race for computer power

[Eugel]

57 minutes ago, TheFreshPrince said:

 The caveat is that DCS in 2K doesn't look that good, despite DLSS/DLAA. If you want good graphics in DCS, you need to go VR and also only play the newest modules and maps.

No. DLSS doesn´t make games look better, it´s the exact opposite ! It decreases picture quality to gain performance.
And again no. DCS looks better on a 2k screen than it does on a 4K VR Headset. Because in VR, you are looking at the screens from like 5 cm distance and looking through a magnifying glass. You will see pixels ! And most likely you will have to reduce quality settings to be able to run in VR in the first place.
So no, it definitely does not look better in VR.

On 11/25/2023 at 11:13 AM, Bounti30 said:

I had better performance with my old hardware in 2.5 than today in 2.9

And someone else had better performance with 2.0, so should they have stopped at 2.0 ? Or 1.5 ?
Increasing hardware requirements have been standard in PC gaming since PC gaming existed.

[Dangerzone]

13 hours ago, Bounti30 said:

I come back to my post after having lamented the forum for several days.
actually I don't think that the hardware is at fault but the software itself.
Many posts concerning mutlitrheading or the DCS graphics engine. Many users having problems despite having high-performance computers.
So here is my question. Wouldn't it be wise for ED to stop escalating DCS functionality and instead address the root causes of dysfunction?
Because personally I'm not interested in having photorealistic clouds or a hyper-detailed driver if the games run at 3 frames per second

Originally I disagreed with your original post, but after reading through your explanations, I think I have a greater understanding/appreciation for where you're coming from. Correct me if I'm wrong, but from what I understand your concern is 'performance creep' (or lack thereof) with subsequent versions of DCS - that as each new version is released, there is a decrease in performance even when you don't turn on, or aren't interested in the new features? 

If so, I have shared the same concerns. I would tweak with settings, lower them, and for a while I was able to work, but then with another update, I found DCS crossing that line of the visuals in VR being unenjoyable due to stutters, and I'd have to try and find if there was anything else I could squeeze to get past the jaggies. At the very end of almost walking way, OpenXR came out, and that made an epic improvement for me. Lower FPS, but yet a smoother experience. Even so, the 'performance creep' kept coming with updates. I got to a stage where I was forced to upgrade from my 2080S because I kept on getting stutters that I couldn't seem to resolve. Strangely, others on less hardware seemed to be happy even running better settings than I. I had low clouds, no shadows, no grass, no MSAA, very low settings, there wasn't much more I could see I could do. 

So yes - I have shared the same concerns. However there are two things I have to note to be fair with this.

The first is that maybe I gave in too soon. Shortly after upgrading my rig, DLSS came out. DCS made a huge step forward, and I suspect with this, I would have had similar performance to what I saw before in 2.7. I also note that Vulkan is also being developed, and hope that this will have just as much of a leap (if not more) than DLSS. So I think that ED are focusing on dealing with the root causes. However, what do you do with the other 150 developers that aren't working in this area, and it's not their area? They of course are going to continue working in their respected areas and bring more features to DCS. 

So it's worth recognising that just because other features are being released, it doesn't mean that ED aren't working as fast and hard as they can on Vulkan or MT either. It's not a matter of throwing all ~200 dev's at performance and the problem will be fixed quicker. There's only so many who can work on the problem.

It's also worth noting that other things like Tas's optimized shaders can also help. So this does prove that there is definitely more that ED could do in order to optimize performance in DCS as it is now. But again, it's important to be fair and note that they have been focused in MT, DLSS, and Vulkan, so it's not like they're doing nothing at the same time. 

My main concern, which I think might be a factor with your experience, is the Stability of Releases at the moment:

MT still has it's bugs, as does the issues with DLSS it appears. The latest stable release in my humble opinion was a bad release while there are still some significant issues that are yet to be addressed.

While I personally would love to see ED prioritize a little more getting DCS more stable before releasing to SR and before adding new features that will require more debugging, I sadly understand why they don't. It seems as though there's a rush to get the new features out that stable releases of late haven't really been bedded down as much as I would like. Almost as like "oh - that's good enough, we want to hurry up and focus on the next feature because the community is waiting" instead of "hold on - we need to finalise these issues first before releasing stable and moving on". 

However, while I have these concerns, I can see that the majority of the community doesn't share this. New features are more important than stability. How can I say this? Because actions speak louder than words, and the majority of the community have chosen to be open beta users.

As a result, and as much as I hate it, I can see it makes sense for ED to lessen the priority of stability with stable release because the vast majority of the community show that they are more keen for new features by their actions, than they are for stability. The communities actions speak louder than their words, and ED is just responding to the majority of the communities wishes. Which leaves me just having accept it, and either work with it, or find something else to do. 

