3600x for VR: Sanity Check

[Ghostrider147]

Hey all,

 

I'm considering an upgrade from my Ryzen 5 2600 to the new Ryzen 5 3600X mainly for IL-2 in VR (and a little DCS). I've compared the performance to other CPU's in this post: 

 

To me it seems like a very good option, especially compared to let's say an i5-9600k. It also means I won't have to upgrade my motherboard. Before I purchase it, however, can any of you tell my about your experiences with the R5 3600X (or other Ryzen 3000 series CPU's) in IL-2? What is the performance like in VR?

My Current R5 2600 is okay but in CPU-intensive moments (mainly lots of aircraft) it has trouble keeping the framerate at 40+, which is annoying as I like to use my Rift S at 40 FPS with ASW enabled. That means I'm left with a stuttery experience on online servers and in career mode. I just wanna make sure before I throw the money at it  I'm planning on using it with 16 GB DDR4-3200 RAM (CL-16). My current system specs can be found in my signature, for reference. 

Thanks in advance and happy 2020 ? 

[SCG_Fenris_Wolf]

The Ryzen 5 3600X won't be much better, you will also get bad FPS with it in VR / IL-2. 

 

Low single core boost clock (you'll want 4.8GHz, multithreading is mostly irrelevant), and the biggest problem is the thing works bad with its handbrake on (AVX instructions - which IL-2 VR calls).

 

Mark my words, you'd be underwhelmed, so don't do it. Better go with Intel's chips, one which you can get to 4.9GHz or better yet 5GHz on a single thread. The overclocked  i5-9600k will be infinitely better for this task.

[Ghostrider147]

6 minutes ago, SCG_Fenris_Wolf said:

The Ryzen 5 3600X won't be much better, you will also get bad FPS with it in VR / IL-2. 

 

Low single core boost clock (you'll want 4.8GHz, multithreading is mostly irrelevant), and the biggest problem is the thing works bad with its handbrake on (AVX instructions - which IL-2 VR calls).

 

Mark my words, you'd be underwhelmed, so don't do it. Better go with Intel's chips, one which you can get to 4.9GHz or better yet 5GHz on a single thread. The overclocked  i5-9600k will be infinitely better for this task.

 

Yeah I've been considering the 9600k as a possible upgrade too, I think I'm a little overwhelmed by the number of choices we have today lol. Thanks for the advice!

[SCG_Fenris_Wolf]

 

Addendum:

What you quoted, chiliwili's thread to gather 2D performance, is not representative for VR. I know he said so, but I disagree, because from 2D to VR the bottleneck switches from GPU to CPU single thread, and the entire pipeline behaves different.

 

There have only been 3 guys providing information on VR, of which 1 guy has an R5 3600X. A a single column which means no actual statistical data is readable yet, and the result strongly contradicts all tests made earlier and what we know about how a 4.25GHz chip would actually behave. With no mentions of a handbrake. 

 

With the second chip you mentioned, you'd be much more on the safe (and faster) side from all I know. Whatever you pick, good luck with the silicon lottery ??

Edited January 2, 2020 by SCG_Fenris_Wolf

[Jaws2002]

13 minutes ago, SCG_Fenris_Wolf said:

The Ryzen 5 3600X won't be much better, you will also get bad FPS with it in VR / IL-2. 

 

Low single core boost clock (you'll want 4.8GHz, multithreading is mostly irrelevant), and the biggest problem is the thing works bad with its handbrake on (AVX instructions - which IL-2 VR calls).

 

Mark my words, you'd be underwhelmed, so don't do it. Better go with Intel's chips, one which you can get to 4.9GHz or better yet 5GHz on a single thread. The overclocked  i5-9600k will be infinitely better for this task.

 

 

While i agree that Performance in game won't improve much, The rest of your post is speculation. I don't have VR but i did play with the 3600x for about a month. You will get about zero performance advantage by switching to Intel I59600k. As a matter of fact, I  got the highest score of all systems running 1080Ti with Ruzen 3600x and Ryzen 3950x. He already has the motherboard and everything all he has to do is swap the chip. The game uses just six cores and can't see the difference between 3600x,  3950x, or any other new chip for that matter. 

Edited January 2, 2020 by Jaws2002

[SCG_Fenris_Wolf]

As you said, you don't have VR.

 

No, it is not speculation. I know. We ran so many tests on that that I know.

