        4.9 Ghz (overclocked)
Cores:               6
Threads:           12
RAM size:        32Gb (4x8GB)
RAM Freq:       3200 MHz
NB Freq:          4500 MHz
RAM timings:  14-14-14-34
GPU:                RX 6950XT

OS:                  Windows 10 Home

VR headset:    Reverb G2

 

                                       Frames               Min              Max              Avg

CPU Test in 1080p:         5169                 75                126               86

CPU Test VR1:                 3486                 47                86                58

GPU Test VR2:                3283                 44                75                 54

Edited May 13, 2022 by BeastyBaiter
correction on Frames for 1080p

[chiliwili69]

On 5/13/2022 at 5:32 PM, BeastyBaiter said:

GPU:                RX 6950XT

 

Thank you very much for running the benchmark before and after putting your new 6950XT.

 

As you can see, those AMD GPUs have a problem with IL-2. TEchnically they should be superior to the 3080Ti and 3090, but like other previous 6900XT and 6800XT there is an issue with IL-2, in monitor and VR.

 

The issue was reported to the Dev team a while ago: 

 

I am sure in other simulators and games the 6950XT is a beast.

[BeastyBaiter]

I ran the 1080p CPU test again with the AMD May Preview drivers as they allegedly make a big difference in some DX11 games. Result below:

 

                                       Frames               Min              Max              Avg

CPU Test in 1080p:         5345                 79                131               89

 

 

It's a very modest improvement. I also monitored GPU usage for a laugh, it never exceeded 27% ?

[1Sascha]

New GPU is up and running, rest of the system is unchanged, except that I backed off on the CPU turbo-offset overclock (from +2 to +1 / 5.1 GHz to 5.0 GHz) to go easy on my PSU (550w, which is at its upper limit now with the RTX 2070 Super in there instead of the GTX 1060 3GB I had before). Seems to have worked since under super heavy load (benchmarks, basically) I got ~25W lower peak CPU draw with the reduced OC. And according to the PSU calculator I'm using I'm now less than 10 Watt under their recommended PSU strength. ~505 max calculated draw, recommended PSU: ~ 557W, actual PSU rating : 550W.

 

 

 

My Specs:
 Motherboard: MSI MAG Z690 Tomahawk Wifi DDR4
 CPU: Intel i5-12600K, slightly OCed via BIOS turbo-offset (+1) on the P- and E-cores, cooled by a Lian Li Galahad 240
 CPU Freq: P-cores max 1080 run: 4.98 GHz, P-cores max in VR test1: 4.78 GHz
 L3 cache: 20 MB
 Cores: 6P + 4E = 10
 Threads: 16
 RAM type: Corsair Vengeance Pro RGB, 3600 DDR4 (XMP is on of course)
 RAM size:  2x8 GB, dual channel
 NB Freq: 3600 MHz
 RAM Freq: 3600 MHz
 RAM Latency:  CL18
 GPU: Inno3D RTX 2070 Super Twin X2 OC, stock. (GPU: 1605MHz, Boost: 1785MHz. GDDR6: 1750 MHz ->14000 MHz)

 

-> according to Passmark's performance test, the 2070 should be pretty much exactly 100% more powerful than the 1060. 3D scores were 11,000 and 22,000.

OS is Win 11 Pro

 

 

 

CPU Test Results / 1080p run:

 

2022-05-22 14:00:21 - Il-2
Frames: 6536 - Time: 60000ms - Avg: 108.933 - Min: 96 - Max: 155

 

Guess my PC is now a lot less bottle-necked by its GPU - which was what I was hoping for.

 

 

Didn't run the 4K test in 1440p this time, but since I now have an Oculus Rift S here to play around with (main reason for the GPU-update TBH ? ...

 

VR Test 1:

2022-05-22 14:21:44 - Il-2
Frames: 3006 - Time: 60000ms - Avg: 50.100 - Min: 37 - Max: 81

 

EDIT: Those were taken with 136% SS in SteamVR, re-ran the test just now with the correct value (184), and:

2022-05-23 11:27:51 - Il-2
Frames: 2542 - Time: 60000ms - Avg: 42.367 - Min: 38 - Max: 62

 

I guess I'm going to have to go back to 100% SS, seeing my results here... at least if I want to turn on a few bells and whistles in VR mode... ?

 

What's a bit odd is that the VR run didn't push my GPU to its power limit - 3DMark and Passmark's performance test both did and caused max GPU power draw of 215 Watts... the VR test "only" needed 189 W. Yes... the required in-game settings for this test are quite low, but 136% SS should be pretty harsh on my puny 2070 I would imagine... Gotta do more tweaking I suppose.

 

 

 

S.

 

 

 

 

 

Edited May 23, 2022 by 1Sascha

[dburne]

1 hour ago, 1Sascha said:

New GPU is up and running, rest of the system is unchanged, except that I backed off on the CPU turbo-offset overclock (from +2 to +1 / 5.1 GHz to 5.0 GHz) to go easy on my PSU (550w, which is at its upper limit now with the RTX 2070 Super in there instead of the GTX 1060 3GB I had before). Seems to have worked since under super heavy load (benchmarks, basically) I got ~25W lower peak CPU draw with the reduced OC. And according to the PCU calculator I'm using I'm now less than 10 Watt under their recommended PSU strength. ~505 max calculated draw, recommended PSU: ~ 557W, actual PSU rating : 550W.

