Stuttering/slugish feeling when FPS drops below FPS Cap G-sync

[dz520]

Hi, I just registered to the forum to reply to this thread and after writing a lengthy post it's marked as spam and doesn't allow me to post halp

P.S now after typing this one sentence it says "Forbidden. You submitted too quickly. You may try again in a few seconds. "

Uhh...

[jorimt]

Hi, I just registered to the forum to reply to this thread and after writing a lengthy post it's marked as spam and doesn't allow me to post halp

P.S now after typing this one sentence it says "Forbidden. You submitted too quickly. You may try again in a few seconds. "

Uhh...

The first few posts of a new user account must be manually moderated and approved. Try submitting your message again.

[RealNC]

Hi, I just registered to the forum to reply to this thread and after writing a lengthy post it's marked as spam and doesn't allow me to post halp

P.S now after typing this one sentence it says "Forbidden. You submitted too quickly. You may try again in a few seconds. "

Uhh...

Hi.

Users with less than 4 posts need to go through an anti-spam moderation queue. Can't be helped, unfortunately. The forum is being targeted by spammers 24/7

[dz520]

Thank you guys, I'm going to copy paste my post from earlier.

Hi, I just had to register to reply to this thread because I'm battling with the same issue in BFV.
Maybe it could help or bring some new info or perspective.
Our PC specs are similar too, I have a 3080 Ti except with a 5900X and 32gigs of RAM running at 3800 with 1:1 1900MHz FCLK.

I've been trying to iron out these microstutters/frametime spikes for the longest time, I've started to question whether this is normal or I'm just hypersensitive to this stuff.
I also reinstalled Windows yesterday and I have to say that did slightly reduce the spikes but that was to be expected since I was running an old install.

Over time I've found that the game responds the best to:
NV VSYNC: Use 3D application settings > Ingame: VSYNC ON > RTSS framecap (in my case 140 FPS for 144Hz monitor)

But I'm experiencing the same issues even though I have a lot of performance overhead.
For example, I lock FPS to 140 but the RTSS counter shows constant dips to 139 and small short spikes on the frametime graph.
The ingame FPS counter shows similarly but sometimes dipping to 138 with spikes to 141 so I'm not sure which counter is more accurate of the two.
If I uncap the FPS it is usually maintained in the 220-280 region and I never see below 200 for the remainder of the game.
Meaning there is no reason for the game to constantly dip to 139 FPS if I never see it below 200 with unlocked FPS.

Lately I've been thinking that maybe monitor is the issue, I'm playing on an ultrawide LG 34GK950-F which doesn't have a Gsync module and is in fact not even certified as a "GSync Compatible" monitor but there have never been any flickering or other issues that I'm aware of and it is generally reported that GSync compatibility works fine regardless.

I've read from other people online that swear that GSync module "feels smoother" and that Compatible monitors are hit or miss in terms of implementation.
But I don't know if I want to spend 4 digits on a monitor only to end up with the same result, native GSync ultrawide monitors are expensive (the other way would be going to higher refresh rates but that would mean a significant monitor downsizing).
And that still wouldn't fix the 139 FPS dip issue, but my reasoning is that tiny dips would be less noticeable with the actual module?
But the module is only supposed to be better at lower framerates below 48ish (freesync refresh range) and that's not the point here, uhh this is so frustrating

After all this time I've spent on trying to get to the bottom of this I'm still at a point of not knowing if this has something to do with the Ryzen architecture, the game engine itself or the monitor.
In an ideal world I would have multiple configurations and monitors at my disposal to find the best version of gameplay but oh well

[biggydeen]

After all this time I've spent on trying to get to the bottom of this I'm still at a point of not knowing if this has something to do with the Ryzen architecture, the game engine itself or the monitor.
In an ideal world I would have multiple configurations and monitors at my disposal to find the best version of gameplay but oh well

Well this looks like the exact same issue I have. The best way to test is by playing the smaller snow map called Fjell 652 (conquest). On this map, all gameplay is buttersmooth without any dips on my system (Steady 140 fps without dips). If you also expierence buttersmooth gameplay on this map without stutter then you have the same issue as me and I can explain what is happening and how to fix this.