 

The game also does not use 6 threads. It has a large main which swaps along the threads, peaks and hops to the next. It may appear that it uses all, but that's the hopping. The game also strongly deviates in its behavior in VR vs 2D, so you cannot pull any information about VR from looking at 2D. For example, the load on the main almost doubles. AVX is called as well (which chokes and beats a Ryzen to its death very quickly). 

 

Take it or leave it, I'm not going to argue about any fanboyisms. I don't care about brands, just facts. And it is indeed just an (unfortunate) fact about the VR in IL-2 (and DCS, if you care about that). 

 

The Ryzen is better with other tasks, and good in 2D indeed. But switch to VR, and you'll be crying havoc at AMD, or basically any VR game that wasn't built for VR from the ground up, but ported instead.

Edited January 2, 2020 by SCG_Fenris_Wolf

[Jaws2002]

1 minute ago, SCG_Fenris_Wolf said:

As you said, you don't have VR.

 

 

No, it is not speculation. I know. We ran so many tests on that that I know. Take it or leave it, I'm not going to argue about any fanboyisms. It is just an (unfortunate) fact about the VR in IL-2 (and DCS, if you care about that).

 

 

That's the reason I din't bother with VR. It's not there yet. Everything is still changing too rapidly and game engines are not updated to properly use it, without crippling performance. Throwing all the extra computing work load on a single core and calling it "integration" proves VR ( or VR in this game), is not there yet. 

  

 

[SCG_Fenris_Wolf]

I've been running it at mostly 90Hz since late 2017 in VR, on an old, overclocked, watercooled i7 7700K @4.9GHz. Break through 4.8GHz with AVX and you're set. In high end headsets like the Pimax5K+.

 

 

IL-2 is pretty damn awesome in VR. 

 

 

The Devs did an amazing job with it. Even in the confines that the quite old engine has given to them, they constantly upgrade and tweak it.

 

 

It's been there since 2017. I have an LG 65" c8 4K OLED TV as an alternative and it holds no ground to VR, trust me, really! ? Different tool, different job.

Edited January 2, 2020 by SCG_Fenris_Wolf

[ICDP]

My experience on a 3900X with RTX2080 and Pimax 5K+ is that you will get playable FPS from 40FPS - 90FPS depending upon the scenario.  This is me using Ultra graphics, HDR off, SSAO off, high shadows, high mirrors.  I'm very lucky that I can play VR even down 40 FPS and not have it bother me.

 

Remember the latest Ryzen CPUs have ~10 - 13% higher IPC than equivalent Intel CPUs.  So comparing clock speeds is like comparing apples and oranges.  I say in theory because it really will depend on the game.  I went from a 2700X to a 3900X and saw zero FPS improvement in Il-2 using my above setup (similar to yours), but that is because a Pimax 5K+ has such high resolution it reaches a GPU bottleneck long before the CPU becomes an issue.

 

You want a bigger FPS jump, get a 2080Ti because in VR the GPU is a much bigger bottleneck than any CPU from the past 3 years.

Edited January 2, 2020 by ICDP

[Gomoto]

If you build for IL2 only, 3600X is not your best option. For everything else you want to do with your pc probably it is. ?

Edited January 3, 2020 by Gomoto

[chiliwili69]

7 hours ago, SCG_Fenris_Wolf said:

What you quoted, chiliwili's thread to gather 2D performance, is not representative for VR. I know he said so, but I disagree, because from 2D to VR the bottleneck switches from GPU to CPU single thread, and the entire pipeline behaves different.

 

Fenris just for clarification:

 

1.- Previous benchmarks in VR (Balapan, etc) shows that VR performance depends on both CPU and GPU. If you have a low CPU or a low GPU there is a fps penalty.

2.- New VR devices with more resolution tend to load more the GPU, and it becames bottleneck more frequently, but the CPU is as important as before

3.- The purpose of the lastest remagen benchmark was to determine the best CPU/RAM for monitor, taking the GPU out of the equation. And I think it reaches that goal.

 

If we then are able to determine the optimum CPU/RAM combination for monitor, my "theory" is that those top CPUs/RAMs will be also optimum for IL-2 VR since they execute the same game code but twice (one for every eye).

 

Once we have the best CPU/RAM, then, we can add now the VR device, the GPU, the SS, the panel freq, the ASW, the VR software, etc and play/tweak with all that to reach a pleasant experience.

Edited January 3, 2020 by chiliwili69

[SCG_Fenris_Wolf]

 Guys, your Ryzens in all honors - but the OP asked a specific question:

Quote

mainly for IL-2 in VR (and a little DCS)

and received a precise answer aimed at that question. I did state that for other tasks, the Ryzen may work better.

