WSAD movement causes irregular framepacing

Mouse camera movement is unaffected <-- By this I mean looking around with the mouse while standing still there is no stutter.

Specs:
Win10 21H1
i7 4790k
RTX 3080
16GB RAM

 

It is not related to a specific game world, location or POI.

All my games are installed to SSD. Please try to reproduce the problem for yourself, don't go "customer support mode" on me as if the issue is on my side because it isn't. I am a trained, experienced IT engineer with a well configured gaming PC without any bloatware that could be causing problems like this. I do not have this issue with other games in my Steam library, running on the same system.

Using RTSS to cap the fps below what the GPU is capable of like you're supposed to do when using g-sync in order to smooth out frame pacing, in 7D2D currently does not have the desired effect of smoothing out the frame pacing while moving with WSAD.

That's because the game itself is causing the irregular framepacing, the PC hardware isn't struggling at all.

(not at max cpu / gpu load)

(not at max cpu / gpu load) <-- By that I mean RTSS shows the system isn't at max load even when the fps is uncapped as shown in the screenshots. When I cap the fps with RTSS the load obviously drops further.

Game version: 20.5 b2

Please read the pinned post on how to write bug reports.

We will be needing your output log posted in pastebin.

 

Blog for A21 has info that there are some optimizations coming... But god knows what its gonna be.

 

Anyway, can't you just disable g-sync in monitor config?

 

 

 

14 hours ago, Diaboliko said:

Anyway, can't you just disable g-sync in monitor config?

G-sync is not relevant. This issue occurs outside of g-sync mode as well. On a technical level there is no reason why it wouldn't.

 

19 hours ago, chikorina said:

We will be needing your output log posted in pastebin.

You will find that if you actually test properly the problem occurs on all hardware and software configurations, because it is caused by how the game's code handles character movement. I don't remember exactly which update patch screwed it up, but the problem was introduced either with A20 itself or one of the patches after A20. It was fine when I played before A20.

 

Even though you don't actually need them, I understand you have a procedure to follow so here's all my logs anyway:
output_log_client__2022-05-17__04-44-48.txt: https://pastebin.com/1jLvNJjD
output_log_client__2022-05-17__06-55-48.txt: https://pastebin.com/d6R352cc
output_log_client__2022-05-17__06-57-49.txt: https://pastebin.com/QNxV0XHd

These are all the log files I have.

I uninstalled then reinstalled the game for testing before creating the bug report. I keep EAC disabled. I tested in singleplayer to avoid server tickrate or connection issues affecting the result.

Quote

You will find that if you actually test properly...

Pretend I didn't use the word "actually" there. English isn't my first language.

7 minutes ago, Roenie said:

because it is caused by how the game's code handles character movement.

bold assumption there but the point is it's the game itself.

> G-sync is not relevant. This issue occurs outside of g-sync mode as well. On a technical level there is no reason

Wont fix Issue of "microstutter"? How come..?

Why would it?

This really isn't the time and place for you to learn about g-sync. It's completely off-topic. I regret even mentioning g-sync in the bug report, becaues all that does is act as a distraction for people who don't fully understand.

 

Performance and optimization will be and is being worked on, but I don't see this at all. Not even turning on/off Vsync. 

 

Isn't the i7 4790k about 10 or 11 years old now? That is your culprit, more than likely.

 

Not a bug but will put it in WIP since optimizations will be done.

Interesting.

 

The CPU is not the culprit though, that's a silly assumption, and a little too easy. Couple of problems with that theory. CPU load went up from 28% to 32% in my screenshots, between stationary and walking. The CPU is basically just twiddling its thumbs.

 

This is the only game that stutters on my system when walking.  For example Cyberpunk does not stutter on the same CPU. If there is one CPU that aged incredibly well, it's the 4790k.

 

I'm puzzled as to what the cause is at this point.

But it is repeatable on my system. Clean boot, nothing else running and all that. Distinction between standing and walking is still very very clear and the stutter when walking is severe.

If there is one CPU that deserves a medal for its outstanding future proofness it is the 4790k. It has aged incredibly well and it's faster clockspeed than most current CPUs that people use for gaming today.

Most CPUs are clocked below 4Ghz and games still benefit more from single threaded performance. We're getting to a point where more than 4 physical cores (8 threads) would be beneficial for other tasks and *some* games but the 4790k is a beast. There is a reason I haven't overclocked it yet: there hasn't been any need to.

(4ghz base clock, 4.4ghz turbo)

Is it possible that the game has issues with HyperThreading?

Back when HyperThreading was a new concept, some games had timing issues. My system (not the GPU) is 7 years old and I've kept HT enabled for all that time without any issues. So it would be quite the throwback to old times if this is HT related. I'll test without HT later.

Untuitively, I don't think it's HT, it's too far fetched.

Can you tell me more about what makes this game unique in how it streams the world into existence when you move?

 

Can you give me specifics about how character movement is handled?

Any differences with that or challenges compared to most games?

  

23 hours ago, Roenie said:

This really isn't the time and place for you to learn about g-sync. It's completely off-topic. I regret even mentioning g-sync in the bug report, becaues all that does is act as a distraction for people who don't fully understand.

Rude.