On other maps, FPS sometimes drops below 140. At that exact moment I notice some kind of stuttering. And the best way to explain this is that the monitor is not receiving a steady pace of images to process. This is not a problem of the monitor nor g-sync. G-sync will not fix any stuttering caused by bottlenecks in the system. I thought for years g-sync would offer buttersmooth gameplay @high fps but this is simply not true. That's why I never found a solution to my problem. I've had this problem with every system I had, with completely different hardware. The reason why I never found a fix is because it was never related with g-sync or any other setting. It's simply caused by the fact that if you do not feed your monitor with a steady pace of frames, you will notice some kind of stuttering if this pace is not steady anymore.

And although the guys from BB mentioned this before, it just hit me when I watched this vid: https://www.youtube.com/watch?v=xsXFUVY ... on%29sense

Do not be fooled by the FPS you are getting when uncapped. Sometimes, fps will simply drop below 140 on bigger maps. For battlefield games this might be related to the CPU. I always have multiple cores @100% load no matter what map or FPS I get. I can only assume that cores @100% load all the time will cause hickup's somewhere. Just look up some vid's on YT with your specs, you will always see that the 1% low is below 140 eventhough avg fps is 200+

I've tested this extensivly over the past few days. And I always notice on the smaller maps in BF1 and BFV everything is buttersmooth because the frametime line is just a complete flat steady line. But when playing on the bigger maps, sometimes things get out of hand on the screen and is creating a cpu bottleneck causing fps to dip below the cap. And even just 1 frame below cap means inconsistancy in the frame pacing and that will be noticable as stutter.

You describe the exact same issue I have so i'm pretty sure it's the same problem. I've tried every setting there is and nothing will fix this. Try capping the FPS to 100 and see if that resolves the problem (it should).

I think I finally understand now that there is no fix other then lowering settings to get a higher 1% low or lowering the FPS cap at or below the 1% low. Do note that battlefield is using 100% cpu all the time and not utilizing GPU that much. So in other games, when the GPU usage is at 100% you might have a different scenario. For example, when playing control at max settings my GPU is running 99% all the time. And although my frames are inconsistant I don't notice any stuttering. So be aware that the frostbite engine is more sensitive to this due to high CPU load or something.

Let me know if your expierence the same thing.

So I was thinking with an 5800x and 3080ti I must be able to get that steady 140 fps in most games with high-ultra settings. And most of the times this is the case only in Battlefield games I instantly notice stutter when the FPS is not steady.

Maybe it's just the frostbite engine + heavy CPU load. Battlefield games are very CPU heavy and do not utilize the GPU that much. My 3080ti is sitting at 50% usage max but some CPU cores are running 100% all the time.

Part of the reason you're seeing CPU-limitation with a 3080ti is because you're only running the game at 2560x1440, and the 3080 Ti is more suited to a 4k resolution. Basically, in certain games, your resolution may be preventing your system from taking full advantage of the GPU's power.

Have you tried increasing the "Resolution Scale" above 100% in the Battlefield advanced video options? It should take some of the strain off the CPU and make the GPU work harder, potentially lowering the CPU usage and increasing the GPU usage in those scenarios.

I've set the resolution scale to 150%. Although my GPU usage went up (as expected) my CPU usage is still 100% on 2-3 cores. I'm not sure if this enhanced performance overall. It's hard to test in multiplayer games. But as mentioned above, I think the only solotuion is to capping the FPS below the 1% low or lowering graphical settings to get that steady amount of frames. I simply don't have the issue on smaller maps. So g-sync is working fine, the engine and settings are fine it's just the inconsistancy when getting less frames below the cap what is causing the stutter. For some reason, the frostbite engine does not handle this well or it's related with 100% cpu usage.