[Goose489]

I want to play DCS but I don't have a PC good enough to play it. Do you know of a good PC that is relatively cheap but won't lag?

If so can you send me the Pc Partpicker link?

[Beirut]

9 hours ago, TheFreshPrince said:

If you want good graphics in DCS, you need to go VR and also only play the newest modules and maps.

 

I don't think there is any truth in that at all.

 

If you have your settings right, you can have great looking flights in older modules on any of the maps. Play with the lighting and the weather and, for example, you can have great eye candy on the Caucasus map with the F-5. 

[norman99]

5 hours ago, Goose489 said:

I want to play DCS but I don't have a PC good enough to play it. Do you know of a good PC that is relatively cheap but won't lag?

If so can you send me the Pc Partpicker link?

I find it hard to believe you can't find anything suitable. 

For context, I play on a 4 year old laptop (i7-8750H @4.0Ghz & RTX 2070 max-q), with a 2K monitor and 3 MFDs exported. All settings on High, DLSS, and rarely ever see lower than 40fps. It's usually much higher.

PC parts from the same vintage, say an i7 9700k and RTX2080 would be even better, and being ~4 years old, are probably quite cheap now.

Obviously if you're thinking of going VR, that's a different kettle of fish.

Edited December 6, 2023 by norman99

[Mr_sukebe]

A few thoughts:

- VR will always require more horsepower than a monitor, simply because it’s handling more pixels

- Remember when that new Civ sim arrived in 2020 and everyone went nuts about its graphics?  That’s simply competition.  ED cannot stand still with an aging platform and just hope for the best.  Want an example, look at Falcon 4.  That even has VR support now, but it still looks pap.  Do you really think that it sells in big numbers, and that has an unbelievable level of community support.

- my belief is that DCS looks better than that Civ Sim, certainly on my rig, which is very similar to yours.  Whenever I go near a city in that Civ Sim, it runs terribly and the buildings look like melted marshmallows

- as most of the community (based upon the last poll on here) still use a monitor, and it’s easier to run, ED are clearly pushing what can be done on a monitor, with top end gear, which is as it should be.  If they didn’t, you’d have users asking that they did.  Translating those same settings into VR is never going to work, because of the additional grunt required

 

In short, you need to learn to compromise on your settings and be thankful for what we currently have.

I was out yesterday in the Gulf on a JDAM strike in my F15E in VR.  Sun setting in the west, low level clouds, looked utterly amazing.

 

 

 

[Bounti30]

I read all the ideas, I see common sense in many of them and I agree on many points.
The expectations of some are not the same for others. But my opinion is that by adding layers of functionalities it will no longer be possible to transform the heart of DCS,
and that developers will be facing a wall. I hope I'm wrong because DCS is the only simulation I use and I've tried many (I'm 51). It makes me think of a house where we add floors without consolidating the foundations, and one day everything collapses.

[Mr_sukebe]

???

That doesn’t really make sense.  The move to multi core will have been a massive undertaking and stands DCS in a much better place long term

It’s worth remembering that the Civ Sim 2024 is primarily to allow introduction of multi core, for which they give you the pleasure of paying for it.  

[Hiob]

3 hours ago, Bounti30 said:

I read all the ideas, I see common sense in many of them and I agree on many points.
The expectations of some are not the same for others. But my opinion is that by adding layers of functionalities it will no longer be possible to transform the heart of DCS,
and that developers will be facing a wall. I hope I'm wrong because DCS is the only simulation I use and I've tried many (I'm 51). It makes me think of a house where we add floors without consolidating the foundations, and one day everything collapses.

ED is continuously improving and working on the game core. What makes you think otherwise? Problem with such work is, that you only ever will see most of it once it's finished.

[Bounti30]

Yes, multithreaing was a step forward for DCS, but not successful, many problems persist. you just have to read all the topics. Solutions are often found by the community, which I thank. But I repeat it, without first finalizing the core of DCS  functionalities such as moving grass, road traffic, clouds etc..., for many of them will be of no use if we cannot benefit from it.
I would like to be able to put all the settings on full and go “Wow”. but the more the versions advance, the more the settings decrease. And the only way to overcome this fact for many is to invest in high-end equipment.
I don't think I'm alone in thinking this.

[BIGNEWY]

3 minutes ago, Bounti30 said:

Yes, multithreaing was a step forward for DCS, but not successful, many problems persist. you just have to read all the topics. Solutions are often found by the community, which I thank. But I repeat it, without first finalizing the core of DCS  functionalities such as moving grass, road traffic, clouds etc..., for many of them will be of no use if we cannot benefit from it.
I would like to be able to put all the settings on full and go “Wow”. but the more the versions advance, the more the settings decrease. And the only way to overcome this fact for many is to invest in high-end equipment.
I don't think I'm alone in thinking this.

