Jump to content

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


chiliwili69
 Share

Recommended Posts

chiliwili69
Posted (edited)
9 hours ago, =VARP=Tvrdi said:

GPU: Gigabyte GTX 1660 OC 6GB (~gtx1070)

 

Moving from 1200p (2.3MPixels) to 1440p (3.7Mpixels) you have basically increased by 60% the GPU load, being CPU load basically the same than before.

 

This, in addition to your settings for clouds (Extreme) and MSAAx2 is the perfect combination to give your  GTX1660 a hard time.

 

I suggest you use FXA x2 and relax the clouds settings. Otherwise upgrade your GPU.

Edited by chiliwili69
Link to comment
Share on other sites

=VARP=Tvrdi

I got more fps but I was avg 77 in mp anyways, now its even more. FPS wasnt a problem. The problem is...on some missions/maps tracking motion with head tracker is super smooth and on some isnt. Not that its stuttering but its noticable. And it seams it happens to ppl with better rigs as I heard.

Edited by =VARP=Tvrdi
Link to comment
Share on other sites

DBCOOPER011

Below is my data except for the 4K test, as I don't have that type of monitor/TV..

 

Motherboard: ASUS Z490-E
CPU:               10700K
CPU Freq:       5.2 Ghz
L3 cache:       16 MB
Cores:             8
Threads:         8
RAM type:      DDR4
RAM size:       32Gb (4x8GB)
NB Freq:         4700 MHz
RAM Freq:      4000 MHz
RAM timings:  16-16-16-36-400
Ram type: G.Skill F4-3600C16-8GVK
GPU:       Gigabyte Vision   3090
HMD: Valve Index
AIO: Artic Freezer 2- 360

 

IL2 Version 4.506, Benchmark V6

 

CPU Test:
Frames: 6421 - Time: 60000ms - Avg: 107.017 - Min: 94 - Max: 139

 

VR Test 1:
80MHZ- Frames: 4726 - Time: 60000ms - Avg: 78.767 - Min: 72 - Max: 82
90MHZ- Frames: 4963 - Time: 60000ms - Avg: 82.717 - Min: 70 - Max: 91
120MHZ- Frames: 4859 - Time: 60000ms - Avg: 80.983 - Min: 59 - Max: 102

 

VR Test 2:
80MHZ- Frames: 4720 - Time: 60000ms - Avg: 78.667 - Min: 71 - Max: 81
90MHZ- Frames: 4920 - Time: 60000ms - Avg: 82.000 - Min: 68 - Max: 92
120MHZ-Frames: 4899 - Time: 60000ms - Avg: 81.650 - Min: 59 - Max: 107

 

GPU Passmark (3d graphics mark): 28013

 

Edited by DBCOOPER011
Link to comment
Share on other sites

chiliwili69
10 hours ago, DBCOOPER011 said:

VR Test 1:
80MHZ- Frames: 4726 - Time: 60000ms - Avg: 78.767 - Min: 72 - Max: 82
90MHZ- Frames: 4963 - Time: 60000ms - Avg: 82.717 - Min: 70 - Max: 91
120MHZ- Frames: 4859 - Time: 60000ms - Avg: 80.983 - Min: 59 - Max: 102

 

VR Test 2:
80MHZ- Frames: 4720 - Time: 60000ms - Avg: 78.667 - Min: 71 - Max: 81
90MHZ- Frames: 4920 - Time: 60000ms - Avg: 82.000 - Min: 68 - Max: 92
120MHZ-Frames: 4899 - Time: 60000ms - Avg: 81.650 - Min: 59 - Max: 107

 

Thank you for your multiple tests at different freqs. They say a lot.

 

Firstly, you can see that VrTest1 and Vrtest2 deliver almost same results. This means that in VRtest1 your 3090 is not really loaded and the limitation is not the GPU, it is the CPU. And secondly that in both tests (1 and 2) you are CPU limited.

 

In VRtest1 at 80Hz you are almost at 80fps, so this 80Hz mode is really good for CPU limited systems. And when you move to 90Hz or 120HZ you hit the CPU limit around 82fps. With my Index I always use this 80Hz mode.

 

Other people with ryzens 5x00X and (3090, 3080 and even 1080Ti) were able to reach around 88 with the VRtest1 at 90Hz with G1&2. This confirms that your 82fps in the 90Hz tests are CPU limited.

 

But this is just a benchmark. You system is quite solid and for sure you will be 99% fine in most of real use case scenarios with 80Hz

Link to comment
Share on other sites

DBCOOPER011

Thanks for the input.  I agree that I'm limited by the CPU, and curious a little bit about what the Intel 11 series CPU have to offer. I might end up with one of those depending upon how the reviews look later this week. Really don't want to get another motherboard for a Ryzen chip.


I re-ran the tests again with the highest stable overclock I could get (5.3Mhz peak with an adaptive VID voltage averaging 1.360). It did improve my FPS a little, and is about the most I can get out of this chip...

