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

[Gambit21]

I love that the Ryzens are blazing fast and optimized out of the box. That’s how it should be.

[Voxman]

9 hours ago, chiliwili69 said:

The NB freq of your first profile is quite low (800). 

The XMP profiles are always an easy a free upgrade. Not neccesarily the higher freq delivers the more fps. Also latencies and timmings.

 

Unfortunately I've run into some BSoD at the XMP settings.  Have been testing most of the day using Memtest64x, but now testing using MeMtest86.  Memory tests fine stock.  It's the XMP that does it in on Memtest64x.  The quick failure is the higher voltage setting.  I've been rooting around the net to see what issues there are with QVL and such.  Patriot's website does not have a QVL for the i7-10700K, but does for my motherboard.  My motherboard does not have QVL for the Patriot memory I'm using...much confusion.

 

I got the following from PCpartPicker "The Patriot Viper Steel 16 GB (2 x 8 GB) DDR4-4400 CL19 Memory operating voltage of 1.45 V exceeds the Intel Comet Lake CPU recommended maximum of 1.35 V+5% (1.417 V). This memory module may run at a reduced clock rate to meet the 1.35 V voltage recommendation, or may require running at a voltage greater than the Intel recommended maximum. "

 

I'm a bit of a loss as to what to do next.  Stay with default settings, or manually adjust RAM voltage, then the timings?  I seem to remember doing things like this 9 years ago.

 

 

Or not bother and start OC'ing the CPU?

9 hours ago, chiliwili69 said:

First thing you need is a good CPU cooler. I prefer the AIO water cooling. Zero problems so far.

 

I'm running liquid cooling.  Fractal Designs Celsius S24-240.

Edited November 26, 2020 by Voxman
addition of text

[Gambit21]

19 hours ago, chiliwili69 said:

 

The current GPU test is also OK, but the only problem is that it needs to be done in 4K monitors (or TVs are also valid). At not all people have 4K monitors. (although probably all have 4K TVs in their homes)

 

No, most people do not have 4K TV’s at this point.

[chiliwili69]

4 hours ago, Voxman said:

The quick failure is the higher voltage setting.  I've been rooting around the net to see what issues there are with QVL and such.  Patriot's website does not have a QVL for the i7-10700K, but does for my motherboard.  My motherboard does not have QVL for the Patriot memory I'm using...much confusion.

 

The XMP profiles are from Intel. You can download an Excel file from Intel webpage if you google "Intel XMP profiles". xmp-memory-for-intel-core-processors-datasheet-20200731.zip

 

These are tested in a number of RAM vendors and Mobo vendors but not all possible combinations of them. The high freq profiles (around 4000MHz) uses a higher voltage (1.4 and hihger) and this might cause unstabilities in the Mobo and then you might have BSoD.

 

You can try other XMP profiles with lower freq, for example 3200MHz or 3600MHz which is also quite good.

 

The other option (instead of XMP profiles) is to do manual RAM OC. But here I am a complete ingnorant. XMP profiles are for lazy people like me (if they work).

 

Regarding CPU OC, I think it is independent from using XMP profiles. You can do one or the other or both.

[Voxman]

7 hours ago, chiliwili69 said:

 

The XMP profiles are from Intel. You can download an Excel file from Intel webpage if you google "Intel XMP profiles". xmp-memory-for-intel-core-processors-datasheet-20200731.zip

 

These are tested in a number of RAM vendors and Mobo vendors but not all possible combinations of them. The high freq profiles (around 4000MHz) uses a higher voltage (1.4 and hihger) and this might cause unstabilities in the Mobo and then you might have BSoD.

 

You can try other XMP profiles with lower freq, for example 3200MHz or 3600MHz which is also quite good.

 

The other option (instead of XMP profiles) is to do manual RAM OC. But here I am a complete ingnorant. XMP profiles are for lazy people like me (if they work).