As I had this problem with completely different systems over the last years, this is the only solution that finally makes sense. And that really got me when watching this vid: https://www.youtube.com/watch?v=xsXFUVY ... on%29sense

[dz520]

Well this looks like the exact same issue I have. The best way to test is by playing the smaller snow map called Fjell 652 (conquest). On this map, all gameplay is buttersmooth without any dips on my system (Steady 140 fps without dips). If you also expierence buttersmooth gameplay on this map without stutter then you have the same issue as me and I can explain what is happening and how to fix this.

I usually play Pacific maps and I haven't played Fjell in a long time but you're right, the map is known for running better than other maps and the gameplay is indeed the smoothest and frametimes most consistent, I immediately noticed tracking enemies was a lot more satisfying.

On other maps, FPS sometimes drops below 140. At that exact moment I notice some kind of stuttering. And the best way to explain this is that the monitor is not receiving a steady pace of images to process. This is not a problem of the monitor nor g-sync. G-sync will not fix any stuttering caused by bottlenecks in the system. I thought for years g-sync would offer buttersmooth gameplay @high fps but this is simply not true. That's why I never found a solution to my problem. I've had this problem with every system I had, with completely different hardware. The reason why I never found a fix is because it was never related with g-sync or any other setting. It's simply caused by the fact that if you do not feed your monitor with a steady pace of frames, you will notice some kind of stuttering if this pace is not steady anymore.

And although the guys from BB mentioned this before, it just hit me when I watched this vid:

There is definitely a difference in using Gsync but when we're talking about these microdips then yeah becomes very hard to chase that final step going from 9/10 to 10/10 smoothness.
I'm familiar with Battle(non)sense, I've watched many of his videos in pursuit of the best settings.
But in that video he simply talks about the "lowest FPS you can maintain" and that's where our issues stem from, it's not the FPS we can maintain, it's the absolute lowest microdips we don't notice with a basic frame counter when it constantly shows over 200 FPS.

Do not be fooled by the FPS you are getting when uncapped. Sometimes, fps will simply drop below 140 on bigger maps. For battlefield games this might be related to the CPU. I always have multiple cores @100% load no matter what map or FPS I get. I can only assume that cores @100% load all the time will cause hickup's somewhere. Just look up some vid's on YT with your specs, you will always see that the 1% low is below 140 eventhough avg fps is 200+

I've tested this extensivly over the past few days. And I always notice on the smaller maps in BF1 and BFV everything is buttersmooth because the frametime line is just a complete flat steady line. But when playing on the bigger maps, sometimes things get out of hand on the screen and is creating a cpu bottleneck causing fps to dip below the cap. And even just 1 frame below cap means inconsistancy in the frame pacing and that will be noticable as stutter.

You describe the exact same issue I have so i'm pretty sure it's the same problem. I've tried every setting there is and nothing will fix this. Try capping the FPS to 100 and see if that resolves the problem (it should).

It looks like this seems to be the case.
I just jumped across a few different maps lowering the cap down to 120 or 100 and this removes about 80%+ of the spikes in the frametime graph but at the same time this doesn't make the experience better because now I've removed about 30% (40 FPS) of the frames from my monitor and that causes noticeably less smooth gameplay overall.
It might be more consistent from the framtime perspective but it's also less enjoyable as a whole.

I think I finally understand now that there is no fix other then lowering settings to get a higher 1% low or lowering the FPS cap at or below the 1% low. Do note that battlefield is using 100% cpu all the time and not utilizing GPU that much. So in other games, when the GPU usage is at 100% you might have a different scenario. For example, when playing control at max settings my GPU is running 99% all the time. And although my frames are inconsistent I don't notice any stuttering. So be aware that the frostbite engine is more sensitive to this due to high CPU load or something.

Let me know if your experience the same thing.

The only thing different in our situations is some of your CPU cores hit 100% usage which is absolutely not the case with my system.
I just checked on multiple maps and I've never seen a single thread hit over 80% usage ever, most of the time they're in the 30-70% range and there's about 14 threads loaded during gameplay with rest in single digits.