 

2nd Test

 

CPU Test:

 

Frames: 6493 - Time: 60000ms - Avg: 108.217 - Min: 97 - Max: 145

 

 

VR Test 1:

 

80MHZ- Frames: 4766 - Time: 60000ms - Avg: 79.433 - Min: 75 - Max: 82
90MHZ- Frames: 5121 - Time: 60000ms - Avg: 85.350 - Min: 74 - Max: 92
120MHZ- Frames: 5463 - Time: 60000ms - Avg: 91.050 - Min: 70 - Max: 110

 

VR Test 2:

 

80MHZ- Frames: 4759 - Time: 60000ms - Avg: 79.317 - Min: 74 - Max: 81
90MHZ- Frames: 5153 - Time: 60000ms - Avg: 85.883 - Min: 76 - Max: 92
120MHZ- Frames: 5432 - Time: 60000ms - Avg: 90.533 - Min: 64 - Max: 110

Edited by DBCOOPER011
  • Thanks 1
Link to comment
Share on other sites

2 hours ago, DBCOOPER011 said:

Thanks for the input.  I agree that I'm limited by the CPU, and curious a little bit about what the Intel 11 series CPU have to offer. I might end up with one of those depending upon how the reviews look later this week. Really don't want to get another motherboard for a Ryzen chip.

See you do have an 8 core, are you inn for an experiment for more CPU performance? PM me, because I first want to have more prove if its working on other systems before publishing this old trick that everyone seems to have forgotten. 

 

Link to comment
Share on other sites

DBCOOPER011

Thanks, PM sent.

 

Just received an email from newegg that they charged/sent an 11700K I pre-ordered a while ago. Kind of pisses me off cause I canceled the pe-order last week. I might be disappointed with it, but we we see...

  • Upvote 1
Link to comment
Share on other sites

J5_NiiranenVR-Gfr

I have an I5 9600 running at 5.2 GHz - 6 core water-cooling .. 

Will I notice if I change to a maybe I9 10xxx something gen 11 something running 5.2 GHz 

What is most important - the GHz , gen , Ix ....??????

 

 

Link to comment
Share on other sites

chiliwili69
11 hours ago, DBCOOPER011 said:

VR Test 1:

 

80MHZ- Frames: 4766 - Time: 60000ms - Avg: 79.433 - Min: 75 - Max: 82
90MHZ- Frames: 5121 - Time: 60000ms - Avg: 85.350 - Min: 74 - Max: 92
120MHZ- Frames: 5463 - Time: 60000ms - Avg: 91.050 - Min: 70 - Max: 110

 

VR Test 2:

 

80MHZ- Frames: 4759 - Time: 60000ms - Avg: 79.317 - Min: 74 - Max: 81
90MHZ- Frames: 5153 - Time: 60000ms - Avg: 85.883 - Min: 76 - Max: 92
120MHZ- Frames: 5432 - Time: 60000ms - Avg: 90.533 - Min: 64 - Max: 110

 

Hey, thanks for this second test at 5.3. You got a significant gain on the VRtest1&2. In all registers, min, max, avg. Very good.

7 hours ago, DBCOOPER011 said:

Just received an email from newegg that they charged/sent an 11700K I pre-ordered a while ago. Kind of pisses me off cause I canceled the pe-order last week. I might be disappointed with it, but we we see...

 

If you reorder we will be very interested in your benchmark...  

 

It looks promising on the single-core department:

https://cpu.userbenchmark.com/Compare/Intel-Core-i7-11700K-vs-Intel-Core-i7-10700K/4107vs4070

 

If the gain is significant, the good thing for FCLGA1200 socket owners is that they will need to simply change the CPU.

1 hour ago, J5_NiiranenVR-Gfr said:

What is most important - the GHz , gen , Ix ....??????

 

The most important for IL-2 is the result of the SYN_Vander test.  😉

 

So, it depends on CPU brand (Intel, Ryzen), the model or gen determine the IPC (Ryzen 3600X vs 5600X, or 9700K vs 10700K vs 11700K), the GHz and the RAM (speed and latency). 

 

Currently the Ryzens 5900X and 5950X are the beasts to beat. And it seems the new 11700K and 11900K will beat them, but we need to test it.

Link to comment
Share on other sites

2 hours ago, Alonzo said:

Anyone got resizable BAR working on an NVidia card yet? I think NVidia are gating this feature behind game testing, so it probably needs NVidia inspector to enable it for IL2. I'll look into this eventually but no time at the moment.

 

https://www.nvidia.com/en-us/geforce/news/geforce-rtx-30-series-resizable-bar-support/

 

It is all enabled on my end as of today.

Doubt our flight sims would make use of it currently.

 

 

Link to comment
Share on other sites

RAAF492SQNOz_Steve
2 hours ago, dburne said:

 

It is all enabled on my end as of today.

Doubt our flight sims would make use of it currently.

 

 

 MS Flight Sim performance was worse when tested with AMD's version of Rebar so that FS result does not look promising.

 

Fingers and toes crossed that NVidia adds IL2 to the list of supported games and that gives us some useful fps improvements. 😧

 

MSI have not released their GPU Bios update at the time I write this so I am a spectator at present.

