Q2 2023 Release Candidate Update is Now Available

Discussion in 'News & Notifications' started by John Ratcliff, Mar 31, 2023.

  1. mantasisg

    mantasisg Registered

    Joined:
    Aug 17, 2016
    Messages:
    2,925
    Likes Received:
    3,871
    rF2 is more dynamic and overall more detailed, more realistic because it has parameters for simulating bits that influence handling that others don't even care for, because they has some data that takes all of that stuff and approximates into some number that neglects bunch of stuff that makes car alive, and only shows what is the performance like. And sadly many can't tell the difference, only real gangstas like us can.

    More over, there usually hardly is any data for things that are highly critical for how car behaves over the very limit, I learned it from one iRacing dev blog post years ago. I talk about sliding.

    Of course S397 might have gotten some great unique data from tire manufacturer, maybe because tire manufacturer extracted it mainly for simulation purposes, and what they gave undeniably has to be better than what is possible by just guesstimating through observation and whatever else there is to rely on, but in both ways good stuff can be achieved, and in both ways knowledge and talent is needed. But it is only possible first of all because physics are schemed in such way that it even takes things in account. At the end of the day you should be able to see in the replays quite similar things what can be seen IRL footage, and it should also vibe pretty hard while actually driving onboard in simulated racecar. I vibe a lot with the March. Not so much with McLaren.

    I'd like to say again that in age of data, AI, computers, simulations and overall technology + herding(school classes, university groups with no intimate connection), human sensing, having imagination and making pictures in mind is very underrated. For a normal person it would sound very unbelievable and strange that one could have a power to just decide something without being told what to think, do, say without some dead exact guide, rule or cornerstone AND do it right. So of course everything is 100% data and laser scanned - with no chance of human error.
     
  2. Kahel

    Kahel Registered

    Joined:
    Jun 6, 2020
    Messages:
    262
    Likes Received:
    185
    I was in the same case as @darkojovanoski1986 ... didn't touch my simulator for more than 3 years...
    but found 2 rtx3080 for a good price... once installed (one)... my system became clearly unbalanced... paired with the ryzen 2700x.

    However... while upgrading to the 5700x, did fix the unbalanced system part... it did not (completely) solved the probleme... with rF2 at least...
    Any other sim or game... got good scaling performance... on all aspect...

    But with rF2... there is clearly a limitation with the engine itself... (on some scenarios)... that can't be solved, no matter how much hardware you throw at it...

    To take a example that will demonstrate the extent of the issue...
    With my CPU upgrade alone... I've improved my Overall performance by 70%... when I benchmark a small race... on average.
    Great you would say... and I agree... that seem very good... But... and there is a HUGE... BIG... BUTT... hum.

    The issue lies in the 0.1% and 1%... they only improved by 5-20%... depending on the conditions/tracks/cars...
    Same when you measure the averages... during the first 30 second at the start of a big grid... very low gains...
    So I wasn't happy by my performance at the start of a race... Well despite the upgrade... and 70% uplift performance... OK, it's better... but I'm still not happy... because those 70% uplift are nowhere to be seen... when it count the most...

    So Sure I could throw at it a rtx4090/13900k... and get 150% uplift... with 300fps on average...
    But that would not completely solve my issue... the perf, would not scale, at a start of a big grid... (not that big btw, only tested with 30 drivers... can't imagine a 70 big LFM grid on the nurb at nigh with rain....Insane...)

    What the point... in improving by even 1000% the performance... if you barely get +30%... when it's the most critical...

    Anyway, no matter what... I'm pretty sure you can't solve those 0.1% low (stutters)... that tend to remain... and poor perf at start.
    Which is the most important thing... for me... and I'm not even a VR users...
    So I'm not surprised to see so much complain... and guide to try everything they can to improve it... those poor VR users...

    A shame... probably the result of having such a complex and old engine... But a the very least people should keep that in mind... before spending 3000$ on a computer... Your experience will (most likely) stilll suck... but with very high fps on average :)



    PS: did not test with ULTRA setting... I did my test with the most Optimized setting I could find... with hours of back and forth... try and errors.
     