But I'd say the fix is in better hardware, maybe not now but down the road. New Intel and AMD CPUs are coming in a few months.
If 1% and 0.1% lows are what's causing this issue then obviously a better CPU or RAM would boost those lows.
Overclocking RAM definitely makes a difference on Ryzen systems, I've noticed some small improvements with higher RAM + Fabric frequency and tighter timings which I'm still in the process of tuning since it's really time consuming for a RAM newbie like me.
I wish I had a 10900K/11900K system to try out, even though they might get slightly lower average FPS, I'm curious how they do with the lows.
It's very difficult to find this info online, the big reviewers can't do multiplayer test for obvious reason of pure randomness of action and people that post gameplay on YT all have very different specs/settings etc.

Aside from that I don't know what else would make a difference, if going 240Hz+ would bring any improvement.
Yes the dips would still be there but there would be more overall smoothness, just like the frame cap at 100 reduced spikes but decreased smoothness maybe the opposite happens at higher fps - bigger spikes but more buttery overall, a compromise.
I've read that higher Hz "break up" the tearing into smaller sections so they're easier on the eyes, but I'm not sure if this is true, maybe the monitor experts could chime in on this one.

[jorimt]

There is definitely a difference in using Gsync but when we're talking about these microdips then yeah becomes very hard to chase that final step going from 9/10 to 10/10 smoothness.

G-SYNC can sometimes take slightly longer to recover from frametime spikes when compared to standalone V-SYNC off, as it must wait for the next scanout to prevent tearing, whereas standalone V-SYNC off can recover more quickly due to the fact that it can interrupt the current frame with the next in the same scanout. This is obviously called "tearing" however.

Regardless, either way, with a frametime spike you will get stutter, period. Pick your poison; G-SYNC for a slightly longer tear-free stutter, or standalone V-SYNC off for a slightly shorter stutter with a tear.

I should also mention both G-SYNC and standalone V-SYNC off at the same FPS within the refresh rate can have the same amount of stutter during a frametime spike. It just depends at what completion level the current scanout cycle is in when the spike occurs.

But I'd say the fix is in better hardware, maybe not now but down the road. New Intel and AMD CPUs are coming in a few months.
If 1% and 0.1% lows are what's causing this issue then obviously a better CPU or RAM would boost those lows.

Better hardware can absolutely reduce frametime spikes, but it can't eliminate them.

Sans spec limitations and the various other issues that can increase frametime spike frequency and duration, It's virtually guaranteed with current PC architecture paired with the open-world nature of most modern games (where they opt to stream instead of preload most of the data, the latter of which would instead result in periodic loading screens as you navigated to a new chunk of the given map; i.e. Half Life 2, etc) that there will always be the occasional frame that takes longer than a single refresh cycle to render, which will create a frametime spike, regardless of system specs, settings, optimizations, etc.

Heck, even PS5, a closed platform which prioritizes SSD bandwidth to decrease asset load time (which, in turn, reduces frametime spikes; something that PC doesn't even take advantage of yet in 99% of games) still experiences them, such as in Ratchet & Clank during portal transitions:

phpBB [video]

There's sometimes simply no getting around more complex scenes having to load in.

So while, sure, maybe you can play Battlefield V on future hardware 10 years from now frametime spike-free, there's likely going to be a current game on current hardware that won't be.

I've read that higher Hz "break up" the tearing into smaller sections so they're easier on the eyes, but I'm not sure if this is true, maybe the monitor experts could chime in on this one.

This is true. See:
https://blurbusters.com/gsync/gsync101- ... ettings/9/

However, the irony is, the higher the refresh rate, the closer G-SYNC and standalone V-SYNC off become in input lag as well.

It's all about the scanout speed; the faster it is, the less time there is for tearing. At 1000Hz, syncing method will become all but unnecessary. G-SYNC is merely a stopgap.

[biggydeen]

I usually play Pacific maps and I haven't played Fjell in a long time but you're right, the map is known for running better than other maps and the gameplay is indeed the smoothest and frametimes most consistent, I immediately noticed tracking enemies was a lot more satisfying.

I also played the pacific maps alot. I did not notice much frametime spikes there. The gameplay was overall pretty smooth. But Fjell is buttersmooth 100% of the time. Same thing with the smaller maps in BF1 and BFV.

