Project Reality Forums

Well you can always come up with all kinds of explanations and i am probably wrong about some of the specifics since i am not well versed in coding. The stuff on my meme was just what i pieced together from talking to people in the PR community that actually know more about this and from reading posts on some certain shady forums.

What i can say with full confidence however is that performance has been steadily decreasing with every update, safe for the launcher outsourcing fix mats did once, which temporarily lifted off some weight.
So instead of trying to focus on defeating my main argument by poking holes in my meme pic which admittedly is probably wrong on some things, how about you explain the general tanking of FPS in PR on urban maps.

This effect is most noticeable on urban maps with lots of objects in the line of sight and at its worst when combined with lots of players being in those areas. Worst offenders by far are Grozny (the city, the outskirts perform fine), fallujah, shahada and the like. Decreases also happened on all other maps but in general are only noticeable there for players with old and weak hardware, as other maps will still generally run at 100 fps for people with modern hardware.

I talked to 3 different other people (not gonna name them don't wanna involve them in autism) that basically have the best consumer hardware available or close to it (e.g. 9700k, fast RAM, 2080ti you name it). While they can run fallujah at several 100 fps when the fps cap is unlocked with console command, the moment a sizeable amount of player characters are introduced in their LOS FPS will stop tanking and will reach their lowest average at approx. 30 FPS on a full server. Players with dated hardware may even find themselves getting low FPS on fallujah on an empty map though.

What is interesting to note though, is that it seems that beyond a certain point better hardware (notably CPU single thread performance, HDD/SSD read/write speeds and RAM quality and speed) will not result in a noticeable increase in average minimum FPS in these kinds of situations. For example i have a ryzen 3600x (rated as one of the top ten in single thread performance on passmark: https://www.cpubenchmark.net/singleThread.html), a fast nvme SSD and 16gb 3200 Mhz RAM. But when i compared the min FPS live on grozny with someone with worse CPU and RAM, standing next to each other looking at the same spot, we found that our min FPS were basically identical within the margin of error.

This seems to indicate that beyond a certain point hardware wise the game engine itself becomes the hard limit/cap and better hardware will then not yield any better min FPS under the same circumstances.

Mineral mentioned that modern CPUs are often lacking in single thread performance compared to older ones, but i generally find this to be not true. The most recent top of the line releases dominate the single thread benchmarks. This only really was kind of true for ryzen 2nd gen and the FX line of processors, afaik intel always pushed more IPC every generation. In any case i made sure to account for RAM quality and CPU single thread performance when asking other players about their rigs.

Mineral also mentioned DX9 might play a part here and that DX9 games will perform worse on newer OS and hardware. On the other hand someone knowledgeable i talked to expressed the opposite and said optimization for DX9 on windows has never been better? So would appreciate someone shedding more light on this.

I will dig up some older threads and post them here which will show people complaining about performance decreases with new updates going back as far as the 1.0 release. I still remember reading such a thread that was created right after 1.0 with lots of people reporting sudden and big FPS decreases. I also still remember trying out PR once many years ago before 1.0 for a few days on my shitty old laptop and not noting any FPS issues then.

For now this thread (which i already posted here a while ago) will do nicely though:
https://www.realitymod.com/forum/showthread.php?t=55054

He reported 100 fps on fallujah with specs that are even worse than my old PC was. The argument that this can be accounted for by the increased player limit from 64 to 100 players after 1.0 is moot, since i couldn't even get more than 45-25 fps on fallujah on an empty server in v1.2-v1.5 with my old PC, which again, had even better specs than his (same CPU, better and more RAM, better GPU).

Writing 45-25 since this was the respective average FPS on a speficic part of the map for the two specified iterations of the game on an empty fallujah. I still remember this vividly since i always used fallujah as a benchmark basically to stress test.

So to reiterate, i can clearly prove a significant performance decrease that happened over the years. You can even go to the first page of this very thread you are reading right now and see many people saying the same thing. Claiming the maps always performed the same is utterly wrong and ignorant!

Threads:

Performance difference between 1.4.1.0 vs 1.5.2.0:
https://www.realitymod.com/forum/showthread.php?t=147179
Interesting bit from this thread:


Amazing. So server side python code could potentially affect client performance after all? Wow, i got double bamboozled, don't know what to believe anymore tbh.

From 2014 complaining about lower FPS since 1.0 release: https://www.realitymod.com/forum/showthread.php?t=130887
Seems people with lower end systems were the first to notice since they had less 'buffer'. Same thing i experienced with my previous potato PC, i would always notice after certain updates when performance got worse.

https://www.realitymod.com/forum/showthread.php?t=122486

FPS drop after 1.0 (57 pages lol): https://www.realitymod.com/forum/showthread.php?t=121940



Will update this post when i find more.

