Whats your average FPS in VAM? 2 Person atom 1 CUA and your computer spec?

[KameronFox]

Mine is like 28fps with couple of vars. This is so bad...

My spec: RTX2080TI, Ryzen 9 5950x, 64gb 3200 ram. PBO enabled.

 

[PhantasyStarfan]

Usually just use 1 model with 2 -3 lights, sometimes 2. It sits at roughly the same. 20-30.

If introduce more characters like male, their appearance is less important so turn stuff like hair and clothe physics off along with soft physics, seems to help abit.

 

[KameronFox]

Usually just use 1 model with 2 -3 lights, sometimes 2. It sits at roughly the same. 20-30.

If introduce more characters like male, their appearance is less important so turn stuff like hair and clothe physics off along with soft physics, seems to help abit.

Whats your computer spec?

 

[PhantasyStarfan]

Whats your computer spec?

AMD Ryzen 5 2600X, 16gb DDR3, GTX 1660 Super

 

[DarkFantasy]

Mine is like 28fps with couple of vars. This is so bad...

My spec: RTX2080TI, Ryzen 9 5950x, 64gb 3200 ram. PBO enabled.

It depends, in more complex scenes mid 20's. With very simple scenes around 50-60 or higher. I have Render Scale set to 1.25-1.5 depending on the scene and also change the "Physics Update Cap" depending upon the FPS I get.

Lights make a huge difference, I try to stick with two or modify scenes to go with two and also make it darker.
Other quick fix things to look at:
Cloths - jewelry items with Sim on steal FPS (like necklaces); Also look at turning physiscs down
Hair Physics - Curve Density and Hair Multiplier
Mirrored Surfaces - if enabled and there are mirrored surfaces there will be a huge impact to FPS
Plugins - try turning plugins off one at a time and wait a 30 sec to see if that impacts FPS
Collisions - If there are bad collisions it seems to impact FPS as well

Personally, I find more realistic scenes that help with immersion are ok between 20-30 FPS if I tweak the speed and the lights.

Can I ask how you are loading VAM? How you load it also makes a difference. For example, if you are going through Steam rather than just Virtual Desktop you will lose a lot of FPS.

Specs: I9900kf; rtx3070, 64G ram, M.2 NVMe PCIe drive

 

[KameronFox]

Can I ask how you are loading VAM? How you load it also makes a difference. For example, if you are going through Steam rather than just Virtual Desktop you will lose a lot of FPS.

Specs: I9900kf; rtx3070, 64G ram, M.2 NVMe PCIe drive

Oh really? I use steam because of AMD FidelityFX SuperResolution for SteamVR, It does some fps improvements. But how so steam decreasing fps?

 

[DarkFantasy]

Oh really? I use steam because of AMD FidelityFX SuperResolution for SteamVR, It does some fps improvements. But how so steam decreasing fps?

Instead of launching Virtual Desktop, then Steam VR, then opening VAM in Steam it seems to work much much better if you just create a Bat file on your desktop and execute it inside Virtual Desktop by double clicking it using your controller.
The batch file should contain:
"C:\Program Files\Virtual Desktop Streamer\VirtualDesktop.Streamer.exe" "F:\Vam\VaM.exe"
Just modify the path to match your configuration.

Here is a thread with more details and discussion:

Increased FPS when using openVR

So a friend who introduced me to VaM just told me about a new open source free plugin that dramatically increased your FPS by reducing the render resolution and interpolating back to full resolution. It's been written by AMD and is apparently really good, giving some impressive performance...

hub.virtamate.com

If you are already doing this then I guess it is down to tweaking the scene.

The other thing to look at that are the custom assets. Some of them are heavy and really impact FPS. Also if there are too many it impacts FPS. I usually go through and turn each off one at a time to see which is causing the drop if there is one.

 

[rabbit1qaz]

Its useless to say whats peopl FPS is. Couse every scene might get completel diffrent. To be most acurate you should ask for default scene FPS , given player settings: 1 physics cap, 70hz ,... and so on.