There is definitely a difference in using Gsync but when we're talking about these microdips then yeah becomes very hard to chase that final step going from 9/10 to 10/10 smoothness.
I'm familiar with Battle(non)sense, I've watched many of his videos in pursuit of the best settings.
But in that video he simply talks about the "lowest FPS you can maintain" and that's where our issues stem from, it's not the FPS we can maintain, it's the absolute lowest microdips we don't notice with a basic frame counter when it constantly shows over 200 FPS.

It's indeed all about that 1% lows. I wish I had known this before. I bought the 5800x because it was the best gaming CPU overall when comparing max and avg FPS but I never looked at the 1% lows. But 1% lows are also game engine related. You can have the best system in the world but if the game engine can't handle it you will always see spikes.

But overal the 5800x or 5900x is the best CPU you can buy right now. There is currently no system in the world which does have 1% lows in BF V above 140 fps. It all depends on how many times 1% low's occur below 140 fps. On smaller maps they never occur and that's why the gameplay is buttersmooth 100% of the time.

The only thing different in our situations is some of your CPU cores hit 100% usage which is absolutely not the case with my system.
I just checked on multiple maps and I've never seen a single thread hit over 80% usage ever, most of the time they're in the 30-70% range and there's about 14 threads loaded during gameplay with rest in single digits.

But I'd say the fix is in better hardware, maybe not now but down the road. New Intel and AMD CPUs are coming in a few months.
If 1% and 0.1% lows are what's causing this issue then obviously a better CPU or RAM would boost those lows.
Overclocking RAM definitely makes a difference on Ryzen systems, I've noticed some small improvements with higher RAM + Fabric frequency and tighter timings which I'm still in the process of tuning since it's really time consuming for a RAM newbie like me.
I wish I had a 10900K/11900K system to try out, even though they might get slightly lower average FPS, I'm curious how they do with the lows.
It's very difficult to find this info online, the big reviewers can't do multiplayer test for obvious reason of pure randomness of action and people that post gameplay on YT all have very different specs/settings etc.

Aside from that I don't know what else would make a difference, if going 240Hz+ would bring any improvement.
Yes the dips would still be there but there would be more overall smoothness, just like the frame cap at 100 reduced spikes but decreased smoothness maybe the opposite happens at higher fps - bigger spikes but more buttery overall, a compromise.
I've read that higher Hz "break up" the tearing into smaller sections so they're easier on the eyes, but I'm not sure if this is true, maybe the monitor experts could chime in on this one.

That is quite strange because our cpu's are not that differtent. Although BF runs best on multiple cores, some cores will always have the biggest load. Unless you have OC'ed your 5900x, I cant explain why you never hit 100% load and I do almost every time.

OCing ram can make a difference but depends on the resolution you play. Im playing at 1440p. At this resolution it does not really make any difference what mem or FCLK you use in gaming. At 1080p it can have a big impact.

240hz will prob not make much difference because you will still have those frametime spikes. Although 240hz is more smooth compared to 140hz. I'm pretty sure you would rather have 100% smooth gameplay all the time rather then very smooth play with dips all the time resulting in stutter. At least that's what I want.

I'm pretty sure the only thing that will really make a difference here is capping the FPS below 1% low. There is no other way to reduce the stutter. Other then optimizing Windows and your system as a whole as much as possible. Using core park, setting BF V to high priority, reducing background apps and all that kind of stuff.

I've tested multiple games now and the best result is always when the frametime line is a flat line with zero dips. Even just 1 fps below cap causes a noticable degrade in performance (stutter) which is instantly noticable. Although not all games have this behaviour. It seems like shooters will suffer from this the most.

[dz520]

My 5900X runs at 4625MHz CCX1 and 4500MHz CCX2, as you can see those are pretty standard numbers and I don't believe a few percent clock speed up or down would make such a difference when we're talking about tens of percents of utilization.
Maybe the system likes to have more threads to spread the load evenly but even 16 threads like in your case should be more than enough for this, so it is a bit strange.