 

Regarding CPU OC, I think it is independent from using XMP profiles. You can do one or the other or both.

 

Thanks for the tips.  Will play around with settings to optimize. 

[Jaegerschnitzel]

On 11/15/2020 at 10:38 PM, chiliwili69 said:

These are my results:

 

CPU test
Frames: 4933 - Time: 60000ms - Avg: 82.217 - Min: 72 - Max: 112

 

GPU test
Frames: 5618 - Time: 60000ms - Avg: 93.633 - Min: 77 - Max: 109
 

VR test with Index at 90Hz, 100%SteamVR-SS (2016x2240x2=9MillionPixels)
Frames: 2911 - Time: 60000ms - Avg: 48.517 - Min: 44 - Max: 68

 

 Motherboard: MSI Z97M-Gaming
 CPU:                 4790K
 CPU Freq:        4.8 Ghz
 L3 cache:        8 MB
 Cores:               4 
 Threads:           4 
 RAM type:        DDR3
 RAM size:        16 GB
 NB Freq:          4000 MHz
 RAM Freq:        2933 MHz 
 RAM Latency: 12 
 GPU:                 1080Ti

 

Anyone with a new Intel/AMD CPU or NVIDIA/AMD GPU???

Previous Intel/AMD CPUs or NVIDIA/AMD GPUS are also very valuable to check if new hardware is worth the upgrade.

Post your test results here and I will update the table.

 

Here are some more results. This time from a PC with relatively new, but medium classed hardware:

 

ASUS Prime H410M-K

 

i5-10600 (non-K, not overclocked, simply out of the box and no options ever changed)

Freq jumping between 4.4 GHz and 4.8 GHz

L3 Cache 12 MB

6 Cores

12 Threads

 

16 GB DDR4

NB Freq: 4100 MHz

RAM Freq: 2124 MHz

RAM Latency: 15

 

RTX 3070

 

These are the results:

 

CPU Test:

Frames: 5015, Time: 60000, Min: 72, Max: 112, Avg: 83.583

 

GPU Test:

Frames: 6965, Time: 60000,  Min: 95, Max: 141, Avg: 116.083

 

VR Test with Reverb G2:

Frames: 3199, Time 60000, Min: 44, Max: 67, Avg: 53.17

 

I think the "untouched" i5-10600 is somewhat of a modern counterpart to your overclocked i7-4790K. I´d love to add VR stats, but I don´t own a headset now. However, I have been thinking about buying a headset for a few month now and there are mainly two options: 1.) Buying a used but always well treated Index from a friend of mine for ~700€ or 2.) Ordering a brandnew G2 (arrival would be in January, I guess). Tough decision to be made ?

 

Thank you all for making these stats! Even if I have not made a dicision which VR-headset to buy, all these stats are realy helpful.

 

EDIT: Got a Reverb G2 and added VR results (06.11.2020)

Edited December 6, 2020 by Jaegerschnitzel

[chiliwili69]

8 hours ago, Jaegerschnitzel said:

GPU Test:

Frames: 6965, Time: 60000,  Min: 95, Max: 141, Avg: 116.083

Thanks you for your tests.

It is the first 3070 tested here. The results is quite good. I assume you used a 4K monitor right?

 

For the CPU result is quite aligned. I assume the clock speed was mantained at 4.8GHz during the test. (The auto mode do that if temps are OK).

 

Regarding the VR, I think it would be better to go directly to the G2, but if you don´t want to wait until January perhaps trying the Index for two months is not bad at all. And probably selling it when you get the G2.

I a couple of week I should recieve my G2, so let´s see which one I keep. Other people who already recieved the G2 and had an Index prefered the G2.

[Jaegerschnitzel]

7 hours ago, chiliwili69 said:

Thanks you for your tests.

It is the first 3070 tested here. The results is quite good. I assume you used a 4K monitor right?

Yep, I used my 4k-TV in the livingroom and connected it to the RTX 3070 directly via HDMI (no TV-receiver or so in between).