    Last edited: Apr 20, 2023
    darkojovanoski1986 likes this.
  3. Bernat

    Bernat Registered

    Joined:
    Jun 15, 2020
    Messages:
    732
    Likes Received:
    570
    I think your performance is way below what's to be expected from your system. With a Ryzen 3600 / RTX 2070 Super / 16 GB RAM I can get 90 fps (capped) on all circuits with some tweaking of the graphics. Most settings at high and PFX at medium.

    It doesn't seem an optimization issue. You should try troubleshoot what's hurting the performance on your system.
     
    darkojovanoski1986 likes this.
  4. Bernat

    Bernat Registered

    Joined:
    Jun 15, 2020
    Messages:
    732
    Likes Received:
    570
    The simulation of the physics of vehicles and specifically tires is perfectly known and mathematically modeled with great detail. There's nothing to guess. These models are used in the automotive industry.

    If computers were powerful enough these models could be used in games to get a very realistic simulation. They need lots of data to work correctly though.

    rFactor 2 implements a simplified and optimized version of this model to be able to run on a domestic computer. But it still needs lots of data to feed from.

    Other sims use a simplistic approach of guessing and applying simple calculations that mimic what we think it should happen. They are completely different approaches and that's the reason why they will never have physics as good as rFactor 2 no matter how much they tweak their calculations.

    In the end it's a matter of choosing: do you want a realistic simulation that requires lots of data that has to be squeezed and tweaked to fit into a domestic computer or do you want a less realistic simulation using simpler calculations based on trial and error and requiring no data?
     
    Last edited: Apr 20, 2023
  5. darkojovanoski1986

    darkojovanoski1986 Registered

    Joined:
    Jan 4, 2022
    Messages:
    135
    Likes Received:
    44
    These words are actually what's on my mind... with same cpu, 1660ti and 16gb ram I get 5% less fps than I get now. Something is wrong in the engine I belive.
    Green bar is 80% full which suggest that gpu is not fully utilised but 4070 is 150% faster than 1660ti so in theory I should get big boost in fps like in other sim titles but no...
    Don't want to repear my self but why putting i5-9600k as a reccomended processor if it couldn't cope with the sim?
    Thanks for all the worda and explanations mate!
     
  6. pilAUTO

    pilAUTO Registered

    Joined:
    Nov 7, 2020
    Messages:
    653
    Likes Received:
    570
    @darkojovanoski1986
    Try set opponent on LOW (VRAM impact I think + CPU impact I think)
    And also set visible opponent to 6 to test, it's ok for me but after you can set it to max 10 (GPU impact I think)

    Also my i7 9700K go to 4,9 Ghz without any issue, for your i5 9600K, try an OC to 4,5 Ghz, it will change everything (number of AI.....).
     
    darkojovanoski1986 likes this.
  7. darkojovanoski1986

    darkojovanoski1986 Registered

    Joined:
    Jan 4, 2022
    Messages:
    135
    Likes Received:
    44
    I tried less opponents visible (12 visible on 19 ai race) but increase in fps os marginal. Also I tries to change texture to medium and results are the same. Even tried turning off all except cirquit, player, textures and af x2 and get barely 20fps which is insane :(
    All suggesting my processor can't handle more than 12 ai on good medium or whatever graphic settings I use which is not normal given the system specs I had.
     
  8. Herve_Versabeau

    Herve_Versabeau Registered

    Joined:
    May 27, 2015
    Messages:
    118
    Likes Received:
    30
    Renommez userdata.old et launch game
     
  9. Alex Sawczuk

    Alex Sawczuk Administrator Staff Member

    Joined:
    Oct 4, 2010
    Messages:
    851
    Likes Received:
    971
    I have a a 9600k and it runs just fine on the whole (not outstanding, but ok). If you barely get 20fps there's clearly some other issue going on.
     
    Rui Santos likes this.
  10. Lazza

    Lazza Registered

    Joined:
    Oct 5, 2010
    Messages:
    12,298
    Likes Received:
    6,543
    I'm pretty sure the "20fps" was about the fps gain (from lowering settings), not the absolute rate.

