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

[HankCzinaski]

Motherboard:   MSI B450 Tomahawk Max

CPU:                  Ryzen 5 3600
CPU Freq:         3.6 Ghz

L3 cache:          2x16 MB
Cores:               6
Threads:           12

Cooling:            Cooler Master Hyper 212 Black Edition

 

RAM:                 AData XPG GAMMIX D10

RAM type:         DDR4
RAM size:          16Gb (2x8GB)

Uncore Freq:    1600 MHz
RAM Freq:         DRAM 1600 MHz   ->   3200 MHz
RAM timings:    16-18-18-36-74

 

GPU:                  Gigabyte RX 5700 XT

 

Radeon drivers: 22.1.1

IL2 Version 4.702c

Benchmark V6

 

CPU Test 1080p:

Frames: 3963 - Time: 60000ms - Avg: 66.050 - Min: 57 - Max: 96

[chiliwili69]

14 hours ago, HankCzinaski said:

CPU:                  Ryzen 5 3600

 

Thank you for your test. It clearly shows that for playing in 1080p at 60Hz it can keep the 60fps quite well.

 

[WallterScott]

CPU test in version 4.702

 

CPU:                  Intel 12900k
CPU Freq:         5.4 Ghz

L3 cache:          30 MB
Cores:               8+8
Threads:           24

Cooling:            Сustom

 

 RAM size:        32Gb (2x16GB)
 RAM Freq:       6400
 NB Freq:          4400 MHz 
 RAM timings:  30-38-38-50 CR2
 GPU:                3090

 Win 11

 

P-core + E-core

2022-01-17 22:55:33 - Il-2
Frames: 9114 - Time: 60000ms - Avg: 151.900 - Min: 137 - Max: 200

 

E-core off HT-off

2022-01-18 00:19:07 - Il-2
Frames: 9130 - Time: 60000ms - Avg: 152.167 - Min: 136 - Max: 195

[chiliwili69]

9 hours ago, WallterScott said:

 RAM Freq:       6400
 NB Freq:          4400 MHz 
 RAM timings:  30-38-38-50 CR2

 

Thank you very much for running those extra test. It clearly show that DDR5 RAM at those speed and latencies is as good as DDR4 RAM of FoxbatRU.

 

The 12900K is clearly the new king!

[WallterScott]

3 часа назад, chiliwili69 сказал:

It clearly show that DDR5 RAM at those speed and latencies

Sorry. My mistake. The memory didn't work on the 6400.. Only for 6200.
I think DDR5 memory is now what limits 12900k. Now the latency of this kit is 6200 cl30 -55-56ns (Aida64)
It is necessary to achieve 50 or less.

[Jaws2002]

12900K  is a real monster.  Impressive jump in performance. Intel is back.  Competition is a good thing. 

[dburne]

Can you game regular at that 5.4 GHz speed or was it set just for the benchmark run?

[WallterScott]

5 часов назад, dburne сказал:

Can you game regular at that 5.4 GHz speed or was it set just for the benchmark run?

5.4 this is the minimum frequency. I even have a linx test on it.
In games, several cores boost up to 5.6-5.7. It's just a select processor (SP91) and good cooling.
The previous 12900k had an SP88. This is about the middle class. He could 5.2 all cores and 5.5 boost 2-4 cores.

[dburne]

3 hours ago, WallterScott said:

5.4 this is the minimum frequency. I even have a linx test on it.
In games, several cores boost up to 5.6-5.7. It's just a select processor (SP91) and good cooling.
The previous 12900k had an SP88. This is about the middle class. He could 5.2 all cores and 5.5 boost 2-4 cores.

 

Good to know thanks.

 

[shirazjohn]

I am hoping to upgrade my CPU to a Ryzen 7 5800x3d when they arrive on the scene (which will be last upgrade on this Mb) so thought i would run the benchmark again on the latest game version  so I have a comparison.

 

The test results are as follows:

Win 10 (tried win 11 but had an average drop of about 10 fps even with the windows and chipset update)

Game version 4.702c

 

Cpu test  average of 5 runs = Avg: 125.793   Min: 110   Max  170

 

Vr test Reverb G2 100% ss =  Avg: 69.500 - Min: 57 - Max: 84

 

Since the last test I have a slight increase in my cpu test not really sure if this is since the new cloud update or a few few bios tweaks I have made but definitely noticed an increase my fps during gameplay since the clouds update so probably this.