Hi, 

I dont think you have a full enough picture to say if it is successful or not, yes we have people having issues and we want them to post here on the forum with as much data as possible so we can continue tweaking, but we also have thousands of users having no issues at all. Multithreading from our point of view has been successful, we have measured results and data from performance increases.  

Your system performance depends a lot on your hardware and settings, like any of us, often there are bottlenecks to contend with and compromises to make to reach the desired frame rate / performance. DCS will always push the performance of any machine, as performance improves new features can be added and the tug of war between visuals and performance continue. Once we have multithreading tweaked some more we can looked towards Vulkan implementation, and again we will be looking for feedback and data, we will all be changing our settings and tweaking to find the best results. 

Best thing to do is post a thread with track replay examples and your dcs log and dxdiag so we can take a look. 

thanks

[Slippa]

 

“post a track”  Sorry, that’s done me after reading all that. The irony. No offense meant, really.

i agree with the point being made though. Get everything working properly before moving on. You can’t move on without moving on either so swings and roundabouts. Not an easy one.

[Bounti30]

Thank you for your response and your interest in this subject. I understand your overall view of the system and I refrain from giving you advice.
But you agree that DCS, and I repeat for myself is the best simulation, pushes the capabilities of computers. Are you not afraid that a certain part of the community will not be able to keep up without a real improvement in performance?
Wouldn't that be detrimental to the future of DCS?
 

 

[Hiob]

5 minutes ago, Slippa said:

 

“post a track”  Sorry, that’s done me after reading all that. The irony. No offense meant, really.

 

See - some people suffer severe problems. I get that. But not all of them. So, since all use the same DCS, the differences in configuration, hardware and so on, must be identified. Tracks and logs are the most reliable information about that kind of stuff. And often times, problems on the user-end can be identified and sorted out.

Whilst it would be ignorant for me to say "I can play without problems, so there aren't any!", it is equally wrong to assume the whole thing is broken because any number of individuals face a problem.

[BIGNEWY]

9 minutes ago, Bounti30 said:

Thank you for your response and your interest in this subject. I understand your overall view of the system and I refrain from giving you advice.
But you agree that DCS, and I repeat for myself is the best simulation, pushes the capabilities of computers. Are you not afraid that a certain part of the community will not be able to keep up without a real improvement in performance?
Wouldn't that be detrimental to the future of DCS?
 

 

We have many settings in DCS that can be tweaked to help achieve the desired look or frame rate. For machines that do not keep up with technology advances there will always be the option of reducing settings, and we test on many machines from high end to low end ones, but there is no stopping technology advancing, Ive been flying DCS for well over a decade, had many PC's in that time and I am trying to stick to a new machine every 5 years. 

19 minutes ago, Slippa said:

 

“post a track”  Sorry, that’s done me after reading all that. The irony. No offense meant, really.

i agree with the point being made though. Get everything working properly before moving on. You can’t move on without moving on either so swings and roundabouts. Not an easy one.

If a particular scenario creates problems on your machine posting a track replay is the only way we are able to recreate the problem. It is essential when reporting issues here. 

thanks

[norman99]

@Bounti30 Think of it this way.

If the core DCS engine was a paid product, released every few years (which I actually advocate, but that's another can of worms...), than with every new 'release', new features and updated technologies would be not just included, but expected. Instead, DCS has a very frequent update cycle, with small continuous improvements the ultimate goal. Over the course of time, both methods ultimately lead to a requirement to provide increased computing power, if one wishes to continue enjoying DCS to the fullest (i.e. maxed out). This is absolutely the same as any other long running series of software. I certainly don't expect MSFS to run maxed out on the same system I ran FSX or early versions of Prepar3D.

At the end off the day it's the price of progress, and it's unavoidable.

Edited December 6, 2023 by norman99

[Eugel]

33 minutes ago, Bounti30 said:

But you agree that DCS, and I repeat for myself is the best simulation, pushes the capabilities of computers. Are you not afraid that a certain part of the community will not be able to keep up without a real improvement in performance?
Wouldn't that be detrimental to the future of DCS?
 

Especially flight simulation games have always pushed computer hardware to their limit. The other big civilian flight sim does the same.
I think it would be much more detrimental to the future of DCS if they stopped pushing further after version 2.5 or whatever.

You wrote earlier that you have an i9 and an RTX 3090. So I really don´t understand your problem. 
But it´s also weird that you raise this point now. Of course not everyone is willing or able to keep up with new hardware. But just now they added DLSS support, which can give a considerable performance boost.