Link to comment
Share on other sites

DBCOOPER011

Greetings,


This is the preliminary IL2 bench I have for the 11700K. This thing pulls a lot of Vcore (1.5V) but strangely it is cooler then what my 10700K was. I was fighting CTD's for awhile and figured out it was because the PCI 4.0 was faulty on my board. Set it to 3.0 and haven't had a CTD since. This is an ASUS Beta Bios (2004) from January, Hope an update corrects it. I really don't see a lot benefit from this CPU over my overclocked 10700k, I do however see much less CPU frame time spikes in VR..

 

Motherboard: ASUS Z490-E
CPU:               11700K
CPU Freq:       5.2 Ghz
L3 cache:       16 MB
Cores:             8
Threads:         8
RAM type:      DDR4
RAM size:       32Gb (4x8GB)
NB Freq:         4000 MHz
RAM Freq:      3600 MHz
RAM timings:  15-15-15-36-300 2T
Ram type: G.Skill F4-3600C16-8GVK
GPU:       Gigabyte Vision   3090
HMD: Valve Index
AIO: Artic Freezer 2- 360

 

5.2 GHZ:
CPU test:
Frames: 6945 - Time: 60000ms - Avg: 115.750 - Min: 102 - Max: 152

 

VR Test 1:
80MHZ: Frames: 4752 - Time: 60000ms - Avg: 79.200 - Min: 76 - Max: 82
90MHZ: Frames: 5269 - Time: 60000ms - Avg: 87.817 - Min: 78 - Max: 92
120MHZ: Frames: 5634 - Time: 60000ms - Avg: 93.900 - Min: 78 - Max: 112

 

VR Test 2:
80MHZ: Frames: 4793 - Time: 60000ms - Avg: 79.883 - Min: 76 - Max: 82
90MHZ:Frames: 5257 - Time: 60000ms - Avg: 87.617 - Min: 78 - Max: 92
120MHZ:Frames: 5603 - Time: 60000ms - Avg: 93.383 - Min: 78 - Max: 112

 

GPU Passmark (3d graphics mark): 27189
CPU Passmark (Single thread): 3668

 

Edited by DBCOOPER011
  • Thanks 3
Link to comment
Share on other sites

biglouis

Hey

8 hours ago, DBCOOPER011 said:

Greetings,


This is the preliminary IL2 bench I have for the 11700K. This thing pulls a lot of Vcore (1.5V) but strangely it is cooler then what my 10700K was. I was fighting CTD's for awhile and figured out it was because the PCI 4.0 was faulty on my board. Set it to 3.0 and haven't had a CTD since. This is an ASUS Beta Bios (2004) from January, Hope an update corrects it. I really don't see a lot benefit from this CPU over my overclocked 10700k, I do however see much less CPU frame time spikes in VR..

 

Motherboard: ASUS Z490-E
CPU:               11700K
CPU Freq:       5.2 Ghz
L3 cache:       16 MB
Cores:             8
Threads:         8
RAM type:      DDR4
RAM size:       32Gb (4x8GB)
NB Freq:         4000 MHz
RAM Freq:      3600 MHz
RAM timings:  15-15-15-36-300 2T
Ram type: G.Skill F4-3600C16-8GVK
GPU:       Gigabyte Vision   3090
HMD: Valve Index
AIO: Artic Freezer 2- 360

 

5.2 GHZ:
CPU test:
Frames: 6945 - Time: 60000ms - Avg: 115.750 - Min: 102 - Max: 152

 

VR Test 1:
80MHZ: Frames: 4752 - Time: 60000ms - Avg: 79.200 - Min: 76 - Max: 82
90MHZ: Frames: 5269 - Time: 60000ms - Avg: 87.817 - Min: 78 - Max: 92
120MHZ: Frames: 5634 - Time: 60000ms - Avg: 93.900 - Min: 78 - Max: 112

 

VR Test 2:
80MHZ: Frames: 4793 - Time: 60000ms - Avg: 79.883 - Min: 76 - Max: 82
90MHZ:Frames: 5257 - Time: 60000ms - Avg: 87.617 - Min: 78 - Max: 92
120MHZ:Frames: 5603 - Time: 60000ms - Avg: 93.383 - Min: 78 - Max: 112

 

GPU Passmark (3d graphics mark): 27189
CPU Passmark (Single thread): 3668

 

Hey that's not so bad. You got like +10% while having worse RAM (3600 cl15 vs 4000c16) and -100mhz.

 

edit; just ran it on my 11900k

 

Motherboard: Asus Rog Strix-E z590 Gaming WIFI
CPU:               11900k
CPU Freq:       stock settings: 125w / 250w, no ABT but HT disabled (it gives me BSODs with latest BIOS...)
L3 cache:       16 MB
Cores:             8
Threads:         8
RAM type:      DDR4
RAM size:       32Gb (4x8GB)
NB Freq:         4000 MHz
RAM Freq:      3733 MHz (Gear 1 mode)
RAM timings:  14-15-15-35-654 2T
Ram type: G. Skill F4-4000C15D-16GTZR*
GPU:       Asus Rog TUF 3090 (quiet mode, 350w, no OC) @ PCI-E Gen 4

 

CPU test:

 

2021-04-07 15:48:27 - Il-2
Frames: 7036 - Time: 60000ms - Avg: 117.267 - Min: 100 - Max: 156

 

Too lazy to do the other tests for now and anyway this is the one that matters most for me since the CPU has always been my #1 bottleneck in VR. I want to tweak my CPU settings since I have a lot of cooling headroom, but I haven't really had the time to do so. It's also really stupid that I'm unstable at stock unless I disable HT. If Asus can't fix this before my return window ends I might just get a good old 10900k that costs a whopping 200€ less.

