Dieter wrote: ↑17 Nov 2020, 20:17
Have you ever encountered EMI problems in your house?
Not that I'm aware of, and it is my suspicion that there are plenty of other people that think they have that haven't either.
Does that rule out the possibility? No, but I think it's far less common than some here would suggest.
Dieter wrote: ↑17 Nov 2020, 20:17
Have you ever wondered what is really causing those Frametimes performance issues?
It's not just EMI that causes frametime spikes. In fact, that is the rarest and least vetted of causes. I've been trying to be nice here, but I have little patience for seriously entertaining that theory, unless you're in a third world country on a generator.
Again, I tend to avoid the subject because for whatever reason, it approaches political levels of sensitivity, so I'll leave it for others to discuss.
Dieter wrote: ↑17 Nov 2020, 20:17
If you wanna fix 100% this issue then you got to fix it from its heart. Not doing mumbo jumbo workarounds.
I for one didn't suggest any workarounds, I instead told him to first consider if what he was experiencing was within the norm.
Here's my question, why do people like this think there
shouldn't be frametime spikes? How do they know what should and shouldn't be expected in the first place? What's their point of reference?
As I've said, the most common cause of a frametime spike is when the system is forced to render a frame that requires a completion time longer than a single refresh cycle, which causes it to miss one or more cycles, triggering the previous frame to repeat once or more in the meantime, causing the appearance of stutter.
It's the classic bouncing ball example, where if even one frame is repeated or skipped, it breaks the illusion in an otherwise continuous animation. This is
all that is happening when users experience intermittent or continuous stutter.
If you want to debate why it is being triggered in all cases, that's another thing entirely. But what stutter is
IS known. How it occurs
IS known.
Frametime performance issues on high-end systems is more typically due to outdated adherence to HDD architecture for asset streaming, poor engine/game optimization, and in online games, netcode and/or ISP limitations. People should first and foremost be complaining directly to engine and game creators, who usually solely focus on achievable average framerate targets, and often don't have a clue, and/or couldn't care less about prioritizing frametime performance.
If someone experiencing frametime spikes wants to determine if they potentially have an issue outside the expected norm, they should use Afterburner, play the affected games, and monitor their sessions; if their average spike (not counting game launch or loading screens) is ~50ms or under on average
in gameplay, with the occasional ~100ms or so spike, it's within expected norm on your average system running your average more demanding AAA (especially online) game.
If it's full on 100-500ms+ spikes occurring every second or more in every game, then yes, you probably have an issue that should be investigated.