 

7 hours ago, chiliwili69 said:

 

For the CPU result is quite aligned. I assume the clock speed was mantained at 4.8GHz during the test. (The auto mode do that if temps are OK).

You are right, during the CPU test the CPU stayed constant at 4.8 GHz.

 

7 hours ago, chiliwili69 said:

 

Regarding the VR, I think it would be better to go directly to the G2, but if you don´t want to wait until January perhaps trying the Index for two months is not bad at all. And probably selling it when you get the G2.

I a couple of week I should recieve my G2, so let´s see which one I keep. Other people who already recieved the G2 and had an Index prefered the G2.

I ordered a G2 today and while waiting for the headset to arrive, I will keep an eye on other peoples reports on the G2, esp. on the tracking (I want to use the headset for games like HL Alyx or Boneworks, too). In fact, the only reason why I didn´t buy an Index till now (besides the high price tag of course), is, that I´m a little afraid that the Index´ overall display quality might give me a hard time reading gauges, spotting/identifying opponents etc... Nevertheless, thanks for your opinion! I´m Looking forward to your first impressions on the G2.

[dburne]

1 hour ago, Jaegerschnitzel said:

Yep, I used my 4k-TV in the livingroom and connected it to the RTX 3070 directly via HDMI (no TV-receiver or so in between).

 

You are right, during the CPU test the CPU stayed constant at 4.8 GHz.

 

I ordered a G2 today and while waiting for the headset to arrive, I will keep an eye on other peoples reports on the G2, esp. on the tracking (I want to use the headset for games like HL Alyx or Boneworks, too). In fact, the only reason why I didn´t buy an Index till now (besides the high price tag of course), is, that I´m a little afraid that the Index´ overall display quality might give me a hard time reading gauges, spotting/identifying opponents etc... Nevertheless, thanks for your opinion! I´m Looking forward to your first impressions on the G2.

 

I am currently going through another play through of Alyx with my G2.

While the controllers I would not consider on par with the Touch controllers, they are sufficient. Not having any major problems with then in Alyx.

[robbiec]

Still tuning mine up but here we go

Motherboard: Asus Crosshiar VIII Hero X570 (Bios 2702)

CPU: AMD Ryzen R9 5950X

CPU Frequency: 3.6Ghz

Level 3 Cache: 64MB

Cores: 16

Threads: 32

Ram Type: DDR4

Ram Size: 2 x 8GB

Uncore Frequency: 1800

Ram Frequency: 2 x 3600MHz

Ram Latency: 16

GPU: AMD Vega 64

 

CPU Test

2020-11-30 22:50:14 - Il-2
Frames: 6077 - Time: 60000ms - Avg: 101.283 - Min: 90 - Max: 143

 

Note: Bios is relatively untouched besides memory settings. PBO Enabled +200, RAM = 1.4V, SB = 1.1, CPU = Auto

 

 

I'm curious about discrepancies between frame count on the various runs, is this an indicator of overall performance? Ie the faster you are the more frames you consume or something else? 

 

 

 

Il-2 2020-11-30 22-50-14-40 fps.zip

Edited November 30, 2020 by robbiec

[chiliwili69]

2 hours ago, robbiec said:

AMD Ryzen R9 5950X

Hey, Thanks! this is the first top Zen3 CPU tested here. But I believe you should achieve more.

You say the CPU freq is 3.6GHz, is was like that during the test?  (you can run MSI Afterburner in background while you run the test and record in a chart your CPU freq or core freqs).

 

Have you tested you CPU with CPUuserbenchmark page. Just to check everything is as expected.

 

The discrepancies on the various runs should be in the order of +-1fps on average. If it is more someting weird is happening.

[robbiec]

Ref the discrepancies, it is some here have overall frame counts of about 5 to 6000 but others have recorded frame counts in the 7000 bracket, wondering why that is?