 

These are my game settings for a comparison to the above( I only play single player career mainly bodenplatte medium difficulty and medium frontline activity.

I achieve an average of 65fps and down to 45fps over cities. Which I find reasonably smooth with some slight stuttering low to the ground over cities.

 

Reverb G2 steam VR settings 100%ss. No reprojection

 

Overall game settings= high

shadows =ultra

mirrors=medium

dist land detail=x4

horizon 150km

landscape filter = sharp

grass=ultra

clouds=high

distant buildings,4k textures, sharpen=enabled 

No AA

No Mods

Edited February 3, 2022 by shirazjohn

[Voyager]

@shirazjohn Will be very interested to see what the 5800X3D does once you've got your hands on it. 

 

I'm in a fairly similar spot, debating if I want to grab a 5950X before the lines shut down or the 5800X3D and sit on my current backbone for a few years.

[shirazjohn]

8 minutes ago, Voyager said:

I'm in a fairly similar spot, debating if I want to grab a 5950X before the lines shut down or the 5800X3D and sit on my current backbone for a few years.

Yes exactly the same for me, I've was also thinking of the 5950x but wasn't sure if it was worth the upgrade from my 5600x, the 5800x3d looks like it may have more mileage in it. Also the same with my gpu I've been waiting for the 3080,s to come down in price but now the 4000 series is on the horizon i think I'll wait it out for a 4080.

 

I think these two upgrades should see me out for a couple of years before a complete new build.

 

[paul_leonard]

Check my results with an Alienware R10 5950x/3090.  No where near 125fps at 2K.  Some advice received here re slow RAM but I've upgraded to 3600 16/20/20/20 RAM and overclocked to 3734.  Overclocking CPU as well and touching 5100 at times.  Still no better results, in fact they've gotten worse.  I do run Win11 but that hasn't changed since started benchmarking (I knew I was going down a rabbit hole...sigh).  That said I run game time on Bodenplatte at a pretty much steady 90fps in G2 VR.  I run:

Overall game settings= ultra

shadows =normal

mirrors=simple

dist land detail=x4

horizon 150km

landscape filter = blurry (seems to be a matter of taste)

grass=normal

clouds=high

distant buildings,4k textures, sharpen=enabled (no sharpen when using FSR but HDR checked)

No AA

No Mods

 

Did learn a really hard lesson.  Using Ryzenmaster I set it up in game mode.  Game mode turns off 8 of the CPU cores!  WTF.  Game mode knocked 10fps of test/practical fps.  Still going down the benchmarking rabbit hole.  Oh my.

 

[shirazjohn]

8 hours ago, paul_leonard said:

Still going down the benchmarking rabbit hole.  Oh my.

 

I know what you mean it can get quite frustrating at time's but also very satisfying when you good results. I'm really no expert at this either i think i struck lucky with my MB, Cpu and ram combo but I spent quite a bit time trying different settings especially memory timings .

 

I found the following guide useful for tuning memory you can probably skip most on it and go to the AM4 section.

 

https://github.com/integralfx/MemTestHelper/blob/oc-guide/DDR4 OC Guide.md

 

I know its not for everybody but i enjoy tuning PC's and find it satisfying especially when it translates into actual improvements while gaming.

 

 

[chiliwili69]

On 2/3/2022 at 5:48 PM, shirazjohn said:

I am hoping to upgrade my CPU to a Ryzen 7 5800x3d

Thank you for your test. Let us know when it arrives. I would be also interested, although CPU (5600X) with the Index (80Hz) is almost never the bottleneck. But perhaps future headsets will support only 90Hz or higher.

[chiliwili69]

15 hours ago, paul_leonard said:

That said I run game time on Bodenplatte at a pretty much steady 90fps in G2 VR

 

How much SteamVR SS% did you apply?

 

if you have 90fps in Bodenplatte with the G2 then your CPU and GPU are working quite well. It is not easy to keep 90fps steady with the G2 and your settings.

 

You can run the SYNVander mission using fpsVR tool (it cost few dollars but it is worth) and visualize who is bottlenecking the track (CPU vs GPU).