 

I thought microstutter meant something like screen-tearing because of miss-synchronization of frames of which I can find on google. At least I can find something roughly looking like it. But if microstutter is about missing a frame cycle or something like that because of performance low spikes, then "oh well, I wasted your time to learn what did you mean".

 

> CPU load went up from 28% to 32% in my screenshots, between stationary and walking. The CPU is basically just twiddling its thumbs.

Why do you expect 7 days to die to utilize all available cores?

You've got 4 cores CPU with hyperthreading; how many games are capable of utilizing at least 4 cores in parallel? Not to mention 8... CPU model architecture can also mean plenty to things like cache hit-rate which used to change big steps dozen years ago... It moves a lot slower now I believe? Though, you never know until you try some other CPU model.

I also wonder if it can have something to do with it missing some of newer instruction sets.

Never had to poke that low, but always was curious on how to debug it in runtime... Oh well, Im unwilling to do such research.

 

Also, I believe HT means nothing to operating system since it views every core as logical unit. So its highly unlikely that issue is there, but you can always test it out by manually setting core affinity. 

That would be awesome to have some performance tuning thread with dev attention though... I also have joke utilization while having huge average framerate drops in, say, big cities.

2 minutes ago, Diaboliko said:

Rude.

Sorry to be so blunt, it's obvious you haven't much of a clue and I would've had to write an essay on g-sync to explain why you were off base. Doing so would've made the thread go off topic, possibly leading to the issue never getting fixed.

 

8 minutes ago, Diaboliko said:

I thought microstutter meant something like screen-tearing

The thought is a good one but in this case it is incorrect, because I know the difference! I chose the wording carefully. Microstutter is microstutter, tearing is tearing. Tearing can appear as microstutter but I can tell the difference.

 

10 minutes ago, Diaboliko said:

But if microstutter is about missing a frame cycle or something

Incorrect again. It's about a bunch of frames taking 10ms each to be produced, then one taking 15 to 30ms, then getting another bunch of frames taking 10ms each, then another one that takes 20ms, and so on. Good framepacing means the average isn't strayed from by more than 1-2ms. In other words at 100fps, you would get 1000 / 100 = 10ms per frame, varying between 9 and 11ms.

 

14 minutes ago, Diaboliko said:

Why do you expect 7 days to die to utilize all available cores?

Jugginator's claim that my CPU is too old falls flat by me pointing out that CPU utilization is very low. You can see the load for each core in my screenshots. It's not even using 50% of any one of them, so I could swap my CPU for either a nuclear powered one or a proverbial potato and it would make no difference.

Well *one* core was at 50-55% from the looks of it. All other cores low. It's not that I expect 7D2D to use all cores, but if it needed more CPU juice then the load would've been near max on at least 4 threads.

20 minutes ago, Diaboliko said:

manually setting core affinity

I guess it can't hurt to do some experimenting. Might try forcing it onto a single physical core to see what happens.

I can also disable the HT feature entirely in the UEFI (bios).

6 minutes ago, Roenie said:

Sorry to be so blunt, it's obvious you haven't much of a clue

About g-sync I mean. It does not affect the game itself, only acts on the output side, the monitor adjusts its refresh rate to only display completed frames when they are ready, not before.

So if anything that reduces perceived stutter by eliminating tearing. Disabling it makes matters worse if anything.

(I did test with it off too, just in case.)

16 minutes ago, Roenie said:

Jugginator's claim that my CPU is too old falls flat by me pointing out that CPU utilization is very low

 

No I did not claim that by pointing out that CPU utilization is very low, don't misquote me. 

 

You're missing one fact about CPU's: architecture matters too. Two different CPU's running at 4.0 are not identical because they're both running at 4.0. Cyberpunk cares much more about the GPU than the CPU, so more processing power can go elsewhere, this game is a bit less unoptimized but regardless it heavily relies on the CPU.

 

The test I took a little bit ago is on:

Ryzen 7 5800X

32GB 3200 RAM with tightened timings

ASRock B550 Steel Legend

 

I also have another rig that's using a Ryzen 3 3200g with 16GB of tightened timings ram that uses either a 1650 or the on board GPU.

 

My partner that I play with a bit/live with I use her rig sometimes for testing, and she has a 1st gen Ryzen (can't remember it off the top of my head). None of them have this issue. The 3200g is the only CPU that isn't hyperthreaded. 

 

So, in summary, you either have a hardware or configuration issue that's causing this.

 

You're also making this borderline toxic, if that continues, I will lock this thread.

 

 

Just as what Diablo said, this game among most if not all games now a days favor 4 cores and will send child tasks to other cores. As you see here.

Edited May 19, 2022 by Jugginator (see edit history)

Just now, Jugginator said:

No I did not claim that by pointing out that CPU utilization is very low

I did not write that, you misread. You missed the word "me".

I wrote: by ME pointing out that CPU utilization is very low

😄

I pointed out your CPU was old because it's over 10 years old. I did not say that because you pointed out that your CPU utilization is low. Misquoted.

This report has become a little toxic and bloated with information not related to the issue; the issue is more suited for the general support section for additional help. Issue has been noted. Feel free to open a post in General Support, where performance has been talked about quite a bit.

 

Locking.

Edited May 19, 2022 by Jugginator (see edit history)