 

 

3 minutes ago, chiliwili69 said:

If we then are able to determine the optimum CPU/RAM combination for monitor, my "theory" is that those top CPUs/RAMs will be also optimum for IL-2 VR since they execute the same game code but twice (one for every eye).

 

Once we have the best CPU/RAM, then, we can add now the VR device, the GPU, the SS, the panel freq, the ASW, the VR software, etc and play/tweak with all that to reach a pleasant experience.

Appreciate your response. Please know that everything is already clear to me. See, the bottlenecks providing geometry don't scale though, they're a hard limiter and not linear to performance like a GPU would be, which would scale the fps upwards, depending on resolution, as long as that beforementioned bottleneck remains cleared. But that Ryzen will send the frametime bouncing all over the place instead, whether you use high or low supersampling, whether you got an RTX2080ti or a lesser evil. The differentiation between frametime and fps here is intended.

 

So the issue is to find clearance for a CPU for IL-2's VR mode. IL-2's monitor mode is hence not relevant to IL-2's VR mode regarding that CPU issue. Here is why: In 2D the CPU has so much overhead that it's not touching a bottleneck yet and only the GPU scales resolution and fps. I am aware of the increased IPC on the 3600X compared to the old i7 cucumber I am using, and I am certain it is providing a better service in the majority of other applications - but it doesn't break IL-2's bottleneck, and neither holds a candle to this old 7700K here in IL-2's VR nor in other ported VR titles like DCS, which is the other game he has asked about. Hell, chili's 4790K probably outperforms the 3600X in that regard.

 

I do not think we have a disagreement. For quite a few titles, the 3600X is more bang for your buck. It bucks you off its back playing IL-2 in VR though, and the man will call bloody murder for the bucket.

 

[dburne]

Yet it all remains very fluid and ever changing...

 

Just fly and have fun I say !!

Edited January 3, 2020 by dburne

[SCG_Fenris_Wolf]

@dburneIf I see your CPU in your signature I am the man calling bloody murder though

[ICDP]

I don't remember you ever running a Ryzen 3600X CPU so you can substantiate those claims Fenris?  Apologies if I'm wrong.  Your problem is that you keep claiming and exaggerating that AMD CPUs are absolutely destroyed by Intel in Il-2 VR.  In this thread I have seen you use phrases like "Intel is infinitely better", or "AVX chokes and beats a Ryzen to its death very quickly".

 

Is Intel faster when you can get a CPU that hits 4.8 - 5GHz, yes.  Is it massively faster as you seem to claim, NO.

 

I have played Il-2 VR on the following PCs over the years.

 

4770K @4.8GHz

Fury X and Vega 64

 

6700K @4.8Ghz.  (A whopping 10% increase in Il-2 VR FPS over the 4770K and left me with a horrible anti Intel and anti mainstream review site taste in my mouth).

Vega 64

 

Ryzen 1700X (stock)

Vega 64

 

Ryzen 2700X

Vega 64 & RTX 2080

 

Ryzen 3900X

RTX 2080

 

I went from CV1 to Odyssey to Pimax 5K+ over the same period.

 

When I compared my 1700X against my 6700K in Il-2 VR the difference was indeed substantial by ~20% better on Vega64.  Despite this I used the 1700X as my main PC because the 1700X killed it in real world productivity.  When I replaced the 1700X with the 2700X it gained ~10% - 15% in VR which brought it roughly 5-10% from 6700K at 4.8GHz using a Vega 64 and an Odyssey HMD.  When I bought a Pimax 5K+ it was clear the Vega 64 was a bottleneck and I upgraded from Vega 64 to RTX 2080 and saw a ~ 50%+ increase in FPS.

 

I eventually upgraded to a 3900X PC and kept the 2700X PC as a spare (now my son's).  As stated above upgrading from 2700X to Ryzen 3900X gave zero improvement in Il-2 VR.  This is because with a Pimax 5K+ and RTX 2080 the GPU is the bottleneck.

 

Long story short, the difference in Il-2 VR between current Gen Ryzen and Intel is negligible when the GPU bottleneck is removed and from reading your posts you have nothing but assumptions on how a Ryzen 3600X handles VR compared to Intel.  You certainly aren't doing your argument any favours by exaggerating that AMD is essentially being murdered by Il-2 VR.

 

If I were pure Il-2 VR only I would save money and get a cheaper CPU and invest the difference in a better GPU.

Edited January 3, 2020 by ICDP

[ICDP]

OP, hopefully someone with an i7 Intel CPU at ~4.8 - 5GHz and similar GPU and Rift S can do a comparison benchmark run.