I'm pretty sure OCing the RAM is always beneficial in CPU-bound scenarios which this definitely is, especially with Ryzen architecture.
Doesn't matter the resolution you're at, as long as you're not GPU bottlenecked (which we aren't) then it's the CPU.
I'm actually on ultrawide 1440p (34% higher pixel count than standard 1440p) and trust me there's mild but noticeable difference with faster RAM since the GPU is only at around 50% usage, it means the CPU is the bottleneck and we know Ryzens are primarily bottlenecked by RAM/Fabric speed and latency rather than core clocks.
These CPUs might be the best in gaming overall (most reviews only focus on average framerates), but they all behave differently with different games and engines.
Look at the benchmark in the attachment, Rocket Lake CPUs lose to top end Ryzens in average but score higher in 99% percentile.
Which makes me think something like a 11900K might've been a better choice for BFV despite losing in other games or tasks.

If you have that 3200C14 memory running at stock XMP profile, sounds like Samsung B-die at those specs, I managed to shave 20% latency from the same 3200C14 spec.
You could run 3800 CL16/1900 Infinity Fabric if your CPU allows it, or worst case scenario 3600 RAM/1800 IF with help of Ryzen DRAM Calculator. (Note: tuning memory can be tricky sometimes)
You can also easily overclock the CPU with CTR 2.1 tool with just a few clicks, my 5900X runs terribly slow with out of the box settings.

This of course won't completely solve the issue, but it can reduce it.

[biggydeen]

My 5900X runs at 4625MHz CCX1 and 4500MHz CCX2, as you can see those are pretty standard numbers and I don't believe a few percent clock speed up or down would make such a difference when we're talking about tens of percents of utilization.
Maybe the system likes to have more threads to spread the load evenly but even 16 threads like in your case should be more than enough for this, so it is a bit strange.

I'm pretty sure OCing the RAM is always beneficial in CPU-bound scenarios which this definitely is, especially with Ryzen architecture.
Doesn't matter the resolution you're at, as long as you're not GPU bottlenecked (which we aren't) then it's the CPU.
I'm actually on ultrawide 1440p (34% higher pixel count than standard 1440p) and trust me there's mild but noticeable difference with faster RAM since the GPU is only at around 50% usage, it means the CPU is the bottleneck and we know Ryzens are primarily bottlenecked by RAM/Fabric speed and latency rather than core clocks.
These CPUs might be the best in gaming overall (most reviews only focus on average framerates), but they all behave differently with different games and engines.
Look at the benchmark in the attachment, Rocket Lake CPUs lose to top end Ryzens in average but score higher in 99% percentile.
Which makes me think something like a 11900K might've been a better choice for BFV despite losing in other games or tasks.

If you have that 3200C14 memory running at stock XMP profile, sounds like Samsung B-die at those specs, I managed to shave 20% latency from the same 3200C14 spec.
You could run 3800 CL16/1900 Infinity Fabric if your CPU allows it, or worst case scenario 3600 RAM/1800 IF with help of Ryzen DRAM Calculator. (Note: tuning memory can be tricky sometimes)
You can also easily overclock the CPU with CTR 2.1 tool with just a few clicks, my 5900X runs terribly slow with out of the box settings.

This of course won't completely solve the issue, but it can reduce it.

Yea you might be right about the RAM when you are CPU bottlenecked. This will result in higher 1% lows. So I should upgrade or OC to atleast 3600. I've got Gskill xflare 3200mhz cl14 which are B-die so 3600 should be no problem. But I also have read somewhere on reddit that one guy managed to increase his 1% lows by 20-40 fps in some games by just running 3800 vs 3600. So i'm thinking if I should go for 3800 cl14 (I doubt my OC will go from 3200 cl14 to 3800 cl16 but can try).

I used PBO2 to undervolt the 5800x + some manual 100 mhz offset.

Yes, well at least now I know that 1% lows are much more important then avg FPS if you want that buttersmooth gameplay. I've been tackling this problem for years and years and now I finally found the missing piece.