 

You can also use MSIafterburner while you run the benchmark or play the game. It is really useful to trend CPU temps, CPU clock, GPU temp, GPU clock, power, etc.

 

The OpenVR Benchmark free tool in SteamVR just to check if you achieve the fps that a  5950X/3090 should achieve using OpenVR.

[paul_leonard]

70%?

I do have fpsVR and MSI Afterburner and HWInfo.   I will check out OpenVR Benchmark.  I haven't posted any results with the new RAM as there has just been too much benchmarking weirdness over the past month, and I've got to get some flying time in as well!  Thanks for advice so far.

 

[Alonzo]

Personally in-game (not for benchmarking) I run SteamVR at 100% and use the FSR mod to run the game at about 70% and upscale. If you run SteamVR less than 100% with the G2 it crushes pixels together and spotting is harder. But if you try to run 100% without the FSR mod there's no GPU on the planet that will do it! ?

[RAAF492SQNOz_Steve]

Ok, Drum roll time..........................

 

Just ran the VR2 SynVander Benchmark 6.0 test and compared the Reverb G2 and Varjo Aero results

G2 settings as per standard VR2 resolution while the Aero Resoution was set to High (i.e. 27 ppd for non foveated renders) & Steam 100% SS.

Average of three runs for both.

1. Reverb G2 average fps = 82
2. Varjo Aero average fps = 80

 

For whatever reason I had a much larger swing in G2 results (+/- 2fps) whereas Aero results were very consistent (+/- 0.2 fps)

Did note a couple of additional microstutters in the Aero run as well.

[chiliwili69]

8 hours ago, RAAF492SQNOz_Steve said:

G2 settings as per standard VR2 resolution while the Aero Resoution was set to High (i.e. 27 ppd for non foveated renders) & Steam 100% SS.

Average of three runs for both.

1. Reverb G2 average fps = 82
2. Varjo Aero average fps = 80

 

Thank you very much for running this first test comparison of G2 and Aero. I think it is very important for many people considering the Aero.

 

Your test shows that you have basically the same performance with the G2 and Aero. But with the Aero you have a much better view and details since it used more physical pixels per degree and more clear lenses.

 

For both tests you used the 100% SS SteamVR, which is what technically delivers the quality recomended by the manufacturer of the headset. 

The point is that the G2, having less physical pixels (2160x2160 per eye, so 9.3 million in total), needs a lot of pixels in the render (3162x3093 per eye, so 19.5 million in total).

Whereas the Aero, having more physical pixels (2880x2720 per eye, so 15.6 million in total, and this is what it really counts for the visual detail), needs less pixels in the render (3140x2692, so 16.9 million in total).

 

So, for both headsets at 100% SteamVR (and Aero High mode which is the default) you obtain basically same performance but much greater detail and clarity with the Aero. The nice thing of the Aero is that using miniLED displays and aspheric lenses it needs a smaller amount of internal supersampling, and this is a great thing. Otherwise we will need 4090 cards to run them.

 

To equalize the number of pixels for the results in the SYNVANDER table we normally run the test with the same total amount of pixels (chosen as 19.5, ie the G2 at 100%SS). So the VR2 test of the Aero will need to be adjusted to 116%.

 

Edited February 13, 2022 by chiliwili69

[Jaws2002]

On 2/4/2022 at 9:02 PM, paul_leonard said:

Check my results with an Alienware R10 5950x/3090.  No where near 125fps at 2K.  Some advice received here re slow RAM but I've upgraded to 3600 16/20/20/20 RAM and overclocked to 3734.  Overclocking CPU as well and touching 5100 at times.  Still no better results, in fact they've gotten worse.  I do run Win11 but that hasn't changed since started benchmarking (I knew I was going down a rabbit hole...sigh).  That said I run game time on Bodenplatte at a pretty much steady 90fps in G2 VR.  I run:

Overall game settings= ultra

shadows =normal

mirrors=simple

dist land detail=x4

horizon 150km

landscape filter = blurry (seems to be a matter of taste)

grass=normal

clouds=high

distant buildings,4k textures, sharpen=enabled (no sharpen when using FSR but HDR checked)

No AA

No Mods

 

Did learn a really hard lesson.  Using Ryzenmaster I set it up in game mode.  Game mode turns off 8 of the CPU cores!  WTF.  Game mode knocked 10fps of test/practical fps.  Still going down the benchmarking rabbit hole.  Oh my.