    But it's hard to keep track, and too many assumptions being made generally (like the CPU not being good enough, which seems to have become its own thing now).
     
    darkojovanoski1986 likes this.
  11. darkojovanoski1986

    darkojovanoski1986 Registered

    Joined:
    Jan 4, 2022
    Messages:
    135
    Likes Received:
    44
    The gain from turning off all other stuff is around 20fps depending of the track. Alex against how many AI you race and how visible? Just to compare with what I get...
    From my understandig I have sufficient Vram and Ram memory to get into trouble of loading tracks and cars textures so that would not thortle the cpu. I think i5 9600k is more capable than results I get. But judging from results from other sim titles i play should run rFactor 2 without issues even in more demanding scenarios. Please, what results you get against ai in terms of fps lowest/highest?
     
  12. Lazza

    Lazza Registered

    Joined:
    Oct 5, 2010
    Messages:
    12,298
    Likes Received:
    6,543
    If you haven't already, can you describe your whole setup? Screens/resolution etc. Actual content you're using matters (track and cars). From my point of view all I'm seeing is snippets (from you, and others) which makes actual comparisons very difficult.

    @darkojovanoski1986 actually, your own thread would make all this much clearer, rather than in here.
     
    darkojovanoski1986 and Bernat like this.
  13. Bernat

    Bernat Registered

    Joined:
    Jun 15, 2020
    Messages:
    732
    Likes Received:
    570
    My Ryzen 3600 has similar performance than your CPU and your GPU is far more powerful than mine. I'm limited by my GPU. You should have way better performance than me. There's something tricky going on here and I wouldn't make fast assumptions about the cause.

    Have you monitored your CPU while running the game? CPU/core utilisation, processes running, etc.

    In case you just find out rFactor 2 is bottlenecking the CPU, I would create a performance log and upload it here. Maybe someone at S397 can take a look to see what's happening.

    I'm inclined to think that it's a system misconfiguration or something behaving badly on Windows but I wouldn't make this assumption neither.
     
    darkojovanoski1986 likes this.
  14. darkojovanoski1986

    darkojovanoski1986 Registered

    Joined:
    Jan 4, 2022
    Messages:
    135
    Likes Received:
    44
    I would not open thread just to cry about it, i love rFactor 2 and I am not that kind of person. Just posted here becouse I saw other complains about performance issues from other users too. Here are my specs:
    CPU: i5 9600k
    GPU: Asus Dual 4070 12gb vram
    RAM: 4x8gb 2666Mhz
    PSU: 750W
    I have SSD and HDD but performance were the same.

    My in game settings:
    Msaa: 2x
    Post process: low
    Cirquit, player, opponent, textures to full
    Special effects: medium
    Shadows: High
    Shadow blur: optimal/quality
    SSR: off
    Rain: low
    Resolution is 2560x1080

    My ~ results vary from track to track and AI against I race and also time of the day. In sunrise/sunset against 19 ai all visible at start of the race I get around 60 fps on most tracks, on Daytona in the first corner after race start I get dip to around 50-55fps. On the middle of the day of course results are better but again even this is not the result I expect if I compare results from other sims I play which are more cpu and gpu demanding like ACC for example.
    Worth noting is that I had 1660ti in my pc until tuesday and results I got with it were around 5% lower than this.
    All I wrote here are results after formatting pc and installing rFactor on formatted drive.
    I really don`t know what to think. I haven`t many background processes becouse I just use this pc for sim racing, only essentiall stuff is installed, not other fancy programs for streaming or anything.
    Thanks guys for hearing me and giving me suggestions...
     