 

My Specs:
 Motherboard: MSI MAG Z690 Tomahawk Wifi DDR4
 CPU: Intel i5-12600K, slightly OCed via BIOS turbo-offset (+1) on the P- and E-cores, cooled by a Lian Li Galahad 240
 CPU Freq: P-cores max 1080 run: 4.98 GHz, P-cores max in VR test1: 4.78 GHz
 L3 cache: 20 MB
 Cores: 6P + 4E = 10
 Threads: 16
 RAM type: Corsair Vengeance Pro RGB, 3600 DDR4 (XMP is on of course)
 RAM size:  2x8 GB, dual channel
 NB Freq: 3600 MHz
 RAM Freq: 3600 MHz
 RAM Latency:  CL18
 GPU: Inno3D RTX 2070 Super Twin X2 OC, stock. (GPU: 1605MHz, Boost: 1785MHz. GDDR6: 1750 MHz ->14000 MHz)

OS is Win 11 Pro

 

 

 

CPU Test Results / 1080p run:

 

2022-05-22 14:00:21 - Il-2
Frames: 6536 - Time: 60000ms - Avg: 108.933 - Min: 96 - Max: 155

 

Guess my PC is now a lot less bottle-necked by its GPU - which was what I was hoping for.

 

 

Didn't run the 4K test in 1440p this time, but since I now have an Oculus Rift S here to play around with (main reason for the GPU-update TBH ? ...

 

VR Test 1:

2022-05-22 14:21:44 - Il-2
Frames: 3006 - Time: 60000ms - Avg: 50.100 - Min: 37 - Max: 81

 

I guess I'm going to have to go back to 100% SS, seeing my results here... at least if I want to turn on a few bells and whistles in VR mode... ?

 

What's a bit odd is that the VR run didn't push my GPU to its power limit - 3DMark and Passmark's performance test both did and caused max GPU power draw of 215 Watts... the VR test "only" needed 189 W. Yes... the required in-game settings for this test are quite low, but 136% SS should be pretty harsh on my puny 2070 I would imagine... Gotta do more tweaking I suppose.

 

 

 

S.

 

The important thing is how it feels for you whilst actually playing the game. No running a trk file is not going to put the same load on your system as running the game itself. I suggest not getting too wrapped up in chasing numbers especially with just a trk file. The main thing this is good for is just comparing with others equipment, not necessarily reflective of the experience you get running the game.

[1Sascha]

11 minutes ago, dburne said:

The important thing is how it feels for you whilst actually playing the game. No running a trk file is not going to put the same load on your system as running the game itself. I suggest not getting too wrapped up in chasing numbers especially with just a trk file. The main thing this is good for is just comparing with others equipment, not necessarily reflective of the experience you get running the game.

Oh yeah.. I wasn't talking about the benchmark.. talking about how the game performs. Trouble is, FPS are pretty much all over the place in VR and you don't have the counter in front of you all the time. It's more like an FPS range I can get from the counter... it'll often sit at or near 80, but I do see 40s and 50s too... since I can't watch the counter all the time in a fight, it's hard to tell whether I'm doing good with occasional drops or doing badly with occasional highs. If that makes sense...

 

Looking at HWMonitor, I seem to be getting close to good settings though... in-game, my CPU isn't really taxed that much... not even breaking 100W max draw. The GPU is near its upper limit in-game... like 190 to 195 W (215W is its limit), GPU load is 96%, GPU memory use is 87/88%. Temps on the GPU itself are good (77° max), but hotspot and memory are a bit scary at 92.8°C each. The GPU's fans are running pretty fast already at load with ~2000 rpm, so I don't know if a custom fan-curve will help here.

 

EDIT: And I just realized that all that was with 136% SS for IL-2... I forgot to reduce it back to 100 or 110 after I'd run the benchmark... oops.

 

S.

Edited May 22, 2022 by 1Sascha

[dburne]

1 minute ago, 1Sascha said:

Oh yeah.. I wasn't talking about the benchmark.. talking about how the game performs. Trouble is, FPS are pretty much all over the place in VR and you don't have the counter in front of you all the time. It's more like an FPS range I can get from the counter... it'll often sit at or near 80, but I do see 40s and 50s too... since I can't watch the counter all the time in a fight, it's hard to tell whether I'm doing good with occasional drops or doing badly with occasional highs. If that makes sense...

 

Looking at HWMonitor, I seem to be getting close to good settings though... in-game, my CPU isn't really taxed that much... not even breaking 100W max draw. The GPU is near its upper limit in-game... like 190 to 195 W (215W is its limit), GPU load is 96%, GPU memory use is 87/88%. Temps on the GPU itself are good (77° max), but hotspot and memory are a bit scary at 92.8°C each. The GPU's fans are running pretty fast already at load with ~2000 rpm, so I don't know if a custom fan-curve will help here.

 

S.

 

I ramp all my fans up to max - I use earbuds and they isolate the noise pretty good for me. I try and keep my rig as cool as I can whilst gaming.