[SCG_Fenris_Wolf]

1 hour ago, ICDP said:

I don't remember you ever running a Ryzen 3600X CPU so you can substantiate those claims Fenris?  Apologies if I'm wrong.

You don't need to apologize, I did indeed and would be worried if you remembered me running IL-2 VR, as that'd be stalking ?. On one of my (20 years younger) brother's computer by the way. Around half an afternoon, average FPS where lower than my old turd of an i7, which can hold 90fps in MP.

 

But that is beside the point.

 

You haven't benchmarked that CPU in VR yet @ICDP. Only your R-2700X benchmarks are in chiliwili's list  https://docs.google.com/spreadsheets/d/1gJmnz_nVxI6_dG_UYNCCpZVK2-f8NBy-y1gia77Hu_k/ and in Samuel3.0 they average in 44.6FPS, while the last I posted in Balapan2.0 were averaging at 86.5FPS. Ceteris paribus in environments that removed the GPU stress, which was the whole purpose, to see the CPUs. I'd be interested in your 3900X VR benchmark though @ICDP.

 

In Remagen, I see one R5 3600X by Gomoto, and its results were underwhelming. 60FPS average with a low of 44FPS! Hello? Trying to get Shady, who is interested in IL-2's and DCS' VR, to buy one of these - that's downright evil. What's the reason, what is the justification?

 

Honestly, why try to get him to spend his money like that, when there are much better options for his case? 

 

You go tell us.

 

And don't forget to benchmark that 3900X, I'm really interested in that one.

 

 

Edited January 3, 2020 by SCG_Fenris_Wolf

[Ghostrider147]

Hey all, massive thanks for the advice. I think I'll wait, safe some money and switch to an Intel-based system in the near future! Salute!

[ICDP]

Unfortunately I can't benchmark the 3900X in VR any more as I sold the Pimax 5K+ while I wait for my 8KX.

 

My Odyssey results in that spreadsheet were with ASW enabled and the minimums were 40FPS on a Vega 64.  At the time there was no way to turn off ASW on WMR devices.  Considering 5GHz Intel chips with 108Ti's were getting similar minimums on lower resolutions with Rift and CV1s, I would call that more than acceptable.

 

The other 2700X in that lists shows 44FPS avg on a Rift and I think that should be doing much better.  Similarly I suspect any result at ~45FPS is with ASW on.

59 minutes ago, =F4U=Shady147_VR said:

Hey all, massive thanks for the advice. I think I'll wait, safe some money and switch to an Intel-based system in the near future! Salute!

 

In mid 2020 Ryzen 4 should be out and will be compatible with your existing motherboard.  Could be a cheaper upgrade path if they can squeeze another 200MHz out of the turbo clock along with IPC improvements.

 

Can I ask what you get running Chili's bench in VR on your current system?

1 hour ago, SCG_Fenris_Wolf said:

In Remagen, I see one R5 3600X by Gomoto, and its results were underwhelming. 60FPS average with a low of 44FPS! Hello?

 

Right above Gomoto 3600X VR result is an identical HMD (Reverb) with identical settings, lower GPU, on a 5GHz i7 9600K and it gets ~15% less FPS compared to the 3600X in VR.  His 3600X is only ~10% slower than an i7 8700K from Goblin's tests at similar settings, with a Reverb HMD and GPU (1080Ti)

This demonstrates my point well, the delta between a modern Ryzen CPU compared to even highly overclocked Intel is not substantial.  You are better getting a faster GPU.

 

Here are the VR results from that spreadsheet.  Contrary to your claim Gomoto's 3600X is holding up quite well in VR compared to the highly overclocked Intel CPUs.

IL2 v.	name	MotherBoard	CPU	Active Cores	HT	CPU GHz	L3 Cache MB	GPU	VR HMD	Freq Hz	SteamVR SS	frames	min	max	avg
4.003	Goblin	Gigabyte Aorus Z370 Ultra Gaming	i7-8700K	6	Off	5	12	1080Ti	Reverb	90	100	12164	49	91	67.6
4.003	Goblin	Gigabyte Aorus Z370 Ultra Gaming	i7-8700K	6	Off	5	12	1080Ti	Reverb	90	100	11801	47	91	65.6
4.002	71st_AH_statuskuo	ASUS ROG Strix Z390-H	i5-9600K	6		5	9	1080	Reverb	90	100%	9372	42	91	52.1
4.003	Gomoto	ASROCK B450 Pro4	R5 3600X	6		4.25	2x16	1080Ti	Reverb	90	100%	10863	44	91	60.35
4.003	chiliwili69	MSI Z97M-Gaming	i7-4790K	4	Off	4.8	8	1080Ti	Index	80	100%	8060	31	81	44.8

 

1 hour ago, SCG_Fenris_Wolf said:

Trying to get Shady, who is interested in IL-2's and DCS' VR, to buy one of these - that's downright evil. What's the reason, what is the justification?

