Vam not utilizing cpu/gpu?

Messages
30
Reactions
5
Points
8
In a demanding scene with 3 people, I'm only getting ~30fps. When I look at utilization in task manager or afterburner though, vam is only using like 30% of my cpu and 40% of my gpu. Changing the settings and using GiveMeFps only effects fps slightly and doesn't affect hardware utilization. No problems in other games. 3080ti and Ryzen 7 2700x. What would be causing this?
 
Make your afterburner show the usage of all CPU threads. If there is any at 100%, that is where your bottleneck is. That's why we generally say that, for games, single thread performance is important. VAM 2 will be optimized to better use the other CPU threads.
 
Last edited:
Nope, no threads above ~40%.
Screenshot 2023-01-29 - Copy.jpg
 
Interesting. Now I'm curious too. Check your vram usage just to be sure. If in other games you are fine, it should not be the temps. Hopefully, someone that knows will show up.
 
Last edited:
Whats your windows power settings set to, cam you screenshot your cpu in task manager
 
Whats your windows power settings set to, cam you screenshot your cpu in task manager
I changed windows power settings from "better" to "best" performance, no difference.

Interesting. Now I'm curious too. Check your vram usage just to be sure. If in other games you are fine, it should not be the temps. Hopefully, someone that knows will show up.
here's another ss with more info.
Screenshot 2023-01-29 141111 - Copy.jpg

in comparison, here's red dead 2, running at half-vsync (155hz monitor) just fine with 75% gpu utilization.
Screenshot 2023-01-29 142105 - Copy.jpg

p.s. I do get up to 80 fps in the "default" scene, but still, only using 40% gpu and 30% cpu.
 
Last edited:
Multiple threads run on each core. So the performance of each thread isn't telling the whole story. And task managers "cpu usage" average value isn't either.

You need to find a tool that displays how busy each core is. You will likely find one or two near 100%.
 
Imo 2700X is lowkey bottlenecking your 3080Ti, 2nd gen ryzen was not major IPC improvement over first gen.
VaM likes good single thread performance more than anything (something 1st and 2nd gen ryzen was lacking, if compared to 3rd/4th gen later on).
That's why intel was/is better choice (when it comes to VaM).

You can't really compare VaM to other games.
Even 6 core 3600 could theoretically perform better in VaM than 8 core 2700X (hate it or love it, It is what it is).

I don't like to use those wild "compare" sites, but anyway here it is as example (take it with grain of salt).
GPUCheck - 2700X vs 3600 vs 8700k vs 10700k vs 12700k
PassMark - 2700X vs 3600 vs 8700k vs 10700k vs 12700k

EDIT: Only thing I can think of, is to make sure vsync or any type of fps limiter or "eco" modes are inactive (it does effect cpu/gpu usage).
60lock.jpg

unlocked.jpg
 
Last edited:
Thanks for the replies everyone, didn't expect this much attention.

@nsfwthrowaway77 Core Temp can do what DJ suggested. if you figure out the reason please let us know what it was
Thanks for the Core Temp suggestion, neat program.

Multiple threads run on each core. So the performance of each thread isn't telling the whole story. And task managers "cpu usage" average value isn't either.

You need to find a tool that displays how busy each core is. You will likely find one or two near 100%.
According to Core Temp that's not the case.
Screenshot 2023-01-29 233021 - Copy.jpg (this is on the basic default vam scene)

Need to first try and get Ultimate power setting option...here is the link
Tried it but didn't notice a difference, think I'll leave it off to avoid unnecessary power usage..

Imo 2700X is lowkey bottlenecking your 3080Ti [...]
It definitely is... cpu is probably the next upgrade on the list.
Also I can't think of anything outside of vam that could be limiting fps.


My next step is to try a fresh install of vam I guess? Will update tomorrow.
 
Maybe Core Temp is only reporting the % load of the thread with the most use on that core? To be honest I don't fully understand how the use of each thread influences how busy that core gets (in terms of bottlenecks). Maybe you could try temporarily disabling threads and checking the CPU usage again, to see if any gets to 100%. (that is a long shot though)
 
Last edited:
Core Temp doesn't work well with SMT/HT (it reports avg. usage for threads).
You can have T0=80% / T1=20% and core temp is gonna report 50% load for that core.

HWiNFO can show usage and utilization for all threads (reported by task manager).
HWinfo.jpg

or go back to task manager and right click anywhere (on CPU graph) and change graph to logical proccesors to list all 16 threads.
 
