43 replies to this topic

[mytilus edulis]

Over the last two years, both Navid and MrVaad independently investigated ways to improve game performance with one fairly simple change: swap the missile and missile flame animated objects, for static ones, resulting in a ~100% increase in performance (from ~25 to 60 FPS in stress tests with 800 missiles in the air).

Currently, every missile fired consists of two objects: an animated missile object, and an animated flame trail object. However, despite being animated objects, neither of these objects have any actual animations, only empty keyframes! So while the missile is in the air, the engine just keeps redrawing the same geometry over and over and over again, with zero changes. And it does this twice for each missile because of the two object composition.

As far as I know, both of them have independently proposed to PGI an easy fix for this that results in zero visual differences (export both objects as static ones instead of animated ones), but that was back in summer.

Is there any chance of this optimization going through?

Would anyone else here support this optimization?

This could be the next best thing since lowered volume in the splash screen videos.

Edited by mytilus edulis, 11 November 2022 - 09:35 PM.

[martian]

Who would be against the better game performance? I guess that nobody would protest.

However, does PGI have the sufficient resources to actually implement this change? We know that the situation is not good.

Edited by martian, 11 November 2022 - 09:47 PM.

[D V Devnull]

Anything that can help performance for older systems and keep more people running MWO comfortably gets my definite thumbs-up!

As a side-effect, that might help change some of the opinion regarding Missiles (particularly the Lock-On kind) where people want them gone because their computer was bogging down too much.

That reminds me... Can we get a fix for the hyper-blinding explosions from Ballistic Weapons for everyone? Being unable to shoot back at opponents and getting explicitly forced to run without exception because you're taking an AutoCannon pummeling simply is unfair & unfun in absolutely every way on the receiving end... It needs to stop.

~D. V. "Visual performance & functionality improvements? All for it!" Devnull

[mytilus edulis]

martian, on 11 November 2022 - 09:47 PM, said:

Who would be against the better game performance? I guess that nobody would protest.

However, does PGI have the sufficient resources to actually implement this change? We know that the situation is not good.

If by sufficient resources you mean "can they copy and paste files that already exist and have already been offered to them"...well, one can hope

Edited by mytilus edulis, 12 November 2022 - 01:57 AM.

[martian]

mytilus edulis, on 12 November 2022 - 01:56 AM, said:

If by sufficient resources you mean "can they copy and paste files that already exist and have already been offered to them"...well, one can hope

Well, let us hope that either Daeron Katz or Matthew Newman can do that.

D V Devnull, on 12 November 2022 - 12:50 AM, said:

Anything that can help performance for older systems and keep more people running MWO comfortably gets my definite thumbs-up!

As a side-effect, that might help change some of the opinion regarding Missiles (particularly the Lock-On kind) where people want them gone because their computer was bogging down too much.

On the other hand, some people should understand that using the same computer, that was possibly mediocre or even sub-standard a decade ago, could be a problem in 2022 (almost 2023).

Ditto as for the Operating system.

Ditto as for the Internet connection.

D V Devnull, on 12 November 2022 - 12:50 AM, said:

That reminds me... Can we get a fix for the hyper-blinding explosions from Ballistic Weapons for everyone? Being unable to shoot back at opponents and getting explicitly forced to run without exception because you're taking an AutoCannon pummeling simply is unfair & unfun in absolutely every way on the receiving end... It needs to stop.

~D. V. "Visual performance & functionality improvements? All for it!" Devnull

With the proper positioning, it is not always necessary to run all the time.

Also, for some weapons such as Rotary Autocannons, that blinding effect is one of their few advantages.

[RickySpanish]

Mind bending that whatever internal testing was performed for the game never picked this up. It should absolutely be implemented.

[Darian DelFord]

RickySpanish, on 12 November 2022 - 08:36 AM, said:

Mind bending that whatever internal testing was performed for the game never picked this up. It should absolutely be implemented.

All IGP cared about then was releasing as much cash grabs as they could with little to no testing....... Clans anyone?

[RickySpanish]

I 'member clams!

[w0qj]

1. MrVaad's work is very well known on MWO Forums, as below:
http://mwomercs.com/...-editor-for-mwo
http://github.com/MrVaad/MWOConfig

2. Question: Where did Navid share is MWO game performance optimisation work? Just curious.

mytilus edulis, on 11 November 2022 - 09:34 PM, said:

Over the last two years, both Navid and MrVaad independently investigated ways to improve game performance with one fairly simple change: swap the missile and missile flame animated objects, for static ones, resulting in a ~100% increase in performance (from ~25 to 60 FPS in stress tests with 800 missiles in the air).

Currently, every missile fired consists of two objects: an animated missile object, and an animated flame trail object. However, despite being animated objects, neither of these objects have any actual animations, only empty keyframes! So while the missile is in the air, the engine just keeps redrawing the same geometry over and over and over again, with zero changes. And it does this twice for each missile because of the two object composition.

As far as I know, both of them have independently proposed to PGI an easy fix for this that results in zero visual differences (export both objects as static ones instead of animated ones), but that was back in summer.

Is there any chance of this optimization going through?

Would anyone else here support this optimization?

This could be the next best thing since lowered volume in the splash screen videos.