 

Honestly, why try to get him to spend his money like that, when there are much better options for his case? 

 

You go tell us.

Because he would need to buy a motherboard and CPU when switching to Intel and the move may not be the holy grail you assert.  Hence my advice to go with a better GPU instead of wasting a lot of money on a CPU/MB upgrade.

Edited January 3, 2020 by ICDP
Left out GPU from table

[Ghostrider147]

20 minutes ago, ICDP said:

In mid 2020 Ryzen 4 should be out and will be compatible with your existing motherboard.  Could be a cheaper upgrade path if they can squeeze another 200MHz out of the turbo clock along with IPC improvements.

 

 

I'll wait for a while and see what Intel and AMD release in the coming months I guess. I mean, I don't have too much time and I usually fly in quick missions for which my R5 2600 is more than sufficient

 

23 minutes ago, ICDP said:

Because he would need to buy a motherboard and CPU when switching to Intel and the move may not be the holy grail you assert.  Hence my advice to go with a better GPU instead of wasting a lot of money on a CPU/MB upgrade.

 

I currently use a 2070 Super, and according to Oculus Tray Tool (OTT) my GPU is fine, it's the CPU that lags behind, usually when there's a lot of action going on in one scene. When I'm alone, flying around in a quick mission without any other aircraft/vehicles I easily reach 80 FPS on low/med settings or 40 FPS w/ ASW on high settings, generally (both in IL-2 and DCS). I'm not saying I'm absolutely right or whatsoever, I could be wrong. I'm just trying to explain what made me think that the CPU in my system was forming a bottleneck. 

 

28 minutes ago, ICDP said:

Can I ask what you get running Chili's bench in VR on your current system?

 

I'll try to do that later, so far I've only benchmarked CPU single thread performance and compared that to the other results

 

[ICDP]

4 minutes ago, =F4U=Shady147_VR said:

I'll try to do that later, so far I've only benchmarked CPU single thread performance and compared that to the other results

 

I would recommend you test exactly as outlined and compare to an Intel I7 overclocked to 5GHz.  This will give you a delta you can use to determine if the upgrade is worth it.

[SCG_Fenris_Wolf]

4 hours ago, ICDP said:

Right above Gomoto 3600X VR result is an identical HMD (Reverb) with identical settings, lower GPU, on a 5GHz i7 9600K and it gets ~15% less FPS compared to the 3600X in VR.  His 3600X is only ~10% slower than an i7 8700K from Goblin's tests at similar settings, with a Reverb HMD and GPU (1080Ti)

This demonstrates my point well, the delta between a modern Ryzen CPU compared to even highly overclocked Intel is not substantial.  You are better getting a faster GPU.

 

Here are the VR results from that spreadsheet.  Contrary to your claim Gomoto's 3600X is holding up quite well in VR compared to the highly overclocked Intel CPUs.

IL2 v.	name	MotherBoard	CPU	Active Cores	HT	CPU GHz	L3 Cache MB	GPU	VR HMD	Freq Hz	SteamVR SS	frames	min	max	avg
4.003	Goblin	Gigabyte Aorus Z370 Ultra Gaming	i7-8700K	6	Off	5	12	1080Ti	Reverb	90	100	12164	49	91	67.6
4.003	Goblin	Gigabyte Aorus Z370 Ultra Gaming	i7-8700K	6	Off	5	12	1080Ti	Reverb	90	100	11801	47	91	65.6
4.002	71st_AH_statuskuo	ASUS ROG Strix Z390-H	i5-9600K	6		5	9	1080	Reverb	90	100%	9372	42	91	52.1
4.003	Gomoto	ASROCK B450 Pro4	R5 3600X	6		4.25	2x16	1080Ti	Reverb	90	100%	10863	44	91	60.35
4.003	chiliwili69	MSI Z97M-Gaming	i7-4790K	4	Off	4.8	8	1080Ti	Index	80	100%	8060	31	81	44.8

 

Because he would need to buy a motherboard and CPU when switching to Intel and the move may not be the holy grail you assert.  Hence my advice to go with a better GPU instead of wasting a lot of money on a CPU/MB upgrade.