 

*I can run the ram with the XMP profile in Gear 2 mode but the latency penalty hurts gaming performance too much to be worth it in any game I've tried. Bandwidth difference is like 55000 (3733) vs 57000 (4000) while latency goes from about 45ns to 50ns in the Aida64 test.

 

Maybe some interesting results could be seen from people pushing high CPU and RAM overclocks on Rocket Lake (5000mhz+ DDR4 is actually possible in Gear 2 mode, and should negate the latency penalty easily). I am not one to do that sort of thing though. I value low noise and stability far more than the last few frames.

 

 

Edited by biglouis
  • Thanks 1
Link to comment
Share on other sites

VA_longboarder131

Hello,

 

Recently got into the game, VR, etc. Wanted to post my results from this test. I did want to note that two days ago I was using a Ryzen 9 3900XT and my first set of results are from that. The second set of results will be from the Ryzen 9 5950x I recently acquired and installed. Pretty interesting changes. Anyway, results are as follows:

 

3900XT

Motherboard: ASUS ROG STRIX X570-E Gaming

CPU: AMD Ryzen 9 3900XT

CPU Freq: 3.8 GHz

L3 cache: 4x16 Mb

Cores: 12

Threads: 24

RAM type: DDR4

RAM size: 64 GB (4x16 Gb)

NB Freq: 1800 MHz

RAM Freq: 3600 MHz

RAM timings: 16-19-19-39-85

GPU: RTX 3080 FE

CPU Cooler: EK 360 AIO

MCLK, FCLK, UCLK: 1800

RAM Latency: 72.9 ns

HMD: HP Reverb G2

 

VR TEST1- SteamVR 50% SS:

2021-04-05 10:37:56 - Il-2

Frames: 3558 - Time: 60000ms - Avg: 59.300 - Min: 49 - Max: 78

 

VR TEST2- SteamVR 100% SS:

2021-04-05 10:32:09 - Il-2

Frames: 3608 - Time: 60000ms - Avg: 60.133 - Min: 49 - Max: 82

 

 

5950x

Motherboard: ASUS ROG STRIX X570-E Gaming

CPU: AMD Ryzen 9 5950c

CPU Freq: 3.6 GHz

L3 cache: 2x32 Mb

Cores: 16

Threads: 32

RAM type: DDR4

RAM size: 64 GB (4x16 Gb)

NB Freq: 1800 MHz

RAM Freq: 3600 MHz

RAM timings: 16-19-19-39-85

GPU: RTX 3080 FE

CPU Cooler: EK 360 AIO

MCLK, FCLK, UCLK: 1800

RAM Latency: 64.8 ns

HMD: HP Reverb G2

 

VR TEST1- SteamVR 50% SS:

2021-04-07 09:50:50 - Il-2

Frames: 5079 - Time: 60000ms - Avg: 84.650 - Min: 64 - Max: 92

 

VR TEST2- SteamVR 100% SS:

2021-04-07 09:58:23 - Il-2

Frames: 4933 - Time: 60000ms - Avg: 82.217 - Min: 60 - Max: 91

 

Interested to hear your take on it. Overall, I'm happy with these results, now I just wanna tweak the visuals to find that sweet spot of good spotting and visual fidelity.

 

This is the first PC I'd built since high school ~20 years ago. I had no idea about RAM and figured "more is better" which apparently isn't the case, and CAS latency (as I've just been learning these last couple of days) is pretty important. 64 Gb is way overkill I now realize. Would you suggest going to a lower amount, with a CAS latency of 14 instead? 

 

Thanks in advance!

Edited by longboarder131
formatting, inclusion of HMD
Link to comment
Share on other sites

DBCOOPER011

Thanks for the input. The most my memory could get in gear 2 mode was 4000MHZ, and 3600 was my tops for gear 1. I also disabled HT, although I could run it with it on, I just didnt see the point in leaving it on just for gaming. The best setting I found so far was to let the ASUS AI do the overclock and then I manually  adjusted the ratio in AI Suite. My CPU core voltage is set for 1.492, but the amount of voltage the cores are pulling averages 1.385 during the test. I'm going to try and lower the core voltage some.  I just re-ran the CPU test and got the below numbers. Still tinkering with this thing...

 

5.1GHZ- Frames: 7167 - Time: 60000ms - Avg: 119.450 - Min: 107 - Max: 158
5.2GHZ-Frames: 7210 - Time: 60000ms - Avg: 120.167 - Min: 105 - Max: 159

 

IL2 Bench HWinfo64.PNG

Link to comment
Share on other sites

biglouis
12 hours ago, DBCOOPER011 said:

Thanks for the input. The most my memory could get in gear 2 mode was 4000MHZ, and 3600 was my tops for gear 1. I also disabled HT, although I could run it with it on, I just didnt see the point in leaving it on just for gaming. The best setting I found so far was to let the ASUS AI do the overclock and then I manually  adjusted the ratio in AI Suite. My CPU core voltage is set for 1.492, but the amount of voltage the cores are pulling averages 1.385 during the test. I'm going to try and lower the core voltage some.  I just re-ran the CPU test and got the below numbers. Still tinkering with this thing...

