FPS - any advice?
-
You will be surprised how different the tabulated numbers are form the “observed”, we tend to see the better numbers.
No idea what that means
Falcon FPS measurement simply measures the time it takes to run the main loop and then convert it to FPS, it’s as simple as that. Although all kind of urban legends I’ve seen through the years, the program cannot bullshit itself as it is the most obvious thing to measure, and it’s the most accurate since it is 100% code based and not some external program kind of measurement (Although that is of course pretty accurate as well, usually).
IMHO, if your FPS counter shows most of the time something between 90-100 then your average will be somewhere between those and not lower - Averaging is again a quite dumb measurement:
Count number of frames until now
Accumulate the counter every frame
DivideThere are no urban legends around FPS measurement, and it’s actually one of the first things every starting programmer should learn (Either it is a game engine or some simple calculations code, you want to know how much time you spent on running “the main loop”)
-
No idea what that means
It means exactly as I stated. And I thought I was reasonably clear. It has nothing to do with the validity of the internal FPS counter !
As a TE progresses the FPS varies, I tend to put a heaver weight on the higher numbers and guesstimate a higher “average” FPS if I only view the counter.
When I actually record them over a number of runs using “Fraps” for example and then Average the results of all runs in a set then I get a reliable Sample.Every run of a TE is rarely identical. And does vary substantially. No matter how well one eliminates user input.
I though you would know this.
-
As a TE progresses the FPS varies, I tend to put a heaver weight on the higher numbers and guesstimate a higher “average” FPS if I only view the counter.
When I actually record them over a number of runs using “Fraps” for example and then Average the results of all runs in a set then I get a reliable Sample.Every run of a TE is rarely identical. And does vary substantially. No matter how well one eliminates user input.
I though you would know this.
Well, I think we are talking apples and oranges here then. If I measure FPS for the sake of measuring FPS then I will choose a very lite TE and try to measure at places where “special factors” will have less effect.
Example of what I mean specifically in the context of this thread:
If we want to help a guy to measure FPS, then we won’t send him to measure FPS in a TE in a busy airbase with 20 ACs waiting for takeoff, right? In order to get a clean measurement, you should create a clean test.I’ll tell you what, in my daily job and also in BMS development, I’m involved a lot with tracking performance, improving, etc. 1 thing I’m sure of, you want to get a clean measure, you make it as clean as possible, then if you wish to find bottlenecks, you proceed until you find it.
So, for me, when I speak of TE for measuring FPS, I mean something clean, I’ll even say “as clean as possible”. And in my books clean TEs should have EXACTLY (or VERY VERY close) FPS, immediate or average. I’ll even say that as I’m working on something currently that involves a LOT with system, HW and performance (I assume you understand what I mean), I havea couple of benchmark TEs where even slight moves from the “known FPS” at different states/areas/views of the TE, may put me on my feet and check what I did wrong or why this change or the other made such a huge effect. Good that I have not only BMS but also a much more dedicated App where I have even cleaner and VERY stable measurements of how much a change affects performance.
Bottom line - I TOTALLY disagree with:
Every run of a TE is rarely identical. And does vary substantially
If you run same TE, exit, exit sim, reenter sim and run same TE, you should get EXACTLY same performance. At least that’s what I see here.
-
I think remark about BMS restart is important.
For example in my Nordic TE (4x Gripen vs 4x F15 Agressor with custom winter textures) I have about 90 FPS (stable, just few acceptable drops when flying through explosions) on my intel HD3000 (i7-2670QM @2.2 GHz) - for first or even second time I fly this TE. If I run it more times it goes quite stable 25fps with some peaks to 75fps.
After restart I have 90 again. -
I think remark about BMS restart is important.
For example in my Nordic TE (4x Gripen vs 4x F15 Agressor with custom winter textures) I have about 90 FPS (stable, just few acceptable drops when flying through explosions) on my intel HD3000 (i7-2670QM @2.2 GHz) - for first or even second time I fly this TE. If I run it more times it goes quite stable 25fps with some peaks to 75fps.
After restart I have 90 again.Wow that a big variation 90fps to 25fps & peaks @ 75fps. Some thing is seriously wrong.
-
Well, I think we are talking apples and oranges here then. If I measure FPS for the sake of measuring FPS then I will choose a very lite TE and try to measure at places where “special factors” will have less effect.
Yes you are, “special factors” is what your looking for while you may be involved in changes to implementations or other change to how Falcon works. Additions & alterations to the Application. A very specific, stable & reproducible results TE is what you need. But that is not Lodestar’s situation.
Example of what I mean specifically in the context of this thread:
If we want to help a guy to measure FPS, then we won’t send him to measure FPS in a TE in a busy airbase with 20 ACs waiting for takeoff, right? In order to get a clean measurement, you should create a clean test.- I TOTALLY disagree with:
This is exactly one of the circumstances that will be applicable. As none of us fly falcon in a dead TE. That’s your job.