 

No, these conclusions are false. Why?

 

The two lines you mention are incomparable, you violated ceteris paribus, the first principle in scientific testing. At the given resolution the difference in GPUs is a limit. GTX1080 is insufficient for the resolution, it doesn't push so many pixels. You can easily tell by looking at the first two rows of Goblin - he is closer to ceteris paribus with having the same 1080ti GPU (rest of information on the system is missing, but hell, let's take this) and does better than statuskuo.

 

Basically, you compare for CPU/Memory, but put GPU into the equation - which warps your conclusion and falsifies it.

 

The problem is that the entire testing done here, in contrary to the older tests, are not done well. The aim, as chili stated, is to find CPU/Mem config. For this we need to take the GPU out of the equation - it warps the result, as you yourself have just seen and been led astray by. So to do that it must not be stressed.

 

 

P.S. Something else: The Reverb users must fix SteamVR's max resolution cap for the horizontal, or play at roughly 94.9% of the actual Reverb's resolution and live with some blurriness. Some probably haven't even been aware of it. That the horizontale resolution in SteamVR's video settings shows 2201 or 2205 respectively does not matter. SteamVR rendered at 2048 per eye, as its horizontal maximum is 4096. It does not go above that. The App "Windows Mixed Reality for SteamVR" has sometimes overwritten this, but often branches back into not overwriting this maximum cap.

 

Here's how to fix this for good:

Spoiler

Step by step guide to bump this up to 8192:

 

1. Open Windows' Command Prompt / Powershell / Bash

2. Change directory to your Steam folder, and in there to \steamapps\common\SteamVR\bin\win32

3. Execute vrpathreg.exe

4. It will print the config path for you

5. Go to the Config Path and backup steamvr.vrsettings , then open it with Notepad++

6. Now scroll down, and inside "steamvr" : {      you change or add the line       "maxRecommendedResolution" : 8192,  

7. Do not put it higher than 8192, the compositor doesn't like it.

8. Save and exit. Make sure no entries are in there two times. Make sure you didn't accidentally delete or overwrote anything else. 

9. Enjoy full clarity in IL-2

 

Addendum: In the past this was multiplied from 4096 by a value called RenderTarget or something similar, which would multiply on this hard cap. Not so anymore though. Led to blurriness for quite a few people. Try this and take a look. A 5% difference is probably not much to see for the Reverb users (unless they aim to supersample on top of that), but some will notice, just like some taste the different aromata in wine, and others don't. 

FYI.

Edited January 3, 2020 by SCG_Fenris_Wolf

[ICDP]

You seem to forget you are the one who brought up the results from Gomoto in that thread to "prove" the Ryzen 3600X was poor at VR.  Yet when I pointed out that the same table show a 8700K @5GHz is only 10% faster than the 3600X with the same conditions, same GPU and even same HMD run at the same settings, they become "false".  You conveniently ignore the 8700K results I also included and focus on my one point about the 1080Ti beating a 1080 despite the CPU used, as if it is the only point I was making.

 

You are attempting to twist my points out of context and ironically are labelling the very evidence you yourself linked to as "false".  You don't get to do that.

 

I'm not responding any more.  I learned a long time ago not to debate with people who can't debate honestly.

Edited January 3, 2020 by ICDP

[Jaws2002]

6 hours ago, ICDP said:

This demonstrates my point well, the delta between a modern Ryzen CPU compared to even highly overclocked Intel is not substantial.  You are better getting a faster GPU.

 

 

Exactly!

 

While i don't know how Ryzen 3000 compares against intel in VR, I know that both Ryzen CPUs i used and tested with this benchmark, (R5 3600x and R9 3950x) got the same performance in game, regardless of the substantial difference in raw power and boost speed. The game uses very little CPU and only six cores are active, So, as long as you have at least six cores any intel or AMD cpu in the last two years will do about the same and your GPU will be the deciding factor.

 

Look at this screenshot:

fallout vault 1

 

Il2 only uses six cores and about five percent of my CPU. Going to the lastes intel CPU will make no difference in game performance, with all other things equal. The only scenario intel chips get a bit of extra boost is when you use very fast memory, above 4000MHz.

  Both my Ryzen3600x and 3950x beat all other intel based systems that used the same 1080ti graphics card. Now this is with the Monitor, not VR, but i doubt VR will add enough workload to the CPU to bottleneck it. The graphics card gets hit much harder than the CPU in VR.