 

Alienware always had horrible cooling. That's not a case for a 16/32 cores CPU and a 3090. Just to test, open the case up see if the performance changes, but even with the case open, if you have the same cpu cooler like in this video you will loose a lot of performance.

 

 

Edited February 16, 2022 by Jaws2002

[paul_leonard]

Nope.  Liquid cooling.  Also tried to run with case open as mentioned in the video.  Bought another fan that you can install in an open HDD slot at front.  But you may be onto something I haven't looked at enough which is CPU temps as it still might be the issue.  I have the software tools so I will be exploring.   In real world I am very happy with performance in VR at Ultra graphic quality and eye candy.

[Jaws2002]

9 hours ago, paul_leonard said:

Nope.  Liquid cooling.  Also tried to run with case open as mentioned in the video.  Bought another fan that you can install in an open HDD slot at front.  But you may be onto something I haven't looked at enough which is CPU temps as it still might be the issue.  I have the software tools so I will be exploring.   In real world I am very happy with performance in VR at Ultra graphic quality and eye candy.

 

Hope you sort it out. It's frustrating to buy such powerful hardware and not be able to get the full potential. The temperature is very important in the long run. You may be satisfied with the performance, but if the CPU or GPU are running very hot, you can shorten their life span.

 

Good luck.

Edited February 16, 2022 by Jaws2002

[paul_leonard]

And I thought just trying to keep up with developer innovation shortened the lifespan ;>)

[-332FG-Cue-Ball]

I'm considering upgrading my graphics card soon, so I wanted to get a benchmark of my current system. It is very un-optimized. I've done no overclocking, no work with memory frame timing (as you can plainly see below), etc.  Tests run using IL-2 version 4.703c.

 

Motherboard:  ASUS TUF Gaming B550M-PLUS WiFi 6
 CPU:                Ryzen 5 5600X
 CPU Freq:       4200MHz
 L3 cache:        32MB
 Cores:              6
 Threads:          12 
 RAM brand:     Corsair Vengeance LPX 2x16GB DDR4 3600

 RAM type:       DDR4
 RAM size:        32 GB
 Unicore Freq:  1799MHz  

 DRAM Freq:     1799.6MHz

 CAS Latency:  18, 22-22-42 timing

 GPU:                EVGA RTX 2060 6GB

 NVIDIA Driver: 497.29

 

CPU test

Frames: 5996 - Time: 60000ms - Avg: 99.933 - Min: 87 - Max: 135

 

GPU test

Frames: 2094 - Time: 60000ms - Avg: 34.900 - Min: 25 - Max: 42

 

Accidentally ran the VR tests at 80Hz mode, but my frame rate is low enough it probably didn't make much difference.

 

VR Test1 with Index at 80Hz, 106%SteamVR-SS
Frames: 4051 - Time: 60000ms - Avg: 67.517 - Min: 39 - Max: 81

 

VR Test2 with Index at 80Hz, 216% SteamVR-SS

Frames: 2364 - Time: 60000ms - Avg: 39.400 - Min: 25 - Max: 42 

 

I normally run Steam VR at 148% (native res for the Index) and run IL-2 using balanced settings + low shadows, low clouds, 70km, 4K textures, no AA. With those settings I'm getting:

Frames: 2699 - Time: 60000ms - Avg: 44.983 - Min: 42 - Max: 47

 

I expect that with a 3080 and some tweaking I could probably get a 50-100% improvement at my current settings.

[chiliwili69]

18 hours ago, -332FG-Cue-Ball said:

Tests run using IL-2 version 4.703c

 

Thank you for this test. It is the first test with the version 4.703. I didn´t include with the other test from previous version 4.701 and 4.702 since I still don´t know if they are comparable. 

 

Do you run Windows 10?

What is the Product ID of your RAM memory (shown in CPU-Z memory tab)?

 

Clearly your 2060 is your main bottleneck in VR. With the 3080 you will run almost always at 80fps with 80Hz mode. (if RAM is enough fast).

 

I might run my 5600X again just to check the 4.703 version.

