Best news for 2020: Focus on CORE of simulation

[mdee]

Hope in one hand...

 

The big VR upgrade was a nothingburger, im not expectimg miracles feom vulkan.

 

Anywhere 3-10% fps gain, depending on how well they implement it. In some extreme cases, performance can be worse. People think Vulcan is some magic thing, it isn't. The best way to "optimize" DCS is to have cutting edge hardware.

[Harlikwin]

Anywhere 3-10% fps gain, depending on how well they implement it. In some extreme cases, performance can be worse. People think Vulcan is some magic thing, it isn't. The best way to "optimize" DCS is to have cutting edge hardware.

 

Yeah, already there on the hardware side. I mean it runs, in VR, at a stable 45fps prior to 2.5.6 at least. Other games run at 90 without a problem...

[rrohde]

Vulkan seems to utilize modern hardware better than DX11 does - isn't that the premise?

[tuxy42]

I swore myself not to buy any new module until core is updated to descently work in VR. I invested in a i7 9700k 32Go and 2080s and valve index, virpil stick, a lot of modules to look at... a diaporama! No offense I love DCS but I really think it's time to forget 90's technologies for the core, I do understand the complexity and the challenge to migrate on a new graphical engine or at least get a significant improvement but I do believe it's an absolute necessity to shine in future. 2.5.6 doesn't go in right direction in my point of view.. we struggled to get a descend fps framerate in 2.5.5, by using a lot of tweaks and custom shaders made by third parties which are more concerned of VR performance than ED and finally, last update put us back on starting point for few lighting improvements, nice but not with the price of 30% fps. I really hope ED will really take care of VR soon, Then I would be happy to buy again :)

[Whiskey11]

Vulkan seems to utilize modern hardware better than DX11 does - isn't that the premise?

 

Vulkan is about efficiency in rendering. One of the biggest advantages is offloading much of the graphics processing to multiple cores in the CPU and reducing overall CPU workload and overhead. On a game like DCS World, where the game is primarily CPU limited in non-VR and a combination of CPU and GPU limited in VR, offloading computation onto multiple cores should make a much larger difference than games that were GPU bound before.

[Al-Azraq]

Anywhere 3-10% fps gain, depending on how well they implement it. In some extreme cases, performance can be worse. People think Vulcan is some magic thing, it isn't. The best way to "optimize" DCS is to have cutting edge hardware.

 

It depends, if ED manage to do some serious multi-threading on the way we will see huge FPS boosts but if it is just a plain Vulkan implementation the performance improvement will be in the range you mention. Anyway, ED stated many times that the main point of implementing Vulkan is multi-threading so I'm hopeful.

[Fri13]

Anywhere 3-10% fps gain, depending on how well they implement it. In some extreme cases, performance can be worse. People think Vulcan is some magic thing, it isn't. The best way to "optimize" DCS is to have cutting edge hardware.

 

If you take a game that was designed for a 1990 hardware, you need to rewrite it to if you want to take benefits of new hardware 30 years later.

 

The big difference is, DCS World is modern, and it is very complex software. So it is not optimally working when you can't by the architecture use all the hardware.

 

DCS has something that many other games don't have, that is simulation.

Now if you add 300 ground units on mission, your PC will almost die there.

Because your single core is doing everything.

 

Now if you split that well, blufor ground units to core #1, Redfor ground units for core #2, your aircraft to core #0 and everything else to core #4.

 

Now cut the timings. If the #2 goes crazy delayed, it shouldn't affect anything else. They become turkey shoot, but it shouldn't affect your aircraft or bluefor units at all.

 

Sounds unfair right? Well, better than that everyone is waiting that Redfor gets over that bridge in orderly fashion.

 

Now if you launch missile and you suddenly enable mirrors, missile flight path shouldn't get affected as it is on #4 core while you are #0.

 

But now there is a big problem that how do you sync such missile guidance and your aircraft radar guiding it? Well, that is the huge challenge...

 

It is same challenge as what to do when one ground unit blocks the bridge from others. Or one gets stuck on the tree in a group formation.

 

You need to cheat, find a solution to get around such hard problems with soft clever solution that makes all work.

 

Example:

[nervousenergy]

That's a stretch, it's quite a bit more complicated than this. We gone over this a lot in the past threads, and DCS is as most applications are, already multi threaded, the question is how good and how good it can be, how good it can be is unfortunately lower than most games because it's a simulator with a lot of serial type of workload that cannot be parallelized, so we need to forget about the level of multi-threading other modern physics-less games can be.

 

That said improvements can still be done in this regard to get everything away and the main simulation thread, the serial workloads that have to be part of the same thread, to have the best an most optimal conditions and have the space all for it self, while other stuff is put on their own threads.

 

I'd only add this that improvements not only can be done, but MUST be done, or the sim will start to collapse under it's own weight. IPC and clock speed improvements have nearly ground to a halt as engineering meets physics head on. The writing has been on the wall for some time - improvements in compute performance are coming wider with more cores, not taller with more IPC and clock per core.

 

Even if the core physics simulation has to remain on one thread (and I think even some of that can be made parallel), there's a TON of compute that can be spun off. AI should be low-hanging fruit... people have been writing self-contained client bots since the Quake days. Avionics systems, weather, damage, etc. are all candidates.

 

As long as the sim is adding more performance draw from a single core each year than cores are going up in IPC/clock (and I think that's now 10% or less per year), then things will keep getting worse. Got to go wider. It's much harder to code this way and takes a lot more resources/talent, and I hope DCS can commit that. Battlefield 5 gets great scaling across 6 logical threads, but they've probably got more engine programmers than DCS has total employees.