    Mydriaz likes this.
  15. Highlandwalker

    Highlandwalker Registered

    Joined:
    Nov 11, 2016
    Messages:
    795
    Likes Received:
    2,240
    I see a lot of people are blaming rF2 for there bad performance when actually it could be more down to a badly optimised PC. On my system rF2 runs no worse than any other sim. I run an I7-8700K at 4600GHz with 32gig and a 3080 at max settings at 4k. I just wonder how much crap people have running in the back ground, Windows is notorious for this. I've been running rF2 since 2016 and any time rF2 was running badly it was the PC and not rF2 but the problem affected all games. Performance could be better in rF2 and also in all games. Virus checkers can cause bad performance, I think Norton is one of the worst. This what I've found over the seven yrs I've been running rF2. This video could help a lot of you, follow the steps out lined and you should get better performance and gain a lot of hard drive space. I apologise if this may upset some of you.
     
    darkojovanoski1986 likes this.
  16. Bernat

    Bernat Registered

    Joined:
    Jun 15, 2020
    Messages:
    732
    Likes Received:
    570
    I really think any issue deserves its own thread. This isn't the right place. If you genuinely want to try to fix the issue, that's not crying, you're helping yourself and others that might be seeing the same. The more information we can collect in one place the easier it will be for anyone wanting to investigate the issue.

    With the information you've given I think anyone will confirm there's something very wrong with your system. Maybe something that rFactor 2 doesn't like about your system. It should be fixable but we need more information.
     
    darkojovanoski1986 likes this.
  17. darkojovanoski1986

    darkojovanoski1986 Registered

    Joined:
    Jan 4, 2022
    Messages:
    135
    Likes Received:
    44
    I opened thread "Low Perfomance" in Tecnical support and wrote the post there...
     
  18. mantasisg

    mantasisg Registered

    Joined:
    Aug 17, 2016
    Messages:
    2,925
    Likes Received:
    3,871
    Our little interaction was about parameters that goes into model not a model.

    These models don't need data to work correctly, they need correct parameters to work correctly and data is supposed to help achieving correct parameters. In some sims data is the parameters, but those parameters aren't necessarily physically accurate to describe how car/tire works.

    Back to 1970s F1 cars topic where we stem from, my bet is that March runs better parameters than McLaren. And it is likely that they had more data to work with McLaren, more and better data doesn't guarantee more realistic physics, you still have to have decent idea what is the target of car performance and what the result handling should be like, data is not being fed up straight into physics, especially in rF2. At least I haven't heard yet about much data that feeds straight into parameters.

    Yes, and from what I have heard while participating in AC modding, and from some people who used to mod rF1 physics, these sims are more straight forward and more simple to work with data, mostly talking about tires. I can remember these opinions from modders like mclarenF1papa and Niels Heusinkveld. They are very respectable modders, but they chose staying with less complex physics models because they understand them better, they have easier time applying data. They do so even if potential of realism with those models isn't as high. Here is where you are wrong - rF2 is more realistic, but there might be more trial and error and less direct application of data.

    rF2 really runs two or even three tire models. Old simple model for AI. QSA model to pre-calculate look-up data that will be used by other model, QSA is mostly tire structural stuff as much as I understand and it is done with ttool, for sure you need data about how tires are constructed, I never messed with ttool. Third model and source of where most of changes to handling happens is Real Time model, I haven't heard from anybody about finding much specific data (or any) for it, perhaps S397 could if they get proper connections with some real tire engineers from the secret world of tires, but probably still they only find some probable ranges where the parameters can be, and then do the fine tuning, it is quite intuitive to fine tune if you understand what the parameters does and how car should behave. At the end RL data might help while evaluating physics, trying to match RL data in game. At the end there might be few ways how car in sim could handle, but still meet the data, here is where comes the ultimate underrated data - observing how real stuff behaved from sources like footage and pictures. IMO when physics are made made it should start and end there. Most importantly such approach makes historic cars more attractive, overly glorified data kills appeal for historic racecars.
     
  19. sman

    sman Registered

    Joined:
    Mar 14, 2013
    Messages:
    81
    Likes Received:
    22
    I Like RF2 a lot but no idea why people like E cars the noisyless cars it gives totally ZERO immersion (also in RL E racing a joke)
     
  20. davehenrie

    davehenrie Registered

    Joined:
    Jul 6, 2016
    Messages:
    7,424
    Likes Received:
    4,340
    Why do some people like cars with fenders? or even a roof? Why are mumbling vacum cleaner sounding turbos preferred over screaming V10's.
    Why isn't the Viper the single greatest sim car in history? EVERYONE has different tastes.
     
    Emery and Superdevil like this.

Share This Page