[mytilus edulis]

w0qj, on 12 November 2022 - 11:36 AM, said:

1. MrVaad's work is very well known on MWO Forums, as below:
http://mwomercs.com/...-editor-for-mwo
http://github.com/MrVaad/MWOConfig

2. Question: Where did Navid share is MWO game performance optimisation work? Just curious.

1. this is something that can't be changed by individual players, it has to be patched in as new files

2. Cauldron Discord

[MrVaad]

I can give a little history on that missile fix

For each missile (LRM, SRM, ATM, NARC etc):
- The missile has 300 triangles and one texture.
- The 3D trail has 768 triangles and four textures. The geometry/form of the trail change between each frame and there's an animation on it using the four textures.
Yes, that trail is animated in game.

Two problems
1- separated objects with separated textures use many drawcalls (cpu->gpu instructions).
2- too many polygons when hundreds of missiles are flying.

Problem 2 can be overcome with a good PC. Problem 1 can be a problem even on a good PC (we get into dx11 limits).

The fixes
I did a first fix in the cauldron around 04/2021.
- I disabled the missile 3d objects,
- Extended the missile engine particle effect to start right after firing a missile
In the end, this replace the 3d objects with a particle effect.
Perf was very good. The only problem was not getting missile bits when firing under min range.

Second fix was done by Nav1d remodelling the missile 3d object. He used an old 3Dstudio max and i sent him the cryengine export tool.
This uses a little more ressources but is more compatible with the existing game (you get the missile bits when firing under min range).

And we did some performance test for both fixes (with an old reshade version that monitor drawcalls)

And we waiiiit
First fix is editing two xmls. Second fix is replacing the missile 3d objects with the new one.
I think PGI wants one of their developper's analysis/approval but no one seems available.

Edited by MrVaad, 13 November 2022 - 05:55 AM.

[martian]

Lockheed_, on 13 November 2022 - 10:41 AM, said:

If we wait for PGI to do that it will take probably forever. It took them over 2 months to fix the pink bug, something that should have been a hot fix as soon as you found the solution. Can't we just distribute those 3d objects among ourselves?
If they don't have any resources fine, but at least they should support the community developing fixes. I also see no issue this late in the games life with uploading a fix just in form of the files and a note that says "Warning, this fix is experimental, use at your own risk and only if you know what your are doing"

Not a bad idea, but I believe that the MWO rules do not allow the use of modified game files (with the exception of a small number of files that are explicitly allowed by PGI).

[Meep Meep]

In the meantime that custom user config I posted does seem to stabilize frames for some users though it's not a fix for the missile slowdown.

[PocketYoda]

Not a single dev or mod comment lol.. I'd be up for any changes that fix the FPS.

[LordNothing]

after all these years they still haven't made full window the default. and that can result in a large number of support tickets and ruined games, especially among new players. if they cant do that right, performance improvements are out of the question.

Lockheed_, on 13 November 2022 - 03:05 PM, said:

1. I expect PGI to upload that fix with the warning. Edited the original post for clarity.
2. Maybe they need change their stance on allowing people to modify game files in circumstances like that. Their product is poorly optimized and people are offering solutions for their product, for free. They already have part of the community working on their game anyways.

the cauldron needs a performance optimization team that can make changes like these.

[confracto]

LordNothing, on 14 November 2022 - 06:16 PM, said:

the cauldron needs a performance optimization team that can make changes like these.

And yet, PGI still needs to actually verify these changes to maintain the integrity of the game. I felt like I was the last one there even capable of understanding if the changes the Cauldron were proposing would present risk. There was already an instance last year where a Cauldron change went in blind and didn't break stuff only because a failsafe was in place. We're all still relying on that failsafe, and an actual fix will be significant work.

Honestly, I hope this missile proposal would help, but I can already see several potential complications, and all of them require a level of capability and support that the Caudron and what I understand is left of the current team aren't able to deliver. To do something, and risk complications ballooning and squandering what little dev time you can invest for a small performance increase, or do nothing and use that time elsewhere? I'd have strongly advocated against this missile fix.

[Mokkor]

confracto, on 14 November 2022 - 11:32 PM, said:

And yet, PGI still needs to actually verify these changes to maintain the integrity of the game. I felt like I was the last one there even capable of understanding if the changes the Cauldron were proposing would present risk. There was already an instance last year where a Cauldron change went in blind and didn't break stuff only because a failsafe was in place. We're all still relying on that failsafe, and an actual fix will be significant work.

Honestly, I hope this missile proposal would help, but I can already see several potential complications, and all of them require a level of capability and support that the Caudron and what I understand is left of the current team aren't able to deliver. To do something, and risk complications ballooning and squandering what little dev time you can invest for a small performance increase, or do nothing and use that time elsewhere? I'd have strongly advocated against this missile fix.

If you don't mind, what are the possible risks of their fixes, that you can see already, without digging deep into the game?

[Meep Meep]

To chime in I have yet to break any part of the game by editing the cvars in the user script. Maybe have some graphics glitches for pushing a range too small or large but the cvars themselves are harmless.

[Meep Meep]

Ok this should show in easy to understand visuals just how much the look of the game can be improved by some simple user side ini edits.

(open each in a new tab to flip between them)

Settings are low for shading and shadows and very high for the rest with post aa enabled.

Vanilla Alpine.



Modified ini Alpine.



Rather striking yes? Now imagine what they could do from the inside using cryedit and someone familiar with the engine and not some old fart like me who dabbles in this stuff.

[-OUTLAW-]

If you have an NVIDIA card, disable AA in-game and enable AA through you drivers (NVIDIA Control Panel game profile). This worked for me, looks better and no more rubber banding.