The result I've shown above will return the same, I've ran it about 5 times and +-1fps. To improve I'll need to lock in an all core overclock. A IL2 bios profile ?

Core clock was bouncing all over the place, when I checked CPUID, that was the figure but it was boosting to 5125 or thereabouts. I was logging the fraps run simultaneously with HWINFO so I'll throw that up too. 

Edited December 1, 2020 by robbiec

[chiliwili69]

6 hours ago, robbiec said:

Ref the discrepancies, it is some here have overall frame counts of about 5 to 6000 but others have recorded frame counts in the 7000 bracket, wondering why that is?

 

you test counted 6077 frames in total, so for a 60 second track, this is 6077/60=101.28 frames/second or avg fps. Other CPUs had higher rates because they were working better. I don´t know why your test is not on pair (or above) other Ryzens zen3. Perhaps the freq is fluctuating up and down. Yes, try a bios profile for IL-2.

 

If your 5 runs return similar numbers (+-1fps) it means that it is something repteable and consitent, which is good.

[Jaws2002]

17 hours ago, robbiec said:

Still tuning mine up but here we go

Motherboard: Asus Crosshiar VIII Hero X570 (Bios 2702)

CPU: AMD Ryzen R9 5950X

CPU Frequency: 3.6Ghz

Level 3 Cache: 64MB

Cores: 16

Threads: 32

Ram Type: DDR4

Ram Size: 2 x 8GB

Uncore Frequency: 1800

Ram Frequency: 2 x 3600MHz

Ram Latency: 16

GPU: AMD Vega 64

 

CPU Test

2020-11-30 22:50:14 - Il-2
Frames: 6077 - Time: 60000ms - Avg: 101.283 - Min: 90 - Max: 143

 

Note: Bios is relatively untouched besides memory settings. PBO Enabled +200, RAM = 1.4V, SB = 1.1, CPU = Auto

 

 

I'm curious about discrepancies between frame count on the various runs, is this an indicator of overall performance? Ie the faster you are the more frames you consume or something else? 

 

 

 

Il-2 2020-11-30 22-50-14-40 fps.zip 299 B · 1 download

That's strange.  

 

I was expecting the 5950x to at least keep up with the zen 3 pack. 

 Now i'm puzzled. Is it the memory, is it the low base clock of the 16 core CPU compared to the other Zen3  chips? 

 That's a head scratcher. 

[R3animate]

31 minutes ago, Jaws2002 said:

That's strange.  

 

I was expecting the 5950x to at least keep up with the zen 3 pack. 

 Now i'm puzzled. Is it the memory, is it the low base clock of the 16 core CPU compared to the other Zen3  chips? 

 That's a head scratcher. 

 

There's a number of potential issues that could be causing this:

 

1. There are still a lot of b550/x570 bios issues that need to be fixed

2.  PBO is known to be buggy/not working properly on Ryzen 5k right now, I also saw performance loss with PBO+200 in IL2

3. Single-rank-8gb dimms x2 (for dual rank, vs 4x8 or 2x16 for quad rank), but this is a small impact (up to 5%)

4. CCD/CCX latency -- remember, on the 5900 and 5950, the second batch of cores have to communicate through the I/O die but this should only be a small impact.

5. You still might be GPU bottlenecked on low with a vega64, my tests were with a 1080 on my 5800x and I fully intend to run the cpu test again when I eventually find a 3080.

 

 

 

[Jaws2002]

1 hour ago, R3animate said:

5. You still might be GPU bottlenecked on low with a vega64, my tests were with a 1080 on my 5800x and I fully intend to run the cpu test again when I eventually find a 3080.

 That may be the case. 

Robbiec, you could easily check if you are bottlenecked by the GPU. Just run the benchmark again, with afterburner on. 

If you see the GPU running at 100% it means the gpu is still the bottleneck.

 

 I was checking a multi cpu bemchmark on FS 2020 and the 5950x was at the top of the pack. That's why i'm scratching my head. Could very well be the graphics card.