And that brings me to another point. I took the trouble to produce results in what was a graphical TE so I could eventually show Lodestar comparable results of pushing His RAM faster. He will grab a few more FPS out of what is a very powerful Rig. And here you are trying to rap my knuckle.I’ll tell you what, in my daily job and also in BMS development, I’m involved a lot with tracking performance, improving, etc. 1 thing I’m sure of, you want to get a clean measure, you make it as clean as possible, then if you wish to find bottlenecks, you proceed until you find it.
“I’ll tell you what” Good one I-Hawk ! That hat is getting chewed. Do you think your the only one who has had to problem solve in a technical or engineering environment.
So, for me, when I speak of TE for measuring FPS, I mean something clean, I’ll even say “as clean as possible”. And in my books clean TEs should have EXACTLY (or VERY VERY close) FPS, immediate or average. I’ll even say that as I’m working on something currently that involves a LOT with system, HW and performance (I assume you understand what I mean), I havea couple of benchmark TEs where even slight moves from the “known FPS” at different states/areas/views of the TE, may put me on my feet and check what I did wrong or why this change or the other made such a huge effect. Good that I have not only BMS but also a much more dedicated App where I have even cleaner and VERY stable measurements of how much a change affects performance.
Bottom line - I TOTALLY disagree with:
I don’t have a problem with the validity of the methodology you use for the purposes you are using them for.
I DO UNDERSTAND. I’m not an IDIOT.If you run same TE, exit, exit sim, reenter sim and run same TE, you should get EXACTLY same performance. At least that’s what I see here.
Around we go.
One of the characteristics that Falcon and its derivatives is that almost every time you go up in the same TE or Campaign mission it plays out differently. Replay-ability.
:tjacked:
And that’s all I have to say about that.
-
Many thanks for all your responses.
The resolution is 1920x1080 75hz. I have Vsync off in addition to mulitisampling.
I have used the TEs within Korea and the campaign has been Balkans. Is it likely the FPS draw is specific to Balkans?
One other thing is that my monitor uses Freesync. Should this be switched off?
I had to save on the RAM but could change this…Will look at the O/C but am on a stock cooler.
All responses appreciated.
If its summer there then there is a good chance its thermal throttling, so get a CoolerMaster EVO 212 cooler, good and good value.
Jump into your BIOS and apply the Extreme Memory Profile (XMP) for your RAM. if you can even push the clock a tad more , all the better.
I’m out of here. :bolt:
-
If its summer there then there is a good chance its thermal throttling, so get a CoolerMaster EVO 212 cooler, good and good value.
Jump into your BIOS and apply the Extreme Memory Profile (XMP) for your RAM. if you can even push the clock a tad more , all the better.
I’m out of here. :bolt:
Changed to XMP for RAM in the BIOS. I’ve messed around with the GPU settings and am hitting 75~ now for Korea Campaign.
Definitely intending to change the stock cooler as l don’t trust the temps with overclocking.
Agree with many that lower FPS isn’t a deal-breaker but on my old PC, TGP and using Mavericks knocked it on its ass.Looking forward to Nordic once it is released for 4.34 as l got decent FPS in that.
-
I must admitt terrain textures are only 256x256, but I have trees on, even enviro reflection (low-res) for water is on, AA on….
For integrated graphic card Gouraud shading (instead Phong) and shadows off are quite essential. -
Yes you are, “special factors” is what your looking for while you may be involved in changes to implementations or other change to how Falcon works. Additions & alterations to the Application. A very specific, stable & reproducible results TE is what you need. But that is not Lodestar’s situation.
- I TOTALLY disagree with:
This is exactly one of the circumstances that will be applicable. As none of us fly falcon in a dead TE. That’s your job.
And that brings me to another point. I took the trouble to produce results in what was a graphical TE so I could eventually show Lodestar comparable results of pushing His RAM faster. He will grab a few more FPS out of what is a very powerful Rig. And here you are trying to rap my knuckle.A guy come with a question if his FPS are OK. The first thing I do is tell him to start clean, that is regardless of development tests, that is for finding bottlenecks and for understanding better where his system starts to fake. If you want to make conclusions about FPS in BMS while flying campaign mission or complex TEs, then what can I say, Good luck with that.
Start clean and proceed slowly, until you hit the problem. You should know it, no?
“I’ll tell you what” Good one I-Hawk ! That hat is getting chewed. Do you think your the only one who has had to problem solve in a technical or engineering environment.
No. But still, I think that I have a more touchy eye than most for noticing performance changes.
I don’t have a problem with the validity of the methodology you use for the purposes you are using them for.