 

Hey ICDP I would be very interested to see your system tested with a monitor, as nobody with 2080,2080Ti  and ryzen 3000 ran the benchmark. I think you should do as good as any intel chip with the same GPU.

[SCG_Fenris_Wolf]

What's your VR headset? Oooops.

 

Literally everything you said in your post is false or irrelevant. The OP has been asking about VR, for IL-2 and DCS. Are you aware that the readout is wrong and result of the main jumping threads? Do you know what I refer to with saying the main. Are you aware that IL-2 in VR pulls your processor's AVX handbrake? Google what that means to your processor then.

 

It's not using 5% of your CPU btw. ?

 

 

You know what, there is so much assembled incompetence in here, that I'm out. Go flame away at me or certify each other in an echo chamber. Your min and avg performance/dollar still won't pass magically what even older Intel chipsets achieve in VR. It's a fact. I don't like it either. It's still a fact.

 

 

I'm glad I could save Shady from the mess presented to him here. But you know what they say when you go ask the public about anything. You'll get 10 folks to answer, 12 opinions, and if you're lucky someone is right.

 

 

Now if the developers could rebuild the engine and parse out enough action to separate threads, the entire picture would shift. But that's a hell of a task, and it's more likely hell froze over than this making a good return monetarily. 

Edited January 3, 2020 by SCG_Fenris_Wolf

[Jaws2002]

50 minutes ago, SCG_Fenris_Wolf said:

You know what, there is so much assembled incompetence in here, that I'm out. Go flame away at me or certify each other in an echo chamber. Your min and avg performance/dollar still won't pass magically what even older Intel chipsets achieve in VR. It's a fact. I don't like it either. It's still a fact.

 

 

 

You don't need to be so defensive. I was just posting my opinion based on the results in that benchmark. Remember, there were only two Ryzern based systems in that list for a long time.  Now that ICDP and Flanker posted their results, with much more powerful graphics cards, i can clearly see the top intel CPUs are better for this game. Now it's pretty much obvious i was wrong. I thought the GPU was still being the deciding factor even at 1080p resolution, but now i know i was wrong. 

   My earlier comments were not based on "fanboyism", but based on the limited information available at the time. This is my first AMD setup. I had only intel based systems until now, so don't be too quick to jump to conclusions about the reasons behind my comments. I didn't build this system for gaming, so I still consider switching to AMD a good decision. Outside gaming this thing is a monster.

 But now i know that at 1080p, in this game, the top Intel chips  have a solid advantage.

 

   

Edited January 3, 2020 by Jaws2002

[ICDP]

2 hours ago, Jaws2002 said:

 

 

You don't need to be so defensive. I was just posting my opinion based on the results in that benchmark. Remember, there were only two Ryzern based systems in that list for a long time.  Now that ICDP and Flanker posted their results, with much more powerful graphics cards, i can clearly see the top intel CPUs are better for this game. Now it's pretty much obvious i was wrong. I thought the GPU was still being the deciding factor even at 1080p resolution, but now i know i was wrong. 

   My earlier comments were not based on "fanboyism", but based on the limited information available at the time. This is my first AMD setup. I had only intel based systems until now, so don't be too quick to jump to conclusions about the reasons behind my comments. I didn't build this system for gaming, so I still consider switching to AMD a good decision. Outside gaming this thing is a monster.

 But now i know that at 1080p, in this game, the top Intel chips  have a solid advantage.

 

   

 

Yes they do as long as we are talking the 9900K at (rare overclocks) of 5.2 - 5.3 GHz and 1080p.  But Shady was specifically looking at a 9600K and the results from similar i7 CPUs show them to be only marginally better than a Ryzen 3600X at the same settings.

 

My experience with all my HMDs in Il-2  has been to take 40%-50% my 2D scores at any graphical settings and I have my rough worst case and best case VR average FPS.  A quick scan down the 2D vs VR numbers in Chili's spreadsheet here shows this to be a reasonable method.

 

https://docs.google.com/spreadsheets/d/1gJmnz_nVxI6_dG_UYNCCpZVK2-f8NBy-y1gia77Hu_k/edit#gid=1719385058

Edited January 3, 2020 by ICDP

[SCG_Fenris_Wolf]

15 hours ago, Jaws2002 said:

You don't need to be so defensive. I was just posting my opinion based on the results in that benchmark. Remember, there were only two Ryzern based systems in that list for a long time.  Now that ICDP and Flanker posted their results, with much more powerful graphics cards, i can clearly see the top intel CPUs are better for this game. Now it's pretty much obvious i was wrong. I thought the GPU was still being the deciding factor even at 1080p resolution, but now i know i was wrong. 