Edited December 1, 2020 by Jaws2002

[robbiec]

I'm just happy to get it at a level that is usable. I doubt I'll be able to reach the same level as a 5800 tuned to the limit as I have double the cores and double the heat to handle but if I have other stuff running then I can just say, here you go, another 8 Cores available Slight tradeoff but one I knew I was making getting the 5950X. I will investigate bumping one of the CCDs higher. A lot of variables at play. Extra mem due in a few days, get that dialled in and take from there. 

The Vega64 does have its usage lights all on like KITT from Knight rider so it is under load at 1080 hence me scouting around for a 6800, 6900 or even a 3080 - I let my Vega Viii go a few months back not expecting too much supply hassle ???

Edited December 2, 2020 by robbiec

[Brandy_Lx]

Hi,

I have an older system, but I have a 5600X on his way with a new Mobo, RAM and Cooler. I guess I might have it running next weekend. In the meantime to set my baseline :

 

Motherboard:       Asus Z87-Deluxe
CPU:                      Intel 4770K
CPU Freq:             4.5 Ghz
L3 cache:             8 MB
Cores:                  4 (number of active physical Cores)
Threads:              8 (HT is on, but Il-2 forced to only run on the 4 physical cores)
RAM type:            DDR3
RAM size:            32Gb (4x8GB)
NB Freq:              4199 MHz
RAM Freq:           2400 MHz
RAM timings:      10-12-12-31
GPU:                     1080Ti

 

My Results:

 

CPU

2020-12-01 21:28:46 - Il-2 - CPU - 1920*1080 - Forced to 4 phys Cores
Frames: 4758 - Time: 60000ms - Avg: 79.300 - Min: 66 - Max: 103

 

VR

2020-12-01 22:28:25 - Il-2 - VR - Index 100%  - Forced to 4 phys Cores
Frames: 3851 - Time: 60000ms - Avg: 64.183 - Min: 51 - Max: 81

 

It seems my VR average is quite nice for that system, so I  doubled checked the settings of headset resolution and in game, but it was all as  recommended.

 

Thanks for the thread and the benchmark file !

 

 

[robbiec]

On 11/30/2020 at 11:13 PM, robbiec said:

 

 

 

Motherboard: Asus Crosshiar VIII Hero X570 (Bios 2702)

CPU: AMD Ryzen R9 5950X

CPU Frequency: 4.8Ghz

Level 3 Cache: 32MB

Cores: 8

Threads: 16

Ram Type: DDR4

Ram Size: 2 x 8GB

Uncore Frequency: 1800

Ram Frequency: 2 x 3600MHz

Ram Latency: 16

GPU: AMD Vega 64

 

Tweaked a  bit more

2020-12-02 03:44:45 - Il-2
Frames: 6479 - Time: 60000ms - Avg: 107.983 - Min: 92 - Max: 153

 

Il-2 2020-12-02 03-44-45-54 fps.zip

Edited December 2, 2020 by robbiec

[SCG_Fenris_Wolf]

You're still GPU bottlenecked, very certain of that.

[robbiec]

4 hours ago, SCG_Fenris_Wolf said:

You're still GPU bottlenecked, very certain of that.

Seem to be. I am not seeing 100% but 90% utilisation coming up if I run with the Radeon monitoring tool. 

Ref the CPU, will be interesting to see if the extra ram releases another 5%, I got it to 3600C15 last night on 2 sticks, single bank. Perhaps another 0.1 on SOC might help. Its fun trying out the various configs ?

 

 

Edited December 2, 2020 by robbiec

[chiliwili69]

17 hours ago, hellfirelx said:

but I have a 5600X on his way with a new Mobo, RAM and Cooler.

 

Thanks. You CPU and VR test are quite good, taking into account you are only at 4.5GHz. But the latency of your RAM is really good. 

Your system present and future is very similar to mines.

About your upgrade, what Mobo and RAM are you picking?