I DO UNDERSTAND. I’m not an IDIOT.My methodology for measuring FPS should be a “base case” for anyone measuring FPS. Start clean and proceed slowly, until you hit the problem.
One of the characteristics that Falcon and its derivatives is that almost every time you go up in the same TE or Campaign mission it plays out differently. Replay-ability.
Apples and oranges, again. You speak from the start about average FPS and involve in “none dead TEs” or even (god help me) campaign missions.
Dead TEs are exactly the clean test I speak of. That must be the base/reference of any FPS measurement.
Gameplay factors of Falcon may have huge effect on FPS. If e.g a 4-ship flight was killed in 2 minutes in the previous mission, and in another test of the same mission they all manage to avoid being destroyed, and they proceed and destroyed a huge factory complex, and now you have smoke rising from that area, you may see real FPS differences compared to the previous test. But what conclusion you can make of that?? None!Even if after the base test (clean TE) you wish to find some kind of measurement for campaign, compare maybe between different systems, you should mainly test specific areas and best to do it in the first minute(s) of campaign, because as more as the mission continue, your current test has more changes to divert from the previous one. And you can’t conclude much from it.
-
So, to check understanding:
1. measure FPS preferably with the in-game counter
2. start and measure FPS in a clean empty TE
3. add or subtract TE features, shaders and/or gpu settings step by step and measure impact on FPS
4. anything / anywhere around 60 FPS is good enoughThough I’m looking for specific advise as per this post: link
I upgraded from GTX 1060 to RTX 2080 Super and tested performance in the benchmark Test TE (both on same cpu). I noticed zero improvement in FPS. I expected at least “some” increase in FPS, like I remember how my older gpu upgrades would increase FPS a bit.
I mean, the 2080 should push the pixels faster right, based on its higher power compared to the 1060?
What am I missing here? Any advise?
Thanks
-
Hi Scuby, good to see you
It could be that at the current GFX engine state, and especially at 1080, the GPU isn’t the bottleneck, hence you may see no difference (Although I agree, it is weird, if you asked me before I’d say there should be at least something).
With 2080 I suggest to push into 4K monitor, there you will for sure see a difference.Also this card will be very valuable for what’s coming in the future.
-
Thanks I-Hawk, good to see you too !
Roger that, I know what is coming! hehe.
Now, for the current BMS exe, I’m happy that you would also expect at least “some” improvement of a performance… that means I’m not totally stupid!
I will tinker some more this weekend. I want to start by clean and re-installing all NVIDIA drivers and BMS itself. Better start with this to be sure I have an absolute clean starting point. Then, I expect maybe some super secret new expensive NVIDIA setting might also be the culprit here…
Will report later. If anything comes to mind in between, please let me/us know here, I’d appreciate it!
-
Thanks I-Hawk, good to see you too !
Roger that, I know what is coming! hehe.
Now, for the current BMS exe, I’m happy that you would also expect at least “some” improvement of a performance… that means I’m not totally stupid!
I will tinker some more this weekend. I want to start by clean and re-installing all NVIDIA drivers and BMS itself. Better start with this to be sure I have an absolute clean starting point. Then, I expect maybe some super secret new expensive NVIDIA setting might also be the culprit here…
Will report later. If anything comes to mind in between, please let me/us know here, I’d appreciate it!
Good luck dude. It takes time and trial and error - l’m still looking, but am at an OK level now.
This is a great place for support.
-
read the ryzen thread on an answer for this, and some aspects you forgot.
Sent from my SAMSUNG-SM-T818A using Tapatalk
-
-
Anyone looking for higher FPS in Falcon should turn off “Focus shadows” that is on default, in the configuration. Its a pig.
Water Enviroment Mapping- uncheck.
Reduce particle system is a good check also.
Measure Fps in clean TE. I get 130 fps over water in clean TE. System: 3770 I7, 1070gtx and SSD. The rest is mainly cpu - gpu issues.
software as GPU-Z running with the tab “sensors” open on your second monitor (if you have it) and you´l see that your GPU is almost idling in Falcon even with FPS at 25-30. I´v tried Nvidias config to Maximum performance on the power config, but it doesnt seem to help. I think this is mainly beacause of a bottleneck CPU- GPU as the calcs need to be run before sneding it to GPU? My tiny understanding.
Maybe a check at the particle sys, skins, etc etc. I think the campaign are mainly what sets the limit, but implementing more and more “eye-candy” in a sim that has an engine from the dark ages is like asking a stone to fly.
I love flying Falcon because its a great, great simulator. It works. It works in MP. It works with cheap systems.
For the guys that love to be able to read the labels on the slammer, (Eyecandy) there are homemade skins and such to download and use; BY CHOICE.
Keep it as clean as possible please.
Cheers!