   My earlier comments were not based on "fanboyism", but based on the limited information available at the time. This is my first AMD setup. I had only intel based systems until now, so don't be too quick to jump to conclusions about the reasons behind my comments. I didn't build this system for gaming, so I still consider switching to AMD a good decision. Outside gaming this thing is a monster.

 But now i know that at 1080p, in this game, the top Intel chips  have a solid advantage.

 

Thank you for sighting the statistics and also this post. Takes a man to admit to that, kudos! 

 

I'd also rather have the game thread out more than in the current situation, as literally every system would more or less profit from this.

Edited January 4, 2020 by SCG_Fenris_Wolf

[ICDP]

For the record I do know and have said so many times that Intel is better in Il-2 for 2D and for VR.  So if anyone is building a new PC from scratch with Il-2 as their main use, then 100% go with Intel and OC to 4.8 - 5GHz.

 

If you already have a Ryzen based CPU such as Shady (the OP), then dropping in a 3600X would increase your Il-2 2D/VR scores for less cost as you can use your existing motherboard.

 

Ryzen 3600X ~240$ (includes cooler and fan)

 

i7 9600K $220

LGA 1151 Motherboard $100 - $150

Decent cooler to keep temperatures in check when overclocking $100 - $150

 

So for someone in Shady's position, an i7 9600K could cost ~$200 - $300 (double the cost) for ~10% better performance in Il-2.  Hence my advice to save that extra money and add towards a better GPU.

Edited January 4, 2020 by ICDP

[Fern]

I've been wondering about the CPU upgrade myself. My Ryzen 5 2600 and 5700 XT isnt cutting it anymore. FPS in Combat Box are horrible. I would hate to get the 3600x and not see an improvement.

 

Shady, whatever you end up doing please post your results.

Edited January 4, 2020 by Fern

[ICDP]

35 minutes ago, Fern said:

I've been wondering about the CPU upgrade myself. My Ryzen 5 2600 and 5700 XT isnt cutting it anymore. FPS in Combat Box are horrible. I would hate to get the 3600x and not see an improvement.

 

Shady, whatever you end up doing please post your results.

 

I assume you are using VR?  You could run the test benchmark in 2D to see how far off you are from the 3600X in that result and it will give a good indication of what you can expect.

 

I would expect a decent improvement in minimum FPS going to a 3600 (or X).  Apparantly the 3600 and 3600X are the same apart from the cooler.

Edited January 4, 2020 by ICDP

[Fern]

31 minutes ago, ICDP said:

 

I assume you are using VR?  You could run the test benchmark in 2D to see how far off you are from the 3600X in that result and it will give a good indication of what you can expect.

 

I would expect a decent improvement in minimum FPS going to a 3600 (or X).  Apparantly the 3600 and 3600X are the same apart from the cooler.

 

Yes, Im using VR. Are you talking the chiliwili69 benchmark? I'll try it.

 

I have some money burning a hole in my pocket. Either a new 2k monitor or a 3600x....

[Fern]

Just something I noticed in CPU-Z...I was running my Memory at 2133mhz instead of the advertised 3000mhz. I updated that in the BIOs...

[ICDP]

Well that's... sub-optimal ?

[Alonzo]

For the OP I think the question is “given budget $X, how should I spend the money to improve my VR experience?” If OP is happy with 40 ASW, it may be that a 3600 chip for $240 will give them a more guaranteed minimum frame rate, and thus smoothness. But if they are shooting for more like 80 FPS most of the time, well unfortunately there’s no chip on the planet that will really do that. My 8086K at 5.0ghz struggles to hit 80 in many situations. If you can eke out 5.2 or 5.3, you can maintain max frame rate in more situations. But sadly given the current hot thread in the engine, many players will need to settle for ASW or motion smoothing. So if you’re already comfortable with that, the 3600 probably will get you smooth motion smoothing (if that makes sense).

[ICDP]

Thanks Alonzo, this is good info and good sound advice.

[ICDP]

On 1/4/2020 at 6:24 PM, Fern said:

Just something I noticed in CPU-Z...I was running my Memory at 2133mhz instead of the advertised 3000mhz. I updated that in the BIOs...

 

Did you see any improvement after fixing the RAM speed?

[Fern]

Yes, it did. I wouldn't say a big difference, but it does help for sure. I wonder what another 16 GB would do.

[ICDP]

Try overclocking the RAM a touch, 3200 may be possible.

[Fern]

I dont have all the fancy cooling fans in my case. Not sure I would want to OC it.