10 hours ago, robbiec said:

Ram Latency: 16

 

Thanks for the update.

I modified the instructions so now I require the timings of the RAM since it might have an impact.

 

1 hour ago, robbiec said:

Seem to be. I am not seeing 100% but 90% utilisation coming up if I run with the Radeon monitoring tool. 

 

Yeap, it seems the GPU is limiting here. I will mark the result in color.

To make your own tests and really see the gain of your RAM sticks, you can run the same test but in a lower resolution 1024x768 for example.

Edited December 2, 2020 by chiliwili69

[nickj123]

Here are my results for the 6800XT - it looks like that at 4K resolution something about the settings cripples the 6800XT - possibly the 8xMSAA? I guess the drivers probably also need further optimisation.

 

 Motherboard: Gigabyte Z390 Aorus Pro Wi-Fi
 CPU:                 9900K
 CPU Freq:        5 Ghz
 L3 cache:        16 MB
 Cores:               8
 Threads:           16
 RAM type:        DDR4
 RAM size:        64Gb (4x16GB)
 NB Freq:          4300 MHz
 RAM Freq:        3200 MHz
 RAM timings:  16-18-18-36
 GPU:                 6800XT

 

CPU Test:

Frames, Time (ms), Min, Max, Avg
  5156,     60000,  77, 121, 85.933

 

GPU Test:

Frames, Time (ms), Min, Max, Avg
  4027,     60000,  57,  77, 67.117

 

VR Test 2156x2112 Reverb 90Hz

Frames, Time (ms), Min, Max, Avg
  3509,     60000,  49,  84, 58.483

Edited December 2, 2020 by nickj123

[robbiec]

I'm getting the feeling that IL2 really likes high bandwidth mem and low latencies + plenty of IPC. On the AMD side, a single CCD model seems better than dual CCD models unless you have chilled water ?, paired with 3600 or better double ranked, low latency sticks. High Fabric helps too. Tonight I'm going after Fblk 1900! 

1 hour ago, nickj123 said:

Here are my results for the 6800XT - it looks like that at 4K resolution something about the settings cripples the 6800XT - possibly the 8xMSAA? I guess the drivers probably also need further optimisation.

 

 Motherboard: Gigabyte Z390 Aorus Pro Wi-Fi
 CPU:                 9900K
 CPU Freq:        5 Ghz
 L3 cache:        16 MB
 Cores:               8
 Threads:           16
 RAM type:        DDR4
 RAM size:        64Gb (4x16GB)
 NB Freq:          4300 MHz
 RAM Freq:        3200 MHz
 RAM timings:  16-18-18-36
 GPU:                 6800XT

 

CPU Test:

Frames, Time (ms), Min, Max, Avg
  5156,     60000,  77, 121, 85.933

 

GPU Test:

Frames, Time (ms), Min, Max, Avg
  4027,     60000,  57,  77, 67.117

 

VR Test 2156x2112 Reverb 90Hz

Frames, Time (ms), Min, Max, Avg
  3509,     60000,  49,  84, 58.483

 

What clocks do you have your 6800XT running at? (if you don't like it, can I buy it off you? ?) 

Edited December 2, 2020 by robbiec

[nickj123]

7 minutes ago, robbiec said:

I'm getting the feeling that IL2 really likes high bandwidth mem and low latencies + plenty of IPC. On the AMD side, a single CCD model seems better than dual CCD models unless you have chilled water ?, paired with 3600 or better double ranked, low latency sticks. High Fabric helps too. Tonight I'm going after Fblk 1900! 

 

What clocks do you have your 6800XT running at? (if you don't like it, can I buy it off you? ?) 

 

The tests were run with stock clocks which I think is around 2100MHz, you can overclock in the AMD app to 2500MHz without too much bother. I will be selling it at some stage if/when the 3080Ti's become available.

[robbiec]

Keep in touch! Is it an MBA model from Sapphire or Powercolor? 