 

5.1GHZ- Frames: 7167 - Time: 60000ms - Avg: 119.450 - Min: 107 - Max: 158
5.2GHZ-Frames: 7210 - Time: 60000ms - Avg: 120.167 - Min: 105 - Max: 159

 

IL2 Bench HWinfo64.PNG

 

Wow, even better! Getting closer to AMD haha (although they still win this round, undeniable). 

 

HT is useless for IL-2 I imagine but it's actually a big deal in many other games nowadays, that's why I personally care (I play many different games). Many modern games run considerably better with HT enabled - sometimes the average fps doesn't look very different but the minimum fps is often greatly improved. An extreme example: Total War Troy (4k full ultra): 50 fps HT off vs 70 fps HT on (11900k). 

 

In theory HT can hurt single thread performance but I have never managed to measure that at all on any games I've tested since the 8700k era (even really old games). Disabling it can let you push higher overclocks (it reduces the load on the CPU) so there is value in doing it for games that use very few threads or if you already have more than enough threads without HT anyway. 

 

Edited by biglouis
Link to comment
Share on other sites

chiliwili69
On 4/7/2021 at 7:48 AM, DBCOOPER011 said:

5.2 GHZ:
CPU test:
Frames: 6945 - Time: 60000ms - Avg: 115.750 - Min: 102 - Max: 152

 

Thank you so much for this frist test of the 11th gen of Intel. It tells a lot.

 

In the 120Hz tests, the CPU is still the bottleneck I believe.

As you say, the gain with respect 10th gen is really modest. Now this chip is aligned with other 5600X in the CPU test, but below the 5900X/5950X.

 

But on both the VR test, you are at 88fps, so better than my 5600X and 3080. Well in fact, in VR your PC with the 11700K (and also your previous 10700K) are having a top performance.

18 hours ago, biglouis said:

edit; just ran it on my 11900k

 

Thanks for this first test of a 11900K at stock speed. It seems that on VR the gain should higher than in the CPU test. If you have time to do the VR test it would be nice.

Link to comment
Share on other sites

biglouis

Yes I hope to find the time for VR testing soon, maybe this week-end.

 

Perhaps after I've done a bit of memory/cpu tweaking (my goal is to find a sweetspot where temperature/noise are still low enough for my tastes - @ stock it is totally silent with my cooling solution). 

Link to comment
Share on other sites

DBCOOPER011

Greetings again,


I'm done tweaking/testing this thing and have what I consider my "daily driver" as its stable in everything I thrown at it so far. I'm done wasting my vacation away messing with this thing. My last tests are below, and I was able to get a 4K benchmark on the living room TV. I also threw in some benchmark screenshots.


I basically let AI Suite do an adaptive OC with 3 cores 5.2GHZ and the rest at 5.0GHZ. At load, the adaptive voltage fluctuates between 1.4 and 1.545 volts, but the temperature of the CPU barely hits 60C. I was really worried about that amount of voltage at first, but everything I read so far indicates that's pretty much normal for rocketlake. I'm also running it HT disabled.


I agree the performance is barely on par with the 5600X for IL2, I just didn't want to purchase a new motherboard, so I'm pretty happy with it so far. What I don't understand is that this thing kicks butt in single core benchmarks, it just doesn't translate that well into IL2.

 

CPU:
Frames: 7038 - Time: 60000ms - Avg: 117.300 - Min: 104 - Max: 154

4K:
Frames: 9792 - Time: 60000ms - Avg: 163.200 - Min: 135 - Max: 202

VR1-90MHZ:
Frames: 5272 - Time: 60000ms - Avg: 87.867 - Min: 79 - Max: 92

VR2-90MHZ:
Frames: 5209 - Time: 60000ms - Avg: 86.817 - Min: 77 - Max: 92

 

 

CB20 Single.PNG

CPUZ.PNG

Passmark.PNG

Vcore.PNG

Edited by DBCOOPER011
  • Thanks 2
Link to comment
Share on other sites

biglouis

Did a first VR benchmark (exact same settings and hardware as my previous post) with the Oculus Quest 2:

 

VR Test 1 (~9.5m pixels)

Frames: 4271 - Time: 60000ms - Avg: 71.183 - Min: 62 - Max: 73

 

(did it 3-4 times and results are consistent)

 

This was at 72hz (there is some bug with latest version of Oculus and SteamVR is stuck at 72hz...) and 11.9 million (4864x2448) pixels. I did a quick calculation so it should be the same as 9.5 @ 90hz if I'm not mistaken (I suck with numbers).

Edited by biglouis
Link to comment
Share on other sites

chiliwili69
On 4/7/2021 at 7:16 PM, longboarder131 said:

Interested to hear your take on it. Overall, I'm happy with these results, now I just wanna tweak the visuals to find that sweet spot of good spotting and visual fidelity.