[1Sascha]

My Specs:
 Motherboard: MSI MAG Z690 Tomahawk Wifi DDR4
 CPU: Intel i5-12600K, slightly OCed via BIOS turbo-offset on the P-cores, cooled by a Lian Li Galahad 240
 CPU Freq: P-cores max 1080 run: 5.1 GHz, P-cores max with 4k settings in 1440p: 4.7 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 (well, 3592.2 according to CPUZ, but it's close enough )
 RAM Freq: 3600 MHz
 RAM Latency:  CL18
 GPU: MSI GTX 1060 3GB Aero ITX running latest GF-drivers, auto-OCed by Afterburner to 1911 MHz in boost-mode

OS is Win 11 Pro

 

 

 

CPU Test Results:

First Run in 1080p

2022-04-12 14:19:33 - Il-2
Frames: 4088 - Time: 60000ms - Avg: 68.133 - Min: 58 - Max: 82

 

... and since I don't have a 4K but a 1440p monitor, I used the 4K settings at that resolution:

Second Run in 1440p

2022-04-12 14:25:02 - Il-2
Frames: 3760 - Time: 60000ms - Avg: 62.667 - Min: 49 - Max: 73

 

And just for the heck of it, I reset the game to my preferences ...

 

 

 

... and ran the benchmark again:

 

2022-04-12 14:59:34 - Il-2
Frames: 4426 - Time: 60000ms - Avg: 73.767 - Min: 59 - Max: 88

 

 

 

S.

Edited April 12, 2022 by 1Sascha

[chiliwili69]

17 hours ago, 1Sascha said:

MSI GTX 1060 3GB Aero ITX

 

Thank you for running the becnhmark. It is in fact the first i5-12600K tested here.

 

But the CPU test (1080p) is heavily constrained by the 1060 card. I tried to design an IL-2 CPU benchmark where the GPU would not limit never the CPU performance, but in you particular case you have a very powerfull CPU coupled with a much less powerfull GPU.

 

In your usual graphics settings, don´t be afraid of maxing-out all settings (except mirrors, clouds and not use MSAA). Your fps will not be impacted to much since you have a powerful CPU.

Edited April 13, 2022 by chiliwili69

[1Sascha]

19 minutes ago, chiliwili69 said:

But the CPU test (1080p) is heavily constrained by the 1060 card.

Tell me about it...

 

I was worried about the discrepancy between the new system and this 1060 I still had lying around unused. But seeing how decently it still performs in all the games I've played so far, I think I'll continue to not spend a ridiculous amount on the 3060 Ti I *really* want and rather invest part of that money in a set of MFG Crosswinds.

 

19 minutes ago, chiliwili69 said:

In you usual graphics settings, don´t be afraid of maxing-out all settings (except mirrors, clouds and not use MSAA). Your fps will not be impacted to much since you have a powerful CPU.

Clouds on "extreme" still work fine with everything else unchanged. They do give me a hit of around 10 FPS compared to "high", and since I don't see too much of a difference in visual quality, I set them to "high" now and enjoy the FPS boost. What really helped was reducing the horizon draw distance and that one setting in the "game" tab re. landscape details (can't recall its exact name). That and a few other changes suggested in the graphic smoothness thread gave me a very decent uplift in offline career missions of 15 to 20 FPS.

 

Plus with higher details, I did run into VRAM issues from time to time (at least I think it was VRAM-related). Sometimes clouds would look funny for a split-second and the game would micro-freeze when POV and/or FOV changed quickly. Looked like the game was trying to shuffle textures around due to my lack of VRAM or something.

 

S.

Edited April 12, 2022 by 1Sascha

[chiliwili69]

10 hours ago, 1Sascha said:

What really helped was reducing the horizon draw distance and that one setting in the "game" tab re. landscape details (can't recall its exact name).

This is interesting. Based on previous experiments (this one and this one) I always thought that Horizon Draw Distance was something just loading the CPU but not the GPU.

The Distant Landscape Details didn´t load the CPU at all. So good to know that this two factors are adding load to the GPU.

 

For playing at monitor at 60Hz your 1060 is quite OK. As you say, better to use the money in those good MFG.

Edited April 13, 2022 by chiliwili69

[1Sascha]

1 hour ago, chiliwili69 said:

The Distant Landscape Details didn´t load the CPU at all. So good to know that this two factors are adding load to the GPU.