[nickj123]

Yes, it's a Sapphire reference board, I'm sure the performance will improve as the drivers become more mature.

[hedger72]

Another new Ryzen benchmark for you. Stock AMD 5600X, haven't done any tweaking, save for enabling XMP profile. No GPU test as I do not own a 4K device.

 

Motherboard: Gigabyte X570 Aorus Elite
CPU:                 AMD Ryzen 5 5600X
CPU Freq:        4.6 GHz
L3 cache:        32 MB
Cores:              6
Threads:          12
RAM type:       DDR4
RAM size:        32 GB (2 x 16GB)
RAM Freq:       3600 MHz 
RAM Latency: 16-18-18-38
GPU:                 Gigabyte Gaming OC RTX 3070

 

CPU TEST

2020-12-02 16:58:20 - Il-2
Frames: 7083 - Time: 60000ms - Avg: 118.059 - Min: 99 - Max: 188

 

Edited December 3, 2020 by hedger72

[Charlo-VRde]

CPU test
Frames: 6199  Time: 60000ms  Avg: 103.317  Min: 91  Max: 140

 

GPU test
Frames: 9338  Time: 60000ms  Avg: 155.633  Min: 136  Max: 202

 

VR test with Reverb G2 at 90Hz, 50% SteamVR-SS (2168 x 2120 x 2 = 9,192,320 pixels)
Frames: 4495  Time: 60000ms  Avg: 74.917  Min: 57  Max: 91

Motherboard: MSI MPG Z490 Gaming Carbon WIFI
CPU: i7 10700K
CPU Freq: 5.1 Ghz
L3 cache: 16 MB
Cores: 8
Threads: 8
RAM type: DDR4
RAM size: 32Gb (4x8GB)
Uncore Frequency: 4801.2 MHz
RAM Freq: 4000 MHz
RAM timings: 17-17-17-37
GPU: EVGA RTX 3090 FTW3 Ultra (thank you, Mrs. Charlo)

Edited December 2, 2020 by Charlo-VR
Correction

[chiliwili69]

13 hours ago, nickj123 said:

CPU Test:

Frames, Time (ms), Min, Max, Avg
  5156,     60000,  77, 121, 85.933

 

GPU Test:

Frames, Time (ms), Min, Max, Avg
  4027,     60000,  57,  77, 67.117

 

VR Test 2156x2112 Reverb 90Hz

Frames, Time (ms), Min, Max, Avg
  3509,     60000,  49,  84, 58.483

Thanks for your tests. All is quite strange for your system.

Please, review these items:

https://forum.il2sturmovik.com/topic/34107-items-to-review-if-you-have-low-performance-in-il-2-benchmark/

 

[nickj123]

I've run the benchmarks that you recommend and performance for my system is where it should be. I think the issue with Il-2 and the 6800XT is most likely a driver issue, I've seen better results in other games and sims. I could rerun the 4K benchmark but with the card overclocked to 2500MHz?

[chiliwili69]

33 minutes ago, nickj123 said:

I've run the benchmarks that you recommend and performance for my system is where it should be

In the CPU test you get 85.9 fps. This is a low value for your 9900K at 5GHz. You should be above 100 at least. Please, verify that your CPU runs a 5.0 fixed freq set it out in your BIOS. You can verify CPU freq during the test with MSI Afterburner.

You should resolve first the CPU test value before going to the 6800XT thing.

12 hours ago, hedger72 said:

Another new Ryzen benchmark for you. Stock AMD 5600X, haven't done any tweaking, save for enabling XMP profile

 

Many thanks. Quite a good result for just 4.6GHz freq.

Regarding your RAM, it is 32Gb. it is 2x16 or 4x8?

[nickj123]

7 minutes ago, chiliwili69 said:

In the CPU test you get 85.9 fps. This is a low value for your 9900K at 5GHz. You should be above 100 at least. Please, verify that your CPU runs a 5.0 fixed freq set it out in your BIOS. You can verify CPU freq during the test with MSI Afterburner.