 

This is the first PC I'd built since high school ~20 years ago. I had no idea about RAM and figured "more is better" which apparently isn't the case, and CAS latency (as I've just been learning these last couple of days) is pretty important. 64 Gb is way overkill I now realize. Would you suggest going to a lower amount, with a CAS latency of 14 instead?

 

Many thanks for this tests. Just by upgrading your CPU to Zen3 you got +22fps in VR in this heavy test with the G2 at 100%. You are having good numbers in VR.

 

If you look the table at WalterScott or others, you will see that you could go even higher if you run your RAM at 1900 and use lower latency RAM.

 

32GB RAM is more than enough to run this game (also 16GB is enough currently). Try to use two sticks and go to 1900 or more. I think your RAM will be more stable with 2 sticks than with 4.

 

You can also run the CPU test, just for completeness. 

On 4/8/2021 at 11:33 PM, DBCOOPER011 said:

What I don't understand is that this thing kicks butt in single core benchmarks, it just doesn't translate that well into IL2.

 

Thank you again for this additional tests after more tinkering. The test doesn´t translate in the CPU test, but in the VR tests (where geometry is calculated twice) this chips deliver very good performance, well above my 5600X and on pair with the 5900X&5950X top performers.

 

The table is shorted by the CPU test, but If I short the table by the VRTest2 you will be at the top.

 

Just one question, did you run your last tests with RAM and NB freq at 4000MHZ? or just RAM at 3600 and NB at 4000 as the first tests.

Link to comment
Share on other sites

chiliwili69
52 minutes ago, biglouis said:

Oculus Quest 2:

 

VR Test 1 (~9.5m pixels)

Frames: 4271 - Time: 60000ms - Avg: 71.183 - Min: 62 - Max: 73

 

Hey, Thanks for posting the first test of the Quest2. I didn´t include the settings in the instructions since I really don´t know how it works or what SS should be applied.

 

So in this heavy test you are almost at 72fps, this means neither your CPU or your GPU is constraining your experience.

 

Whenever the bug is corrected let us know how it run at 90Hz.

Link to comment
Share on other sites

Gambit21
On 4/8/2021 at 2:33 PM, DBCOOPER011 said:

What I don't understand is that this thing kicks butt in single core benchmarks, it just doesn't translate that well into IL2.

 

 

Watch/monitor your CPU utilization while running IL2.

 

 

 

Link to comment
Share on other sites

thermoregulator

I don't know what happened, but I have just tried to rise steamVR SS to 100% (reverb g2), and found out I am basically locked on 90 Hz in career mode (everything on ultra, clouds and shadows on medium, no AA). This was never the case before. Some wmr update? SteamVR update?

Link to comment
Share on other sites

DBCOOPER011
On 4/10/2021 at 11:26 AM, chiliwili69 said:

Just one question, did you run your last tests with RAM and NB freq at 4000MHZ? or just RAM at 3600 and NB at 4000 as the first tests.

Chili, On that last test my memory was at 3600 and NB at 4000. I recently discovered that my most recent Asus bios was holding me back. I seen a recent bios on the overclock.net forums and it allowed me to put in a higher gear 1 memory clock with tighter timing, amongst other things. Overall I'm very impressed by it, even though its not a sanctioned Asus product. However, I still crash in VR whenever PCI 4.0 is enabled for my video card, and have to go to PCI 3.0, hope a bios update fixes it. I think my sweet spot is a fixed 5.0MHZ OC with a 4.1MHZ ring. I'm running 3733 memory with tightened timings. I re-ran the cpu/vr tests and gained a little bit. I'm impressed with the chip so far as the CPU frametime spikes have gotten a lot less.


Motherboard: ASUS Z490-E
CPU:               11700K
CPU Freq:       5.0 Ghz
L3 cache:       16 MB
Cores:             8
Threads:         8
RAM type:      DDR4
RAM size:       32Gb (4x8GB)
NB Freq:         4100 MHz
RAM Freq:      3733 MHz
RAM timings:  14-14-14-36-654 2T
Ram type: G.Skill F4-3600C16-8GVK
GPU:    Gigabyte Vision   3090
HMD: Valve Index
AIO: Artic Freezer 2- 360
CPU Passmark (Single thread): 3529


CPU test: Frames: 7154 - Time: 60000ms - Avg: 119.233 - Min: 107 - Max: 158
VR Test 1: 90MHZ: Frames: 5310 - Time: 60000ms - Avg: 88.500 - Min: 80 - Max: 92
VR Test 2: 90MHZ:Frames: 5305 - Time: 60000ms - Avg: 88.417 - Min: 80 - Max: 92

 

 

  • Thanks 1
Link to comment
Share on other sites

DarKcyde

I got my results on a slightly older rig, but it holds up very well given the crazy amount of memory tweaking I did.  I went very deep down the RAM timings hole, tweaked almost every single primary, secondary and tertiary timing.  I'm on some cheap B-die, so it won't clock insanely high, but it does OK.  The system punches pretty good for its weight class.

 

I have a Rift S, and I dont use steam at all (like most Oculus users I run OpenComposite).  To get close to the intended resolution, Rift S users need to set Pixel Density 1.4 (using Oculus Tray Tool).  It gives final resolution of 2112x2272 = 9.596Mpixels.  Might want to update the first post, it doesn't mention any Oculus-centric way of getting there.

