Animation Time Graph Spikes Normal ?

[NHP]

Hi everyone,

I am sorry if this isn't the right place to ask about this, but hopefully it is.

I am wondering if such spikes on the Animation Time Graph are normal and not to be worried about ? I Fresh installed windows, installed necessary drivers, but no RGB software or anything else optionnal, and still got them.

Although, I still manage to get a flat graph for a few seconds, the screenshots are really taken in moments wherethese happen, but in between there would be 30, 20 seconds of flat grapph

Screeshot 01

Screenshot 01.png (38.39 KiB) Viewed 470 times

Screeshot 02

Screenshot 02.png (36.23 KiB) Viewed 470 times

Screeshot 03

Screenshot 03.png (26.07 KiB) Viewed 470 times

(Screenshots in attachement)

[RealNC]

Yeah, it's fine. Especially when not using the "ultimate" power plan in Windows (preventing CPU power saving, core parking, etc.) and having something else open as well, or Windows background processes that use the CPU from time to time. The browser itself as well (it periodically writes data to disk, at a pace of like 200kb/s, all the time non-stop.)

I would say getting a perfectly flat graph for an extended period of time is pretty much impossible on modern Windows versions and modern browsers. They just do way too much crap in the background all the time.

[NHP]

Thanks a lot for your reply RealNC. That's helpful!

Good to know, I am using balanced mode because I have 7950X3D, but I'll give the high Performance mode a try!

[Chief Blur Buster]

This graph is generally pretty much fine.

This is generally good if you do not see very frequent (every 1 second) red spikes with an idling browser as the foreground app.

Browsers are simply a canary in the coal mine, they can often be more sensitive to stutters than many DirectX/Vulkan/OpenGL games. It's great when you need to diagnose a "I've got a really nasty background process that even a webapp can 'see' the jitter of" problem (like background RGB software).

Some demo kiosks and display reviewers use custom slimmed-down Windows installs (even dual boot to them) as well as game-optimized OSes to try to calm down this graph, to make TestUFO run really smooth.

Performance Mode and Ultimate Performance help, as does raising the priority of foreground app. That said, Chrome run a lot of background apps that have varying levels of priorities.

[NHP]

Hi there!

Thanks for answering! Yeah, when idling, I don't ge any stutter warnings and spikes never exceed green color, unless I mvoe my mouse around the tabs or things like that where it feel like it makes sense tha it unsyncs.

However a quick question, the reason i'm looking at all these blur buster tests, in some games, on my 144Hz monitor, when capping fps, there is a massive amount of screen tearing (Counter Strike 2 ; League Of Legends ; ...) However, when I unlock FPS it's much much smoother. Isn' this counter-intuitive ???

Sorry if this thread is going somewhere else haha!

[RealNC]

However a quick question, the reason i'm looking at all these blur buster tests, in some games, on my 144Hz monitor, when capping fps, there is a massive amount of screen tearing (Counter Strike 2 ; League Of Legends ; ...) However, when I unlock FPS it's much much smoother. Isn' this counter-intuitive ???

Without vsync (or any form of sync) there's gonna be tearing. FPS capping and syncing are different things. When tearing, the higher the FPS, the less noticeable tearing becomes. But it's always gonna be there, unless you start using some form of sync (vsync, g-sync, freesync, etc.)

[NHP]

However a quick question, the reason i'm looking at all these blur buster tests, in some games, on my 144Hz monitor, when capping fps, there is a massive amount of screen tearing (Counter Strike 2 ; League Of Legends ; ...) However, when I unlock FPS it's much much smoother. Isn' this counter-intuitive ???

Without vsync (or any form of sync) there's gonna be tearing. FPS capping and syncing are different things. When tearing, the higher the FPS, the less noticeable tearing becomes. But it's always gonna be there, unless you start using some form of sync (vsync, g-sync, freesync, etc.)

I totally understand what you're saying, however, I am using the same screen I was using on a previous 1080ti build and I am noticing so much more screen tearing with this new 4080super build. I don't get it, it's like waaay too obvious. Perhaps, my monitor is bad ? (MSI g241)

[RealNC]

I totally understand what you're saying, however, I am using the same screen I was using on a previous 1080ti build and I am noticing so much more screen tearing with this new 4080super build. I don't get it, it's like waaay too obvious. Perhaps, my monitor is bad ? (MSI g241)

Maybe you had g-sync enabled?

Edit:

Oh, wait. 10 series cards don't support MPO (multiplane overlays.) So you might have been vsynced with the 1080 Ti when not using exclusive fullscreen. The 4080 S supports MPO, which means it's possible to have vsync off even without exclusive fullscreen, especially in Windows 11 where this happens even in non-borderless windowed mode. (In Windows it's possible with some games in borderless windowed mode.)

Or you might have been using g-sync and it has nothing to do with MPO or Windows 11.

[dervu]

I noticed this test not having any yellow/red spikes at all when running over 10 seconds with nvidia power management mode set to full performance for chrome, but when I set nvidia power management mode to normal there are both yellow and red spikes pretty often.
It's reproductible every time.
Would GPU performance matter so much?

It's not like PCI-E speed would drop or anything. Only GPU memory and core speeds drop.

[RealNC]

I noticed this test not having any yellow/red spikes at all when running over 10 seconds with nvidia power management mode set to full performance for chrome, but when I set nvidia power management mode to normal there are both yellow and red spikes pretty often.
It's reproductible every time.
Would GPU performance matter so much?

It's not like PCI-E speed would drop or anything. Only GPU memory and core speeds drop.

It's probably the actual clock switching that might cause the spikes.