You should resolve first the CPU test value before going to the 6800XT thing.

 

I think the RAM speeds may be a factor in that score, I have a lot of RAM and it's not particularly fast. In the Userbenchmark test it was in the middle for the 9900K CPUs. Il-2 does seem to be heavily dependant on memory speed from what I've seen in the benchmarks. I used the Easytune app to overclock all the cores on my CPU, and to be honest I don't want to lock my CPU to 5GHz as it's only air cooled.

[chiliwili69]

9 minutes ago, nickj123 said:

Il-2 does seem to be heavily dependant on memory speed from what I've seen in the benchmarks.

Memory speed influence IL-2 performance, that´s for sure. But it alone doesnt explain your number.

Voxman went from 2200 to 4400MHz in RAM and got +8fps in the CPU test. 3200 is not a bad frequency for CL16.

I think your CPU clock might go below 5.0 during the test. You can use MSI Afterbuner running in background to monitor CPU clock and CPU temps while you run the test.

 

What CPU cooler do you have?

I would also try to put only 2x16Gb memory sticks and run only the CPU test.

[nickj123]

8 minutes ago, chiliwili69 said:

What CPU cooler do you have?

I would also try to put only 2x16Gb memory sticks and run only the CPU test.

 

It's a Noctua DH-15, so fairly big. I've just disabled HT and will give the CPU test another run....

[chiliwili69]

13 hours ago, Charlo-VR said:

GPU test
Frames: 9338  Time: 60000ms  Avg: 155.633  Min: 136  Max: 202

 

VR test with Reverb G2 at 90Hz, 50% SteamVR-SS (2168 x 2120 x 2 = 9,192,320 pixels)
Frames: 4495  Time: 60000ms  Avg: 74.917  Min: 57  Max: 91

Many thanks for these test. It is the first 3090 tested here and the first G2 as well. You have a pretty good system.

 

The 155.6 fps in 4K test it is amazing!! This is really a good card! (thanks Mrs Cahrlo!)

 

The 74 in VR is also good, although I believe that the limitation is not the GPU.

[HunDread]

10 minutes ago, nickj123 said:

Noctua DH-15

 

This is pretty much on-par with average water cooling.

[chiliwili69]

22 minutes ago, nickj123 said:

It's a Noctua DH-15, so fairly big. I've just disabled HT and will give the CPU test another run....

 

This is quite good. You should have no problems to mantain 5.0GHz.

In an 9900K with aircooler NH-D15S we have at work, we disabled 4 cores (leaving only 4) in the BIOS and were running at 5.2GHz with HeavyLoad (stress tool) and the temp was round 55 Deg C.

I think that 4 or 6 cores it quite enough for most of the games, specially IL-2.

Disabling cores in the BIOS is quite simple. And you generate quite less heat and allows higher OC.

Edited December 3, 2020 by chiliwili69

[nickj123]

OK, so I disabled HT and Speedstep and the cores are running at 5GHz but it's still averaging 85fps, so not sure what's going on but as it seems to be only Il-2 that is affected, I'm not too concerned as it could be something with the AMD card that is causing the slowdown.

 

EDIT: I just set AVX to 0 in the BIOS - no change to fps.

Edited December 3, 2020 by nickj123

[hedger72]

14 hours ago, chiliwili69 said:

Many thanks. Quite a good result for just 4.6GHz freq.

Regarding your RAM, it is 32Gb. it is 2x16 or 4x8?

 

I'm running 2 x 16GB sticks chili...I've updated my post to reflect this.

 

Oh, and I forgot to add; this is a brand new build from the ground up, new mobo, cpu, GPU, RAM, SSDs, the whole enchilada. This PC has Win10 and IL2, period. Perhaps a factor in the rather good showing at 4.6GHz.

 

Edited December 4, 2020 by hedger72