Too Much FPS, need help!!
-
What’s the best way to record test runs? Is FRAPS still a thing or is there a better way to test?
-
@Atlas I like NVidia FrameView. It’s a wrapper around Intel’s PresentMon dev tool – works ok with AMD, Intel etc too (not NVidia specific).
Emits a CSV file you can load into Excel or Google Docs etc.
But fundamentally these tools all plug into the DirectX call chain to record frame-timings, in pretty much the same way…
-
@airtex2019 Seems like FRAPS is gone… I could use MSI AB/RivaTuner but it only really gives min/max/1%/0.1% but I want an output that can be converted to a graph to show FPS over time. I’ll give FrameView a try.
-
or you can tell us you minimum average and maximum
-
@airtex2019 said in Too Much FPS, need help!!:
@I-Hawk said in Too Much FPS, need help!!:
Force Anisotropic x16
I notice there’s a cfg var … is there any difference to setting it to 16x in NVcpl?
set g_nAnisotropicValue 0 // The max anisotropic filter value to use if anisotropic filtering is ON in the setup UI. Valid values: 0 = max available (default), 2, 4, 8, 16
set transparency AA to Multisample
Interesting I never thought much about this – but how does this interact with canopy-reflection … and HUD/HMCS? Is it an extra pass of AA for everything you see out-the-window? (And presumably a brutal fps impact?) Something fun to try out, for the Nvidia 3090 pilots.
About AF don’t know I’m just used to it from Driver, I think never touched it because it worked for me very good for years.
Yes 3090 here too on a 2K screen, no contest, so I just searched what can I do to get better picture and enabled all those I mentioned above. Well at least until I will get a G2 for VR
-
-
@I-Hawk
i cant believe that you still try to optimize the current version when you can play with the next@ SemlerPDX
i dont know i never used fraps only msi afterburner -
@vfp said in Too Much FPS, need help!!:
@ SemlerPDX
i dont know i never used fraps only msi afterburnerlol - wasn’t a question. Atlas asked about “how to graph” and you replied to him to just tell us his min avg and max. His point of asking how to graph would be to collect more accurate and applicable min/max/avg. though use of a spreadsheet and (hopefully) a repeatable TE or Campaign Save instance to compare one run to another, rather than just looking at some live numbers in the top left of the screen and assuming trying to “do science” with that for improved FPS. One is testable and repeatable, the other is “shoot from the hip” style where a bit too much is assumed with regards to variables and potentially producing inconsistent results from test to test.
Point of file output is to run through the test multiple times, having changed one GFX setting in-game or in-driver, to see if there is a net positive result (whatever results are being sought, that is). Someone may want better picture quality without sacrificing FPS, so their goal would be to change something and then see if that looks better, and then check the graph to compare to other runs.
-
@Atlas For what it’s worth, those are pretty much my exact Nvidia CP settings for my 2070S (except for G-Sync, I don’t have a G-Sync display, and don’t enable V-SYNC for anything - just driving a 1080p 144Hz display, nothing fancy).
Not sure you’ve missed anything, though I’m not aware of whether ALLOW or CLAMP on Negative LOD bias is better in BMS, never tested it.
-
@Atlas said in Too Much FPS, need help!!:
Is there a significant difference between default and JanHas work?
I don’t know, I haven’t used them yet.
-
So I’ve used FrameView and set it to run for 5 minutes. The “no settings” was just my nVidia defaults plus my BMS settings as shown above. The “new settings” was the nVidia control panel set up as shown above and the BMS settings also as above.
I ran the benchmark TE 3x from the ramp, 3x from the runway, and as a comparison, I also ran TE14 set to AP steering and altitude hold to steerpoint 7, with the TGP and Mav pages on the MFDs and pressed the “3” key twice so the MFDs are shown on screen. I would reset the PC after each run and just FrameView and BMS running, no TIR, no Helios, no Voice Attack.
Each test was started at the same time based on the DED time.
The numbers on the right are from each test, the numbers on the left is the average for each type of test. It seems like the “new settings” actually increased my framerates for the most part, rather than decrease it.
FrameView also spat out another excel file but not sure if it recorded a second-by-second framerate or something like that.
System is a 5900X, 32GB RAM, 3080, 120Hz 4K OLED, Win10. OS is on a separate nVME, BMS is on a different nVME, both drives are PCIe 4.0.
I’ll do some testing on the AAR TE tomorrow; it would be interesting to see the performance difference.
-
@Atlas what were the diffs, before vs after? just the texture-filtering settings? and transparency-aa? interesting.
do you still have canopy-cues turned off? personal pref of course… but would be interesting to see if transparency-aa + canopy-reflction takes a big hit. I’ll try it later this evening.
also “prefer max performance” … I’d reached the conclusion that didn’t do much for modern systems with modern bios … but I could be wrong!
-
@vfp said in Too Much FPS, need help!!:
@I-Hawk
i cant believe that you still try to optimize the current version when you can play with the next@ SemlerPDX
i dont know i never used fraps only msi afterburnerI almost never use the current version, only the next, for too long time now already
I mean, no complains, but I want the next to become current
-
@airtex2019 said in Too Much FPS, need help!!:
@Atlas what were the diffs, before vs after? just the texture-filtering settings? and transparency-aa? interesting.
do you still have canopy-cues turned off? personal pref of course… but would be interesting to see if transparency-aa + canopy-reflction takes a big hit. I’ll try it later this evening.
also “prefer max performance” … I’d reached the conclusion that didn’t do much for modern systems with modern bios … but I could be wrong!
Not sure if I understand your question 100% but before would be having everything set to follow global setting and after is having nVidia settings as shown in this post. The in-game settings have not been changed.
-
@Atlas what does it happen if you change the anti-aliasing samples to 2 in BMS graphics setup?
-
@suhkoi69 What do you mean? And why would I do that?
-
@Atlas because I was strongly impacted in term of FPS few months ago…but after writting this post, I tried and I noticed now an impact rather low … 2-3fps between samples 2 & 8 with the TE benchmark
on the other hand, I had noticed that one could modify only the AA via the menu of BMS … because the menu via the graphics card NVIDIA didn’t influence l’AA in game