 

 Motherboard:  Gigabyte Z390 I Aorus Pro Wifi (ITX)
 CPU:                 9600K
 CPU Freq:        5.0 Ghz
 L3 cache:        9 MB
 Cores:               6
 Threads:           6
 RAM type:        DDR4
 RAM size:        16Gb (2x8GB)
 NB Freq:          4700 MHz
 RAM Freq:        3800 Mhz
 RAM timings:  16-20-20-36-304

 AIDA Latency:  42.9 ns
 GPU:                 1080

 

Optional:

CPU Cooler: Custom water loop, CPU and GPU, 5x140mm thick rads
RAM Model: Corsair CMK16GX4M2Z3200C16
GPU Model: MSI 1080 Sea Hawk EK

 

CPU test:  Frames: 6178 - Time: 60000ms - Avg: 102.967 - Min: 91 - Max: 137

VR test 1:  Frames: 4132 - Time: 60000ms - Avg: 68.867 - Min: 39 - Max: 81

 

I managed a 104.1 fps while running 5.1ghz, it would bench but wouldn't survive stress tests.  So I'm not counting that one.

Link to comment
Share on other sites

chiliwili69
3 hours ago, DarKcyde said:

AIDA Latency:  42.9 ns

 

Thank you for your test. You did indeed a good job tunning your RAM, this AIDA latency is quite good.

 

On the VR side for the Rift-S, since it only run at 80Hz (and not 90Hz) I increased the number of total pixel by a factor of 90/80 so the number of pixels rendered per econd is comparable with other test of devices at 90Hz. That´s why the test has to be done to render 10.7million pixels instead of 9.5 million. But that´s OK, comparing performance on different devices with a benchmark is a tricky thing.

Link to comment
Share on other sites

DarKcyde

Ohh, I was wondering why you went higher on that one.  That is a tricky one... because when you are running below 80fps you skew the results lower (or maybe higher, since it moves closer to a GPU limitation?).  Considering all of us so far are getting lower averages, I think it might be best to just use the same rez, and we get naturally capped to 80.

 

The real solution to this issue is to use FCAT VR, which can show max attainable FPS, despite the device refresh rates.

Link to comment
Share on other sites

biglouis

I just did simple math, hz multiplied by pixels on screen = number of pixels per second. The GPU load should be roughly equivalent if you stick with one value for that. That is how I came up with 12 million pixels for 72hz = 9.5 @ 90hz.

 

I know Oculus did the same math when saying Rift S was no more demanding than Rift CV1 despite the higher res.

Link to comment
Share on other sites

Thetxsheriff

IL-2 Version:      4.506

NVIDIA Driver:   461.72

AMD Chipset:    2.11.26.106


Motherboard:    ASUS TUF GAMING X570-PLUS (WI-FI)
CPU:                  AMD Ryzen 9 3900X
CPU Freq:          4.25Ghz All-Core @ 1.28125v
L3 Cache:          4 x 16 MB
Cores:                12
Threads:            24
RAM Type:         DDR4
RAM Size:          32GB (2x16GB)
NB Freq:            1600 MHz
RAM Freq:         3200 MHz
RAM Timings:    16-18-18-36-75
GPU:                  2070 Super    (@ Factory speeds for model)

VR Headset:      HP Reverb G2

 

Optional:        
CPU Cooler:      Noctua NH-D15 (two fans)
RAM Model:      CORSAIR Vengeance RGB Pro (PC4 25600)
GPU Model:      GIGABYTE 2070 Super GAMING OC 3 x WINDFORCE Fans 
MCLK:               1600
FCLK:                1600
UCLK:                1600
Mem Lat:           77.4 ns

 

CPU Test:

Frames: 4821 - Time: 60000ms - Avg: 80.350 - Min: 70 - Max: 108

 

GPU Test (4K):

Frames: 4830 - Time: 60000ms - Avg: 80.500 - Min: 61 - Max: 98

 

VR Test 1 50% SS:

Frames: 3222 - Time: 60000ms - Avg: 53.700 - Min: 46 - Max: 68

 

VR Test 2 100% SS:

Frames: 3187 - Time: 60000ms - Avg: 53.117 - Min: 45 - Max: 65

Link to comment
Share on other sites

chiliwili69
6 hours ago, Thetxsheriff said:

CPU:                  AMD Ryzen 9 3900X

 

Thanks for this test. you system is both constrained in CPU (and GPU) for the G2 at 100%.

If you just replace you 3900X by a Zen3 processor you will get a nice bump.

Link to comment
Share on other sites

paul_leonard

Did the latest update break the mission?  Doesn't load now. Stuck on Mission description is loading.

Link to comment
Share on other sites

2 hours ago, paul_leonard said:

Did the latest update break the mission?  Doesn't load now. Stuck on Mission description is loading.

 

Delete the .msnbin file and it should load correctly.