Eg: if i get default scene non vr i get about 150-300FPS. but as i change hair to long, set it to 64 curve and 64 density it goes to 60-90 with just one person and simple clothing (10900K, 32GB4200 Ram, 3090, Fast SSD)

 

[KameronFox]

Instead of launching Virtual Desktop, then Steam VR, then opening VAM in Steam it seems to work much much better if you just create a Bat file on your desktop and execute it inside Virtual Desktop by double clicking it using your controller.
The batch file should contain:
"C:\Program Files\Virtual Desktop Streamer\VirtualDesktop.Streamer.exe" "F:\Vam\VaM.exe"
Just modify the path to match your configuration.

Here is a thread with more details and discussion:

Increased FPS when using openVR

So a friend who introduced me to VaM just told me about a new open source free plugin that dramatically increased your FPS by reducing the render resolution and interpolating back to full resolution. It's been written by AMD and is apparently really good, giving some impressive performance...

hub.virtamate.com

If you are already doing this then I guess it is down to tweaking the scene.

The other thing to look at that are the custom assets. Some of them are heavy and really impact FPS. Also if there are too many it impacts FPS. I usually go through and turn each off one at a time to see which is causing the drop if there is one.

What do you mean by virtual desktop? I run VaM (OpenVR).bat since that AMD Fidelity.

 

[DarkFantasy]

What do you mean by virtual desktop? I run VaM (OpenVR).bat since that AMD Fidelity.

You need someone smarter than me then. I don't know and don't use that method nor the Fidelity DLL. (I am paranoid when it comes to replacing DLL's with third party DLL's!)

 

[HolySchmidt]

Its useless to say whats peopl FPS is. Couse every scene might get completel diffrent. To be most acurate you should ask for default scene FPS , given player settings: 1 physics cap, 70hz ,... and so on.

Exactly.

If you want to compare hardware used with VaM, everyone posting their FPS has to ...
- use the same quality settings
- use the exact same scene

Everything else is kind of useless. I don't know (who knows, anyone??) when VaM 2.0 comes out but hopefully it won't take too long and after that all of this is obsolete anyway.
If you still wanna compare hardware then someone has to create a benchmark scene.

This should contain:
- different amount of lights
- different count of persons
- maybe all combined with an animation (cause of physics)

Physics alone is creating an undefined amount of variations in a scene depending on what is in motion and colliding with something. And so on and on and on ...

 

[TToby]

As HolySchmidt says... as long as we have no "VaM Benchmark Scene", you can't compare FPS.

As I am relatively sensitive for low FPS below 60 FPS, I try to allways stay above this.
I have my Valve Index at 144Hz, so that I most of the time will end with half that Hz in FPS.
As for me content is more important than lighting, I stay most of the times with only one light.
With manipulating intensity, range and the "scene lighting", you can IMHO get some very good looking scenes, too.
Multiple light sources for rendering are a HUGE factor for low FPS, much (!) more than several CUAs (without baked in lighting) at one time (i've tried this).
In addition to this, we have to speak about how you will look at the scene or if there are animations or other movement.
When only looking straight on at a scene, without moving my head too much, I have much more FPS than if the figures or my head are moving.

Answering the question:
a) I had ca 60-120 frames while looking at a scene with no movement with my old 4790K / gtx 980ti, with two persons plus full clothes and hair, 1 CUA (how big???), one light, supersampling at 100% and other settings to mid-low... but around 30 frames when moving one of that person around like crazy....

b) With my actual 5800x / 6900xt, I have ca 70-130 frames with no movement and the same scene, but with all settings on highest and SS at 1,5 to 2 / 200%.
When moving around one person like a lunatic, I will end at around 30 fps, too.
Turning down all graphic settings like with my old PC, I will get only a few FPS more.
This shows IMHO relatively good, how bad VaM is at utilizing the full potential of your PC. You will still have bad FPS, but therefore you can maxx out the graphical settings with a better GPU.