In fairness: I can't be 100% certain of the individual performance hits because I didn't make the changes one at a time but all of them at once (according to the "guide to graphic smoothness"-thread). IIRC I also enabled 4K textures (counterintuitive, especially with a card that's low on physical memory, but that's what the devs seem to recommend), switched off "full screen" and one or two other minor tweaks I don't remember. 

 

However: As you've stated yourself, I also suspect that a GPU this out of date from the rest of the system might act as a hindrance to even the fastest CPU. If the CPU has to wait for the GPU to finish its work, a CPU with more cores/threads, newer architecture or higher frequencies probably only will help so much before it gets "chocked" by the graphics card.

 

 

 

 

S.

[chiliwili69]

I have just run the CPU test and GPU test (4K) with the latest version 4.703c.

 

CPU test(1080p):
Frames: 6636 - Time: 60000ms - Avg: 110.600 - Min: 95 - Max: 146
Frames: 6618 - Time: 60000ms - Avg: 110.300 - Min: 92 - Max: 148
Frames: 6488 - Time: 60000ms - Avg: 108.133 - Min: 94 - Max: 145

 

GPU test (4K):

Frames: 7246 - Time: 60000ms - Avg: 120.767 - Min: 87 - Max: 145
Frames: 7291 - Time: 60000ms - Avg: 121.517 - Min: 94 - Max: 155
Frames: 7275 - Time: 60000ms - Avg: 121.250 - Min: 93 - Max: 144

 

Comparing them with the previous tests in 4.701 I obtain about 5fps less in the CPU test and 20fps less in the GPU test!!

 

I am not too worried about the CPU drop but the 20 fps drop is amazing. This means I will be more constrained by GPU in VR. I have not tested in VR.

 

I don´t know if anyone can re-run the GPU test in 4K just to confirm this. Or anyone with a 3080 can just run the GPU test in 4K.

 

Could be this due to new sky tech introduced lately?

Edited April 19, 2022 by chiliwili69

[TheSNAFU]

I have most definitely lost significant performance in vr since the last update and suspect it is due to the new sky. Nothing else added in the update would likely contribute to so much lost performance. I have struggled to find settings that offer good visuals and performance ever since. 
 

I didn’t run the benchmarks and my comments are based strictly on gameplay. Very disappointing since in vr I don’t see an improvement in the sky worth what I have lost in performance. My 3080 is constantly running at 93-99% load since the update. 

[-332FG-Cue-Ball]

On 4/11/2022 at 11:34 AM, chiliwili69 said:

 

Thank you for this test. It is the first test with the version 4.703. I didn´t include with the other test from previous version 4.701 and 4.702 since I still don´t know if they are comparable. 

 

Do you run Windows 10?

What is the Product ID of your RAM memory (shown in CPU-Z memory tab)?

 

Clearly your 2060 is your main bottleneck in VR. With the 3080 you will run almost always at 80fps with 80Hz mode. (if RAM is enough fast).

 

I might run my 5600X again just to check the 4.703 version.

Sorry for the slow reply. I'd ordered a 3080 and was waiting for it to arrive before I updated. 

 

I'm running Win10 Pro 21H2 build 19044.1645.

I don't see a Product ID for memory in CPU-Z, but I've got Corsair Vengeance LPX 2x16GB DDR4 3600 (PC4-2800) C10 sticks.

 

Reran all four tests changing from an RTX 2060 to a 3080 (EVGA XC3 Ultra 3080 12GB). Results are as follows:

 

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

 

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

 

VR Test1 (Index @ 106% SS, 90Hz):
Frames: 4870 - Time: 60000ms - Avg: 81.167 - Min: 66 - Max: 92

 

VR Test2 (Index @216% SS, 90Hz):

Frames: 4701 - Time: 60000ms - Avg: 78.350 - Min: 53 - Max: 91

 

Quite an improvement over the RTX 2060. GPU 4K text is nearly 4X as many frames. VR tests are both about 50% improvement on minimum numbers, and maxing out at the 90Hz set in Steam VR. If I set Steam VR at 144Hz mode, would likely see slightly higher maximums and averages.

[chiliwili69]

4 hours ago, -332FG-Cue-Ball said:

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

 

Thank you for this tests. It confirms similar values on the GPU test.