[DLEGION]

honestly, without changing any setting, with 2.5.6 i have on gudauta airfield around 30 FPS, while i had in 60-70 range in 2.5.5 .

in the air its less noticeable, but on the ground its a pain, also for low level flight.

i care more on gameplay than graphics, so for me its a serious downgrade.

[Harlikwin]

honestly, without changing any setting, with 2.5.6 i have on gudauta airfield around 30 FPS, while i had in 60-70 range in 2.5.5 .

in the air its less noticeable, but on the ground its a pain, also for low level flight.

i care more on gameplay than graphics, so for me its a serious downgrade.

 

Yeah, I got around 12 fps around sochi IIRC in the air coming in for a landing online last night. Pretty bad, but it bumped up to 30's when I landed. Was running solid 45+ on 2.5.5....

[Targanon]

they have been workingon the core already that why we have the current problems with 2.5.6.

[Lenux]

Should be expect to have the new clouds / weather / until May, June?

What do you think, guys ?

[Svsmokey]

Should be expect to have the new clouds / weather / until May, June?

What do you think, guys ?

 

Prolly the best answer we can get....

 

 

[Harlikwin]

Prolly the best answer we can get....

 

 

 

That they have failed 2x already isnt too encouraging. But at least they are honest about it, so thats a plus.

[Lenux]

Thanks for sharing the link

[Snappy]

Should be expect to have the new clouds / weather / until May, June?

What do you think, guys ?

 

Much as I can understand how you re looking forward to these features, with the current massive performance issues of the 2.5.6 update I really don‘t want to know how the sim runs with even more strain added by these features..

Of course its possible ED can fix the performance issues and I certainly hope so.

 

 

But the trend with the last updates has in my personal experience mostly been negative in regards to performance and later updates seldom recouped earlier performance losses for me.

 

 

Don‘t have the answer to that.

 

 

Regards,

 

 

Snappy

[Fang333333]

Yay!

[Mouse_99]

This is great news, I am anticipating the ED crew will be making significant progress on all these things that are real issues for so many. I think they have more money coming in from all these newbies joining in so, this allows them to allocate more resources to the "core", so it's happy days.

[USA_Recon]

This was the best news, from my angle, in the Upcoming 2020 development statement:

 

"Most of the Eagle Dynamics team will be focused on improvements to the DCS World core that includes performance improvements, more realistic lighting, a new cloud and weather system, an outstanding damage modeling, a new airfield air traffic control system, air-to-ground radar and Forward Looking Infrared (FLIR) sensors, improved missile dynamics, more life-like AI, new AI units, and a dynamic campaign system."

 

My emphases added.

 

Good work, ED! :thumbup:

 

1. performance improvements. Well, last patch, killed fps in VR. So that went backwards

2. lighting - is terrible. Bloom effects overdone. Flew F-15 in NTTR last night, radar and rwr lines where thicker, it didn't look very good at all. I think this went backwards. Any chance we can have this as a setting ?

3. missile dynamics - each release the aim-7 has different characterics (we test this, chaff resistence, etc...). Dsync continues to be horrible

 

ED, obviously move forward, but don't move backwards. Don't say perf improvements then throw in a bunch of fps killers in new patches.

[strikers_blade]

1. performance improvements. Well, last patch, killed fps in VR. So that went backwards

2. lighting - is terrible. Bloom effects overdone. Flew F-15 in NTTR last night, radar and rwr lines where thicker, it didn't look very good at all. I think this went backwards. Any chance we can have this as a setting ?

3. missile dynamics - each release the aim-7 has different characterics (we test this, chaff resistence, etc...). Dsync continues to be horrible

 

ED, obviously move forward, but don't move backwards. Don't say perf improvements then throw in a bunch of fps killers in new patches.

 

3 comments and 3 complaints. What about posting into the "bugs and problems" instead?

 

And yes, ED said they will work on the core so it would be futile to think it will always be "good" on the first trial. Mistakes will happen, which is why BETA exists....

[BlacleyCole]

IMO ED needs to complete the mods it has started, work on the core items mentioned above before releasing another mod maybe two to three mods I work at one time. How many do we have now? 5 or six ? Plus fix the huey before redoing the ka-50.

[doclucio]

hi guys, and think about the supercarrier module...i just purchased it (2080ti and valve index), i wanted to help the team...i think that taking off from the new carrier will be...5fps?!?

anyway i love the game but agree they should prioritize performance now

[DerFangzahn]

The whole 2.5.6 debacle is best thing what could happen. In the past ED was able to hide behind all the shiny fancy modules without putting any substential effort into the simulation itself.

 

This model is falling apart. 2020 will be a turning point for DCS. For good or bad we will see...

[Cornelius]

The whole 2.5.6 debacle is best thing what could happen. In the past ED was able to hide behind all the shiny fancy modules without putting any substential effort into the simulation itself.

 

This model is falling apart. 2020 will be a turning point for DCS. For good or bad we will see...

 

Please do not underestimate the forgetfulness of DCS customers, especially when new boombastmusic trailers spoil upcoming modules.

Then the "shut up and take my money" hymn will be sung in the choir again.:pilotfly:

[GunSlingerAUS]

It's so cute you kids think this is all going to be possible on ED's non-existent financial model. "See children, if you wish for something long enough, dreams do come true!".

 

Personally, this claim got boring for me some time back In 2014.