Other than only looking at FPS, there are some more things like latency and stuttering.... sometimes I have good FPS, but while turning my head, the picture is hanging for a very short moment and each time making my stomache rotate a bit. When looking at a scene and having the menu open, the FPS drops plus I have some stuttering, too, or even more while the file browser is loading... Creating a hairstyle for 30 minutes with the menue open and constantly rotating my head between UI and figure, is hell. Not as bad as running up the 732 steps to High Hrothgar in SkyrimVR, but bad enough to be unpleasant ;-)

 

[Jiraiya]

What do you mean by virtual desktop? I run VaM (OpenVR).bat since that AMD Fidelity.

Virtual Desktop is a way to get wireless VR from the PC to the Oculus Quest and Quest 2 headsets. It is ONLY if you have those headsets.
If you are running VD you do NOT need to involve SteamVR. In fact, that slows things down and makes the experience worse than without it.
So in the case that you use a Quest (1/2) headset and have Virtual Desktop it is always best to run VaM directly using the batch files repeated here.
You launch VD from your headset, and from within Virtual Desktop can then run the batch file. That launches VaM without SteamVR at all and gives a faster smoother experience.

 

[KameronFox]

Virtual Desktop is a way to get wireless VR from the PC to the Oculus Quest and Quest 2 headsets. It is ONLY if you have those headsets.
If you are running VD you do NOT need to involve SteamVR. In fact, that slows things down and makes the experience worse than without it.
So in the case that you use a Quest (1/2) headset and have Virtual Desktop it is always best to run VaM directly using the batch files repeated here.
You launch VD from your headset, and from within Virtual Desktop can then run the batch file. That launches VaM without SteamVR at all and gives a faster smoother experience.

I've oculus rift s, Running VaM (OpenVR).bat because of that AMD Fidelity will work on steamvr. Does that mean steamvr making my game fps worse?

 

[Jiraiya]

I've oculus rift s, Running VaM (OpenVR).bat because of that AMD Fidelity will work on steamvr. Does that mean steamvr making my game fps worse?

No. Because you are not running a Quest or Quest2 with Virtual Desktop.

I will say it again, clearly. VD is ONLY for those running Quest or Quest2 headsets.
The reason it's worse is this.
VaM->SteamVR->VirtualDesktop->Headset
which is clearly worse than
VaM->VirtualDesktop->Headset

You have VaM->SteamVR->RiftHeadset.

Does that make it clear?

 

[KameronFox]

No. Because you are not running a Quest or Quest2 with Virtual Desktop.

I will say it again, clearly. VD is ONLY for those running Quest or Quest2 headsets.
The reason it's worse is this.
VaM->SteamVR->VirtualDesktop->Headset
which is clearly worse than
VaM->VirtualDesktop->Headset

You have VaM->SteamVR->RiftHeadset.

Does that make it clear?

Okay.

 

[Onionpeeler69]

not enuff

I get 20-30 fps on 1080

It sucs ballz

 

[Jiraiya]

At this point I think we need to pick a free reference scene and base settings to compare.
The numbers are kinda meaningless otherwise.

 

[TToby]

I think we need to pick a free reference scene

I think this should be done by someone who understands how VaM is exactly using ressources and which things and actions are the real performance killers in every day use. In addition to this the scene should be well balanced in a way, that we will get numbers that are not meaningless, or numbers that are more or less similar from one PC to another, because they are too close together. But getting fancy numbers that have no real-life value because of a scene that a sane brain would never ever use, is meaningless, too. I think picking a reference scene sounds easier than it is... and certainly AFTER doing this, there will be many user who claims to know it better.

Though, every randomly picked, self running scene (ideally with no references and easy to use, maybe one of the old build-in ones?), would be way better than comparing numbers that simply can't be compared.

 

[KameronFox]

I think this should be done by someone who understands how VaM is exactly using ressources and which things and actions are the real performance killers in every day use. In addition to this the scene should be well balanced in a way, that we will get numbers that are not meaningless, or numbers that are more or less similar from one PC to another, because they are too close together. But getting fancy numbers that have no real-life value because of a scene that a sane brain would never ever use, is meaningless, too. I think picking a reference scene sounds easier than it is... and certainly AFTER doing this, there will be many user who claims to know it better.

Though, every randomly picked, self running scene (ideally with no references and easy to use, maybe one of the old build-in ones?), would be way better than comparing numbers that simply can't be compared.