DogACTUAL wrote:Performance difference between 1.4.1.0 vs 1.5.2.0:

I did performance testing for PR Team after 1.5 release. Here are the results:

DogACTUAL wrote:Amazing. So server side python code could potentially affect client performance after all? Wow, i got double bamboozled, don't know what to believe anymore tbh.

He said 'playing locally'.
I know not everyone is greatly educated in english and computer science - and I am not either - but I guess that by 'playing locally' he meant playing on a local server.
You know, when you are playing on your local server, all the python stuff is running on your PC, therefore it might affect your FPS...

i have a 1060 6gb, 16 gb ram and a prosssesor on i5-9something but i stil have a fps drop when i look at big areas. is that normal?

selman wrote:i have a 1060 6gb, 16 gb ram and a prosssesor on i5-9something but i stil have a fps drop when i look at big areas. is that normal?

Depends, I have slightly better and things like the UG on beruit causes me me fps drops.

I'm not exactly sure what you want Dog. Nobody has ever denied that pr's performance is worse then vBF2. Or that over the years we have added more complex maps and more stuff (content) on screen causing worse performance. But there are so many factors to this. Some in our control and some simply aren't.

If you want some 'break/pause' of general development and want for us to go in some optimization sprint I can tell you that this simply won't happen. Or to stop making complex maps. It's not going to happen.

Occasionally we look into performance as a team or individual team members. Mostly we just stick to our workflow guidelines. But we are still a bunch of volunteers locked in an old outdated game.

This just reads as you complaining you want good fps on all maps. Fair enough. We would like the same. But so far we haven't found any real performance benefit to the many things we tried over the years with basic experiments. Maybe you will have more luck? Let us know if you do. Until then we just gonna continue with what we do. Feel free to download an old PR version if you think it will perform better. But this getting extremely boring. So I suggest you move on.

He said 'playing locally'.
I know not everyone is greatly educated in english and computer science - and I am not either - but I guess that by 'playing locally' he meant playing on a local server.
You know, when you are playing on your local server, all the python stuff is running on your PC, therefore it might affect your FPS...

Yeah reading it now it is obvious you are correct. I was amped up on five big cups of coffee and checked lots of threads quickly, i saw his post and my preconceived bias got the better of me, also didn't help that i missed the context of the guy testing in local.

While i was kinda rubbing it in i was also genuinely confused since what alon said about python made sense and i kind of scratched my head too and asked myself how stuff running on the server would affect performance on the client system. But went with it because it seemed to me like that was what mats really meant there.

i have a 1060 6gb, 16 gb ram and a processor on i5-9 something but i still have a fps drop when i look at big areas. is that normal?

This is very generic and non descriptive info, you don't even know if your cpu is an i5, i7 or i9 let alone which generation and exact model number. RAM speed/frequency and timings can also noticeably bottleneck your CPU if they are too bad. GPU is more than enough though. I'd still say that your experience can be said to be normal and expected given how PR runs on new and a bit older hardware.

@Mineral: Well it is pretty much accepted that for shooters on PC stable 60 fps is pretty much the minimum requirement to have an environment where you can have a decent gameplay experience.

Adventure and RPG games and some console titles can also be played at 30 fps, but they usually utilize some tricks to smooth out the gameplay and most importantly, they will deliver very stable frame rates and frame times in general. PR on the other hand will not have stable frame times or frame rates at all in the situations described, especially when you need it most, while moving your aim(fov), which leads to a rather nasty effect.
Rapidly shifting FPS that quickly alternate from say 30 to 50 fps can look and play much worse than stable 30 fps.

Also i do not need to know the exact workings of the engine and your code constructs to know that moving PR into this territory performance wise was a choice you made, when you play tested the maps or updates and said 'this is fine'. Until one day it wasn't and too many people complained. You threw them a bone, did some optimization work that took you just a handful of days to finish (to your credit your probably worked hard on that). Something tells me you could do it again, but that is just a pure guess based on observation and nothing else.

The links speak for themselves, the same story repeats again and again after several updates, some people, especially ones with limited hardware, notice decreased FPS and they just get told to 'unpark' their cores or overclock/get a better CPU.
The big difference now is that it has reached a point where better hardware beyond a certain point doesn't yield further increases in performance anymore. Which maybe should be some cause of concern because if the trend continues you may find yourself in the same situation again. Also not helpful for player numbers alienating the ever increasing arab/asian/african playerbase, one of PRs biggest draws should be an ability to run on older hardware instead of the opposite.

Oh and btw even if i had the motivation to become an R-DEV, which i really do not, my time would be better spent learning something else, because the much thrown around claim 'whoever puts in the work becomes an R-DEV' has a slight asterix next to it that says it holds only true if you either tow the party line or you are a non controversial guy who stays out of any circlejerk and just delivers pure content that doesn't affect game balance or meta, like the dutch and russian devs.
Sorry tovarish but i've seen one too many times aspiring R-CONS get stonewalled when they got the wrong ideas concerning game balance. Oh and why board a sinking ship, your own core DEV team seems to be content with PR biting the dust sooner rather than later, it is a self fulfilling prophecy at this point.

