Powerful system (9800x3d, 5070 Ti), but Overwatch 2 on the lowest settings has suboptimal 0.1% / 1% lows

[TN_fun]

Hey guys, I don't understand why you are ignoring my message. Please test with replay code. Raw gameplay can cause frame drops from network code and packet loss.

[Maelstrom]

Hey guys, I don't understand why you are ignoring my message. Please test with replay code. Raw gameplay can cause frame drops from network code and packet loss.

I'm testing in the practice range, frankly a replay code test seems a little pointless when the possibilities are:
issue present in replay code -> ok what now
issue not present in replay code -> ok so the netcode in the game is botched, nothing I can do to fix it
But I can test it later tonight.

[Fpsgamer]

Had same issue on cs2 with my 7800x3d.
My FPS were going from 600 on mirage to 300 the next game.
I switched to w11 and got my bios optimized and my fps are pretty consistent now.

[TN_fun]

Hey guys, I don't understand why you are ignoring my message. Please test with replay code. Raw gameplay can cause frame drops from network code and packet loss.

I'm testing in the practice range, frankly a replay code test seems a little pointless when the possibilities are:
issue present in replay code -> ok what now
issue not present in replay code -> ok so the netcode in the game is botched, nothing I can do to fix it
But I can test it later tonight.

The replay code is easier to work with. If there is a problem in the code, we can ask someone else with a high-performance system to check it.
Then it will be clear whether there is a problem on another system.

If there is no problem in the replay code, then everything becomes more complicated. Because it is either the system network stack or the game network code.

[kyube]

Sure but I'm taking the OP at face value when they claim that they are aware of other users with significantly better 1% lows. If it's just a game specific issue then I'd not expect a discrepancy between systems. Your other suppositions fall under "it's possible that it's just a more common setup issue than we think", so I'm not sure why you start your comment disagreeing with me.

Yes, I agree with that sentiment that it's a common setup issue.
However, I don't agree that it's caused by anything specific to X3D CPUs. I think it's a RAM or BIOS issue, along with the usual awful frame pacing that in-game frame limiters provide.

At this point I'd just like a single positive example where someone is benchmarking OW2 on an X3D CPU with good 1% and 0.1% lows before (recommending) spending significant time and effort diagnosing their system since it's more likely that there is nothing behaving incorrectly.

Understandable.
Regardless of quantifying a objective performance metric for the X3D CPUs in OW2, I don't see a (theoretical) reason why a particular CPU architecture would be detrimental to performance over Intel's 13700k or newer options. Games can only benefit from larger L3 cache, especially a low latency one (~10ns)
This is why I suggested to rule out the possibility of HW instability, especially with spikes as severe as OP's and the comments he has made afterwards.

I'm testing in the practice range, frankly a replay code test seems a little pointless when the possibilities are:
issue present in replay code -> ok what now
issue not present in replay code -> ok so the netcode in the game is botched, nothing I can do to fix it
But I can test it later tonight.

There are 2 replay codes which are repeatable benchmarks for OW2. I find the 1st one to be a bit better in terms of run-to-run variance.
I'd definitely suggest to anyone to try & run the 1st one with PresentMon CLI in the background to evaluate performance.

HEZHJ — https://workshop.codes/HEZHJ
DK3VN — https://workshop.codes/DK3VN

[mel]

Enable c-states and dont use any custom power plans

[netborg]

Testing in practice range isn't the real thing, because shaders eventually are all compiled and you won't get any stutters there. I tried testing replays too, but for some reason they also can't reproduce the real thing. So in the end I just logged every frame in real gameplay which makes the most sense to me. Also dxvk has a message popping up in the hud every time new shaders are compiled, which made me aware those are the troublemakers. When playing like 5 matches in a row, those eventually get less and less, but every time you restart OW, shaders need to get compiled again (why no one looked into shader caches?).

Not too experienced with the x3d cpus, but larger cpu caches surely aren't helping with compiling shaders.

I agree I wasn't too happy with Overwatch's fps limiter, my dxvk low latency limiter is performing much better.

[TN_fun]

I tried testing replays too, but for some reason they also can't reproduce the real thing.

If your gameplay replay codes do not show problems, then it is a network code or a network stack problem on your side.
But since there are videos on YouTube where 1 and 0.1% are also low, I think it is a network code problem

[netborg]

If your gameplay replay codes do not show problems, then it is a network code or a network stack problem on your side.
But since there are videos on YouTube where 1 and 0.1% are also low, I think it is a network code problem

The replay plays back with completely different GPU usage and thus was not suitable to gather data from for my frame pacer. I need 100% GPU usage with ~500 fps (which then gets converted to 95-99% GPU usage to minimize input lag - or less if the fps limiter is engaged).

[Tell]

I tried testing replays too, but for some reason they also can't reproduce the real thing.

If the problem happens during gameplay but not during a replay then I think this might point to high polling rate mouse/keyboard causing it since you're not moving them during the replay. You can temporarily lower the polling rate of both down to 1000Hz for testing purposes.