That's not even necessary. We just have to build heavy scene and test it with our PCs with different settings then compare it.

 

[MacGruber]

I'm putting out a 50-100$ commission for a "Benchmark Scene". The idea is a dedicated free scene everyone can use to measure VaM performance, enforcing identical settings via plugin, so results are actually comparable. Details here:

50-100$ commission for "Benchmark Scene"

As a side project, I was looking at PC components for upgrading my old machine. That made me browse through various forum threads on the Hub discussing FPS in VaM. Everyone is posting random FPS values, but barely anybody shares which exact settings and scene was used. But those differences can...

hub.virtamate.com

 

[TToby]

@KameronFox In this case I disagree with you. Please maybe read the text of MacGrubers Benchmark Scene comission. He has put it to words much better than I have done it, writing those lines being at work.

@MacGruber I am very happy that you have accept the challenge! From just reading your commission I am sure this will be professionally done and of course successful. To be honest, when writing "this should be done by someone who understands how VaM works...", I was secretly pointing at you, regarding the other thread we were discussing this.

I have three little conflicting concerns, I am not fully able to wrap my head around:
- If there would be a benchmark scene using some extreme stuff like fancy particle effects, six light sources, badly made hair/clothes or an complicated environment with own light/reflection probes, aso, the benchmark would maybe only measuring how good the PC is at doing those certain extreme stuff, not reflecting every day use.
- If the benchmark scene doesn't use extreme stuff, we maybe would see no or only a very small difference between PC systems which are very different at price and performance.
- If we would see no or only a very small difference between very different PC systems, this could maybe exactly reflecting the bottlenecks of VaM and how bad it is in using PC resources.

You are in the uneviable position of finding a well balanced mix.
Maybe @meshedvr would like to give a clue which technical aspects should be typically stressed and to what percentual extent.

 

[supacavalier]

VAM already includes benchmark tests in the root of the installation folder? Running these files (they are .bat files) will open VAM with a specific scene and do whatever benchmark it's assigned to do. There's benchmarks for Baseline, CPU, ClothSim, GPU, Hair, HairRender, and even more. Why don't you just go with one of those?

 

[Jiraiya]

VAM already includes benchmark tests in the root of the installation folder? Running these files (they are .bat files) will open VAM with a specific scene and do whatever benchmark it's assigned to do. There's benchmarks for Baseline, CPU, ClothSim, GPU, Hair, HairRender, and even more. Why don't you just go with one of those?

Those don't seem to open in VR, if you are testing VR performance that's kind of important.
They are also not representative of actual scenes and performance you would expect while using VaM.
They tent to target one specific area and be fairly heavy in that regard.

Don't get me wrong, they are useful, but the scene being discussed is more of a "general public benchmark for VaM" so people can run it then say "I got xxxFPS on VaMark with this PC"
This allows comparisons of hardware with known settings and scene. People can then make better decisions when buying new machines or upgrading.

It also helps people spot when they have obvious problems with a setup not performing as it should or get real information on improvements from things like FidelityFX.

 

[TToby]

Hi,
thank you for reminding, I allmost forgot about that @supacavalier .
Those benchmarks are very early and basic stuff, that doesn't reflect the everyday use of todays VaM.
It is also not in VR and AFAIK doesn't start all that stuff like for instance 20GB of VAR files or 5000 custom morphs, which will affect your usual VaM performance.
What we are thinking off is a self running nice looking representative scene that can be started out of your actual VaM installation in VR and desktop mode, using modern stuff, different quality settings and putting out a nice log file with some numbers that can be shared.

What is interesting is the selection of different tests.
This should be taken over to the benchmark scene IMHO.

- Baseline Benchmark
- Cloth sim Benchmark
- CPU Benchmark
- CPU High Physics Benchmark
- GPU Benchmark
- Hair Benchmark
- Hair Render Benchmark
- Hair Sim Benchmark
- "Crypt Benchmark" with a moving camera, animations and different light sources.

Edit: Whoop crossposting together with Jiraja at the same time.
With "should be taken over" I don't mean "in the same form" but obviously those tests are reflecting the different technical bottlenecks of VaM.