Vsync vs fastsync: question about frame-drops.

Everything about latency. Tips, testing methods, mouse lag, display lag, game engine lag, network lag, whole input lag chain, VSYNC OFF vs VSYNC ON, and more! Input Lag Articles on Blur Busters.
Post Reply
aweigh0101
Posts: 8
Joined: 30 Oct 2018, 22:08

Vsync vs fastsync: question about frame-drops.

Post by aweigh0101 » 11 Jan 2019, 03:05

What if I'm capping the frame rate to 45fps, because for whatever reasons this is the FPS I want?

- 80hz Monitor (it's a 75hz monitor OC'd to 80hz)
- 1980p resolution
- FAST SYNC Enabled
- RTSS frame-rate cap set to '45fps' (40fps has too much ghosting when panning the camera, and the game(s) I'm playing don't reach a consistent 60, so I found 45fps to be a good value)

Scenario: Superposition Benchmark demo set to custom/medium settings. It runs at a steady 45fps (capped to it) throughout the entire demo... EXCEPT at precisely scene 8 or 9 it will momentarily drop to 41-42fps then immediately go back to 45 until it finishes.

My goal is to make it never drop a single frame from the 45fps, to begin the demo and end it with 45fps locked. I know the frame-drop isn't related to graphics settings cos I ran the benchmark with easier settings and the frame-drop still happens in the same place/scene of the demo.

Is the frame-drop related to my settings (fast sync enabled with cap of 45fps in RTSS)? I try it of course with regular vsync instead of fast sync and the frame-drop happens in the same scene/place (still using same cap of 45fps); so with both regular vsync and with fast-sync the Benchmark demo drops a frame in the same place.

I'm going nuts here trying to figure out what I can do to not drop any frames, and I also tried all the values for max frames to render ahead as well from "Let App decide" to '1' all the way to '8' and no change in the frame drop. Scanline-sync is out of the picture because I can only use it to lock to 80Hz or to half of that, which would be 40, and I don't want to run game(s) at 40fps.

Am I just nuts and the frame drop could be caused by any number of factors that are completely unrelated to the kind of vertical synchronization I am using or to the fps Cap? I considered this... but I dunno, seems to me a benchmark that can easily run in +70fps from beginning to end, IF WHEN LOCKED TO 45fps, it stands to reason it should therefore _never drop below 45_.

Right? Hehe, thanks in advance for humoring me to anyone who answers.

EDIT: Oh, forgot to mention that the same happens (the frame drop) WITHOUT Vsync, i.e. with vsync/fast-sync off and only using RTSS frame cap. In fact, ONLY using the RTSS frame cap of 45fps made 1 additional frame drop happen sooner in the demo sequence of scenes.

User avatar
RealNC
Site Admin
Posts: 3737
Joined: 24 Dec 2013, 18:32
Contact:

Re: Vsync vs fastsync: question about frame-drops.

Post by RealNC » 11 Jan 2019, 04:45

Try the nvidia limiter instead. RTSS will prevent the CPU from running ahead of the GPU by keeping all pre-render buffers empty. This has low input lag, but in this case it interferes with your goal. The nvidia limiter might be more suitable here, since you don't care about input lag when running a bench demo.
SteamGitHubStack Overflow
The views and opinions expressed in my posts are my own and do not necessarily reflect the official policy or position of Blur Busters.

aweigh0101
Posts: 8
Joined: 30 Oct 2018, 22:08

Re: Vsync vs fastsync: question about frame-drops.

Post by aweigh0101 » 11 Jan 2019, 04:50

thanks, will try right now.

aweigh0101
Posts: 8
Joined: 30 Oct 2018, 22:08

Re: Vsync vs fastsync: question about frame-drops.

Post by aweigh0101 » 11 Jan 2019, 05:00

RealNC wrote:Try the nvidia limiter instead. RTSS will prevent the CPU from running ahead of the GPU by keeping all pre-render buffers empty. This has low input lag, but in this case it interferes with your goal. The nvidia limiter might be more suitable here, since you don't care about input lag when running a bench demo.

....IT WORKED!!!!!

Min: 44.95 (essentially it never moved)
Avg: 45
Max: 46.xx

I get what you mean by RTSS preventing "run ahead" and stuff, as RTSS never had this kind of going past the specified number (45 in this case); in RTSS it would never go over 45.

HOWEVER, while nvidia limiter did go up to 46 sometimes, the minimum never "dropped" and the benchmark reported it as 44.95, and more importantly: I *personally* did not see a single "frame drop" / "stutter" throughout the benchmark like I had been seeing with RTSS.

For my specific purposes this means I will be using nvidia limiter in games more often now, though It's yet to see if this will actually translate to "smoother pacing" (no dropped frames) in actual game play, as it is just a demo. Still, it worked! Thank you.

EDIT: BTW I used the NV Inspecter's "Default" application of the Limiter instead of the "V2" options because since you said it was related to the way RTSS minimizes lag/delay I figured I should use the Limiter with the least lag compensation (V2 options say they reduce lag).

EDIT 2: Also I noticed that the NV Limiter worked my GPU less, I saw slightly lower %'s for GPU usage and GPU temp. Not a lot less, really just 1 or 2 numbers less, but definitely less usage/temp.

User avatar
RealNC
Site Admin
Posts: 3737
Joined: 24 Dec 2013, 18:32
Contact:

Re: Vsync vs fastsync: question about frame-drops.

Post by RealNC » 11 Jan 2019, 06:36

Note that when letting the CPU run ahead of the GPU you will get more input lag though. The lower the FPS cap you're using, the more severe the input lag becomes when using the nvidia limiter.
SteamGitHubStack Overflow
The views and opinions expressed in my posts are my own and do not necessarily reflect the official policy or position of Blur Busters.

Post Reply