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

chiliwili69

This last weekend, with the previous version v4.506 I run some tests before upgrading to a new BIOS F13G. I got a very few extra fps on the CPU test:

 

bios F12 (Update AMD AGESA ComboV2 1.1.0.0 D)
Frames: 6975 - Time: 60000ms - Avg: 116.250 - Min: 100 - Max: 151
Frames: 6976 - Time: 60000ms - Avg: 116.267 - Min: 101 - Max: 152
Frames: 6938 - Time: 60000ms - Avg: 115.633 - Min: 102 - Max: 152

 

new BIOS F13G (Update AMD AGESA ComboV2 1.2.0.1 PatchA)
Frames: 7344 - Time: 60000ms - Avg: 122.400 - Min: 108 - Max: 165
Frames: 7211 - Time: 60000ms - Avg: 120.183 - Min: 105 - Max: 153
Frames: 7136 - Time: 60000ms - Avg: 118.933 - Min: 98 - Max: 155
 

Also during the weekend, with the new BIOS F13G I just tested if SMT was giving any advantage. SMT is the hyperthreading version of AMD chips.

I run 10 test for each:

Stock SMT ON
Frames: 7309 - Time: 60000ms - Avg: 121.817 - Min: 103 - Max: 162
Frames: 7352 - Time: 60000ms - Avg: 122.533 - Min: 108 - Max: 165
Frames: 7176 - Time: 60000ms - Avg: 119.600 - Min: 104 - Max: 158
Frames: 7088 - Time: 60000ms - Avg: 118.133 - Min: 106 - Max: 155
Frames: 7464 - Time: 60000ms - Avg: 124.400 - Min: 108 - Max: 172
Frames: 7140 - Time: 60000ms - Avg: 119.000 - Min: 103 - Max: 160
Frames: 7259 - Time: 60000ms - Avg: 120.983 - Min: 107 - Max: 161
Frames: 7230 - Time: 60000ms - Avg: 120.500 - Min: 106 - Max: 159
Frames: 7202 - Time: 60000ms - Avg: 120.033 - Min: 105 - Max: 153
Frames: 7350 - Time: 60000ms - Avg: 122.500 - Min: 108 - Max: 163


Stock SMT OFF
Frames: 7297 - Time: 60000ms - Avg: 121.617 - Min: 106 - Max: 164
Frames: 7493 - Time: 60000ms - Avg: 124.883 - Min: 110 - Max: 172
Frames: 7454 - Time: 60000ms - Avg: 124.233 - Min: 108 - Max: 173
Frames: 7244 - Time: 60000ms - Avg: 120.733 - Min: 106 - Max: 161
Frames: 7151 - Time: 60000ms - Avg: 119.183 - Min: 105 - Max: 154
Frames: 7314 - Time: 60000ms - Avg: 121.900 - Min: 108 - Max: 161
Frames: 7211 - Time: 60000ms - Avg: 120.183 - Min: 106 - Max: 158
Frames: 7165 - Time: 60000ms - Avg: 119.417 - Min: 101 - Max: 154
Frames: 7218 - Time: 60000ms - Avg: 120.300 - Min: 105 - Max: 162
Frames: 7151 - Time: 60000ms - Avg: 119.183 - Min: 105 - Max: 159

 

The Avg of the 10 test were 121 for both, so no gain or loss in using SMT for my 6 cores CPU.

This is the graph:

1481986574_SMTON-OFF.png.f31fe1adfdddcf04739dd11d9295e9fc.png

Link to comment
Share on other sites

chiliwili69

Today I also have run 10 test with the new version 4.601 (you have to delete the .msnbin file in order to run again the mission, thank Alonzo) and the results are:

 

new 4.601, Stock SMT OFF
Frames: 7266 - Time: 60000ms - Avg: 121.100 - Min: 108 - Max: 169
Frames: 7258 - Time: 60000ms - Avg: 120.967 - Min: 107 - Max: 172
Frames: 7491 - Time: 60000ms - Avg: 124.850 - Min: 111 - Max: 174
Frames: 7110 - Time: 60000ms - Avg: 118.500 - Min: 104 - Max: 156
Frames: 7022 - Time: 60000ms - Avg: 117.033 - Min: 101 - Max: 152
Frames: 7166 - Time: 60000ms - Avg: 119.433 - Min: 106 - Max: 163
Frames: 7319 - Time: 60000ms - Avg: 121.983 - Min: 108 - Max: 169
Frames: 7185 - Time: 60000ms - Avg: 119.750 - Min: 106 - Max: 159
Frames: 7413 - Time: 60000ms - Avg: 123.550 - Min: 109 - Max: 171
Frames: 7067 - Time: 60000ms - Avg: 117.783 - Min: 102 - Max: 156

 

The total avg with the previous version was 121.1 fps and with the 4.601 is 120.5 fps. So no change at all (giving the measuring error and variability).

Therefore new tests will be mixed with previous.

Shown now in a graph:

998172430_new4_601.png.2217dfed5334190ba56029989ec2f2ed.png

 

  • Thanks 1
Link to comment
Share on other sites

chiliwili69
4 hours ago, =VARP=Tvrdi said:

When I press mission start nothing happens. It says loading mission description.

 

You have to delete the .msnbin file of the SYN_VANDER files in the mission folder as explained by Alonzo above.

  • Like 1
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...