Core Temp doesn't work well with SMT/HT (it reports avg. usage for threads).
You can have T0=80% / T1=20% and core temp is gonna report 50% load for that core.

HWiNFO can show usage and utilization for all threads (reported by task manager).
View attachment 206012
or go back to task manager and right click anywhere (on CPU graph) and change graph to logical proccesors to list all 16 threads.

I see. He showed a screenshot from Afterburner showing that his most active thread was around 45%. DJ suspects that this does not mean that the core is not 100% busy. I searched for a tool that could show the core load rather than the thread usage, but that proved to be not so easy. ChatGPT suggested CoreTemp, but it appears that it's not useful either. Maybe disabling threads and testing can reveal something?

@nsfwthrowaway77 Another thing, as the CPU seems to be the main suspect, could you test again on RDR without vsync and with the graphics set to minimum to see if any of your threads reach 100%? Just to make sure there is no problem with the CPU
 
Last edited:
You are in a CPU bottleneck, simple as that. That's why your GPU usage is that low, cause CPU is at max. in this game, in this scene. It's bottlenecking your GPU. Doesn't matter how the workload looks like. 3 people with maxed out physics, plus hair, plus clothings is overkill for every CPU right now. 13900K (mine) is doing around 50-60fps with it, 2 females and one male.

Remove one female person and your fps will rise to heaven and your GPU usage will increase as well.

https://hub.virtamate.com/threads/faq-frequently-asked-questions.21909/post-87125

In general: if your GPU usage is rising when you "simplify" a scene, you are in a CPU bottleneck. That goes especially for physics, hair (physics) and clothings (physics).
 
Last edited:
You are in a CPU bottleneck, simple as that. [...]
In general: if your GPU usage is rising when you "simplify" a scene, you are in a CPU bottleneck. That goes especially for physics, hair (physics) and clothings (physics).
This was definitely not the case in the test I just did. Demanding scene with three people, GPU usage hovering around 50%. Deleted one person, fps nearly doubled, but gpu usage stayed right at the 50% mark. Same story after deleting another person.

[...]
@nsfwthrowaway77 Another thing, as the CPU seems to be the main suspect, could you test again on RDR without vsync and with the graphics set to minimum to see if any of your threads reach 100%? Just to make sure there is no problem with the CPU
Here's rdr2 with some settings lowered, nearing 100% on most threads
Screenshot 2023-01-30 184634 - Copy.jpg

Of note btw, running a clean install of vam made a crazy difference for me, a 50% or more fps increase. Don't even want to start troubleshooting why that is. Unless it's just a big addonpackages folder reducing fps, which would be ridiculous.
That being said, CPU still isn't getting over 40% utilization. HWinfo showed this:
Screenshot 2023-01-30 234645 hwinfo - Copy.jpg
I'm still not convinced vam is actually using all of a single core.
 
This was definitely not the case in the test I just did. Demanding scene with three people, GPU usage hovering around 50%. Deleted one person, fps nearly doubled, but gpu usage stayed right at the 50% mark. Same story after deleting another person.

For this test, it's probably best to disable things that use more the CPU than the GPU (instead of removing the whole character). I believe soft body physics uses more the CPU. You can try disabling that and see if GPU usage increases

If adding things that use a lot of CPU (such as clothes with physics?), decreases GPU usage, that could also be a sign that you're CPU bound.
 
Last edited:
For this test, it's probably best to disable things that use more the CPU than the GPU (instead of removing the whole character). I believe soft body physics uses more the CPU. You can try disabling that and see if GPU usage increases

If adding things that use a lot of CPU (such as clothes with physics?), decreases GPU usage, that could also be a sign that you're CPU bound.
Ok yeah, good call on that. I was at 30%CPU 40%GPU with soft physics on, and 40%CPU 70%GPU with soft physics off. So I'm convinced it's a CPU bottleneck, but I'm also convinced vam isn't even using all the processing power available. Hopefully vam 2 fixes that when it releases in like 6 years.
 
You can already try an alpha and see what you'll probably get regarding cpu performance.

Like I already said: physics eat up cpu power. It's better to get to know what exactly eats up your cpu, and what your gpu. And what you can use in a scene to not be in a cpu limit to fully make use of your gpu power. Cause limiting the cpu in this game is fucking easy. Just add 3 persons with heavy physics.
 
Meshedvr should be implementing unity's DOTS + Burst compiler, so, performance should drastically increase
 
Back
Top Bottom