[BIGNEWY]

15 minutes ago, Eugel said:

But just now they added DLSS support, which can give a considerable performance boost.

Yep this year alone we have added multithreading, and upscaling options to DCS, this has helped a lot for many people. 

Vulkan is coming in the future and we will continue to advance. 

[Beirut]

6 hours ago, Mr_sukebe said:

In short, you need to learn to compromise on your settings and be thankful for what we currently have.

I was out yesterday in the Gulf on a JDAM strike in my F15E in VR.  Sun setting in the west, low level clouds, looked utterly amazing.

 

Exactly that!  

[Goose489]

16 minutes ago, BIGNEWY said:

Yep this year alone we have added multithreading, and upscaling options to DCS, this has helped a lot for many people. 

Vulkan is coming in the future and we will continue to advance. 

Is that official that the Vulkan is coming?

[BIGNEWY]

2 minutes ago, Goose489 said:

Is that official that the Vulkan is coming?

Yes we announced it a while ago. Work on vulkan is in progress, when we are ready we will share more news. 

[SharpeXB]

On 11/24/2023 at 5:56 AM, Bounti30 said:

I am using a Quest 2 and I recognize that there are a multitude of factors that can influence performance.

This is a big part of your troubles. DCS is primarily designed to be run in 2D so VR will always struggle here. It’s necessary to lower your settings quite considerably and even with the most powerful machine money can buy you will still need to make compromises. If that situation bothers you, consider running the game in 2D. Performance problems will always plague VR when trying to run games that aren’t intended for it. 

On 11/25/2023 at 5:20 AM, Dragon1-1 said:

VR is rather inefficient in DCS right now. Once ED finally moves to Vulkan, it'll be better, because DX11 is really stupid about how it does VR, it effectively renders the entire scene twice.

I don’t see how Vulkan can avoid stereoscopic rendering since this is inherent in how VR works and a primary reason why it’s so demanding. The benefit of Vulkan as I understand it here is that it offloads some tasks from the CPU to the GPU, what that means for DCS remains to be seen. But the performance struggle in PC gaming will continue regardless. 

[Dragon1-1]

39 minutes ago, SharpeXB said:

I don’t see how Vulkan can avoid stereoscopic rendering since this is inherent in how VR works and a primary reason why it’s so demanding. 

It doesn't avoid stereoscopic rendering. What it does is being smarter about how it's accomplished. DX11 renders two completely separate scenes, one of right eye, one for left. Vulkan, meanwhile, renders a single scene and then shows it from two different angles. The latter is vastly more efficient (because the hard part is only done once) and will mean that VR performance will come closer to flatscreen performance at the same resolution. Which will, of course, still require a beefy rig, since rendering a scene even once in 4K is quite a task, but it'll be more doable.

[SharpeXB]

40 minutes ago, Dragon1-1 said:

What it does is being smarter about how it's accomplished. DX11 renders two completely separate scenes, one of right eye, one for left. Vulkan, meanwhile, renders a single scene and then shows it from two different angles.

That still seems like the same thing. I guess we will wait and see. Simple math says it will always be more demanding to draw things twice. 

41 minutes ago, Dragon1-1 said:

Which will, of course, still require a beefy rig, since rendering a scene even once in 4K is quite a task, but it'll be more doable.

That’s the second part of the trouble. VR would require resolution in excess of 4K in order to be sharp enough to replicate 20/20 vision. 2D can accomplish that with zoom view. 

[Dragon1-1]

3 hours ago, SharpeXB said:

That still seems like the same thing. I guess we will wait and see. Simple math says it will always be more demanding to draw things twice. 

This is exactly why Vulkan is going to be better: it doesn't draw things twice. It draws them once, then shows it from two angles. It's not exactly the same as drawing a single 4K picture, but it's sure closer than what DX11 does.

It's not the same thing because things like lighting, culling and so on are calculated only once, because the GPU only renders a single scene. DX11, for most part, does everything twice, even if things like position of the light sources is exactly the same for both eyes, and the viewpoints are only a few centimeters apart.

3 hours ago, SharpeXB said:

That’s the second part of the trouble. VR would require resolution in excess of 4K in order to be sharp enough to replicate 20/20 vision. 2D can accomplish that with zoom view. 

For the Aero/Reverb G2 market segment, 4K will be a target for a long time to come. This is where monitors stand, too, so I believe it's a proper place for these things to be. Eventually, tech will move forward and resolution will increase. When GPUs start advertising 8K or 12K gaming, with appropriate wall-sized monitors (we'll get there, just give it some time), mid-level VR will be able to go there, too. Headsets replicating 20/20 vision and full human FOV won't be consumer hardware for quite a while.

Edited December 6, 2023 by Dragon1-1