I've also done my share of free of charge work in my life and i don't know about you, but once i start i usually try to get it right and don't leave it in a half finished state, it is a matter of principle. Either don't do it, nobody will blame you for it, but if you do it, do it properly, otherwise people will end up wishing you wouldn't have done it in the first place.

DogACTUAL wrote:Yeah reading it now it is obvious you are correct. I was amped up on five big cups of coffee and checked lots of threads quickly, i saw his post and my preconceived bias got the better of me, also didn't help that i missed the context of the guy testing in local.

While i was kinda rubbing it in i was also genuinely confused since what alon said about python made sense and i kind of scratched my head too and asked myself how stuff running on the server would affect performance on the client system. But went with it because it seemed to me like that was what mats really meant there.


This is very generic and non descriptive info, you don't even know if your cpu is an i5, i7 or i9 let alone which generation and exact model number. RAM speed/frequency and timings can also noticeably bottleneck your CPU if they are too bad. GPU is more than enough though. I'd still say that your experience can be said to be normal and expected given how PR runs on new and a bit older hardware.

@Mineral: Well it is pretty much accepted that for shooters on PC stable 60 fps is pretty much the minimum requirement to have an environment where you can have a decent gameplay experience.

Adventure and RPG games and some console titles can also be played at 30 fps, but they usually utilize some tricks to smooth out the gameplay and most importantly, they will deliver very stable frame rates and frame times in general. PR on the other hand will not have stable frame times or frame rates at all in the situations described, especially when you need it most, while moving your aim(fov), which leads to a rather nasty effect.
Rapidly shifting FPS that quickly alternate from say 30 to 50 fps can look and play much worse than stable 30 fps.

Also i do not need to know the exact workings of the engine and your code constructs to know that moving PR into this territory performance wise was a choice you made, when you play tested the maps or updates and said 'this is fine'. Until one day it wasn't and too many people complained. You threw them a bone, did some optimization work that took you just a handful of days to finish (to your credit your probably worked hard on that). Something tells me you could do it again, but that is just a pure guess based on observation and nothing else.

The links speak for themselves, the same story repeats again and again after several updates, some people, especially ones with limited hardware, notice decreased FPS and they just get told to 'unpark' their cores or overclock/get a better CPU.
The big difference now is that it has reached a point where better hardware beyond a certain point doesn't yield further increases in performance anymore. Which maybe should be some cause of concern because if the trend continues you may find yourself in the same situation again. Also not helpful for player numbers alienating the ever increasing arab/asian/african playerbase, one of PRs biggest draws should be an ability to run on older hardware instead of the opposite.

Oh and btw even if i had the motivation to become an R-DEV, which i really do not, my time would be better spent learning something else, because the much thrown around claim 'whoever puts in the work becomes an R-DEV' has a slight asterix next to it that says it holds only true if you either tow the party line or you are a non controversial guy who stays out of any circlejerk and just delivers pure content that doesn't affect game balance or meta, like the dutch and russian devs.
Sorry tovarish but i've seen one too many times aspiring R-CONS get stonewalled when they got the wrong ideas concerning game balance. Oh and why board a sinking ship, your own core DEV team seems to be content with PR biting the dust sooner rather than later, it is a self fulfilling prophecy at this point.

I've also done my share of free of charge work in my life and i don't know about you, but once i start i usually try to get it right and don't leave it in a half finished state, it is a matter of principle. Either don't do it, nobody will blame you for it, but if you do it, do it properly, otherwise people will end up wishing you wouldn't have done it in the first place.

Knifing enemies is also an issue with low fps
You should check you aim though because people often think its their hitreg
Seriously dog, no one cares thb

DogACTUAL wrote:Sorry tovarish but i've seen one too many times aspiring R-CONS get stonewalled when they got the wrong ideas concerning game balance.

Like who?

DogACTUAL wrote:Oh and why board a sinking ship, your own core DEV team seems to be content with PR biting the dust sooner rather than later, it is a self fulfilling prophecy at this point.

Noice theory. Any proofs for that?

DogACTUAL wrote:I've also done my share of free of charge work in my life and i don't know about you, but once i start i usually try to get it right and don't leave it in a half finished state, it is a matter of principle. Either don't do it, nobody will blame you for it, but if you do it, do it properly, otherwise people will end up wishing you wouldn't have done it in the first place.

I'm glad to see a person who doesn't want ww2 to be released asap without considering the state it is currently in.


Nothing gives us more will to continue developing than this kind of talking, dog... like really, do you want to upset us so much up to a point where your theory will be right?

tl;dr
"I don't know shit about computer science but DEVS BAD"