 

In the CPU you obtain less. I wonder why having the same CPU. (although different memory).

 

Could you post a screenshoot fo the Memory tab and SPD tab of the CPU-Z app?

[Alonzo]

Hey all. I haven't checked this thread in detail for a while, but would be interested in your thoughts on my setup.

 

I got an AMD 5800X along with some recommended G.Skill 3600-CL16 RAM, and I did do the benchmark, but I think my results are quite old now. I wasn't able to overclock the 5800X or the RAM -- I didn't invest too much time into it and it felt like fiddling with them caused crashes in day-to-day use of the PC, which is absolutely not OK due to me using the machine for work. A hard crash is a real bummer.

 

My flying use case is competitive multiplayer in VR, with a Reverb G2 (FSR mod at about 70%) and an RTX 3080 10GB. I often run into frame drops when there are lots of players in an area, especially if that coincides with low-level flying near cities.

 

Recently the new AMD 5800X3D was released, which made me think about my system again. Questions / thoughts:

• How far behind is the 5800X compared to the latest stuff? I see the 12900K with DDR5 is crushing it in the benchmarks.
• Has anyone tried the 5800X3D ? That would be a drop-in replacement for me vs a more spendy 12900K + DDR5 upgrade.
• Should I try again with overclocking? I didn't have much joy so I set everything back to vanilla. Can anyone recommend a reliable method for AMD CPU + RAM overclock?
• Is there a software overclock I can apply only when using my dedicated VR Windows install? I reboot between Windows installations, one for general use and work, and one dedicated for VR. So if I can overclock using software, I could apply that just for VR and not affect my day to day stability.

Thanks!

[shirazjohn]

5 minutes ago, Alonzo said:

Has anyone tried the 5800X3D ? That would be a drop-in replacement for me vs a more spendy 12900K + DDR5 upgrade.

 

The 5800X3d is out tomorrow i believe so i am looking to purchase and hopefully receive before the end of the week. I will be testing with chilli's benchmark and will report the results when i get the chance.

[chiliwili69]

27 minutes ago, Alonzo said:

Should I try again with overclocking? I didn't have much joy so I set everything back to vanilla. Can anyone recommend a reliable method for AMD CPU + RAM overclock?

 

I wrote this sometime ago: 

 

 

My experience with overclocking CPU with 5600X: Didn´t got any measurable effect. But I didn´t go very deep. The same with RAM, I didn´t really try too much.

 

I love the always-work XMP profiles. So, best option is always to buy Mobo and RAM is a consisten way: a high frequency and low latency with XMP profile supported by the mobo.

For example, if I would go to a new build now, I would grab the EVGA X570 Dark Mobo since it is the only Mobo for Ryzens which support a 4000Mhz CL14 out the box:

https://www.gskill.com/qvl/165/326/1620976207/F4-4000C14D-32GTZN-QVL

 

That Mobo&RAM with the upcomming 5800X3D could really go over the 12900Ks. Let´s see how the 5800X3D goes.

 

BTW, I saw your previous test with the 8086K but not with your 5800X, did you reported?

[-332FG-Cue-Ball]

16 hours ago, chiliwili69 said:

 

Thank you for this tests. It confirms similar values on the GPU test.

 

In the CPU you obtain less. I wonder why having the same CPU. (although different memory).

 

Could you post a screenshoot fo the Memory tab and SPD tab of the CPU-Z app?

 

16 hours ago, chiliwili69 said:

Could you post a screenshoot fo the Memory tab and SPD tab of the CPU-Z app?

 

[chiliwili69]

Thanks, I only wanted to check you model RAM number (CMK32GK4M2D3600C18) to be sure no other XMP profiles can be applied.

 

The only difference between your system and mine is latencies (16-16-16-36) versus (18-22-22-42), and this alone is 10 fps in the CPU test.

 

With your Asus Mobo I searched in the Gskill webpage and found some modules that are OK for you Mobo (no for my Mobo):

 

https://www.gskill.com/configurator?page=1&cls=1529635169&manufacturer=1524725352&chipset=1603955141&model=1603956830&adSearch2=Capacity§16GB (8GBx2),Tested_Speed§3600 MT/s,Tested_Latency§14-14-14-34,

 

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.