Benchmark Result Discussion

[everlaster]

hmmm, what happened to my baseline 3 benchmark? very low, and I have fps 20-30 in many scenes

What do I need to update? CPU or more RAM?

When benchmarking the system says mostly 70-90% is GPU , RAM 50% > 90%
View attachment 436327

What do you mean by what happened - was your result higher before?

More RAM won't help for fps, but RAM speed and latencies will have some tiny effect on fps.

Upgrading the CPU will definitely help with the Baseline 3 bench and any multi-person and/or soft physics enabled scene. Depends on your motherboard what CPUs it supports.

 

[ldog]

If VR requires 90 fps for a playable experience, does this mean 4090 is the only option for VR?

Can anyone share how the VR experience is with a card below 4090?

 

[HolySchmidt]

If VR requires 90 fps for a playable experience, does this mean 4090 is the only option for VR?

Can anyone share how the VR experience is with a card below 4090?

Who said you have to have 90fps for VR? Even with a 4090 that's not gonna happen if you have a detailed environment and some nice clothings.
Like I said, said, said, said and will continue to say: VaM is mostly bottlenecked by the CPU due to physics.

Anyways, your question is a little pointless cause like-I-said: there's different headsets with different resolutions. I can tell you that I had a lot of fun with a 3080 on a Valve Index.

What many people actually miss (I think) is that you can win a lot of fps by adjusting settings. Soft body physics (the most fun) kills the most performance. Glute physics is another one. Hair, clothings, lights, environment, even body textures, everything cost performance. If you wanna have 90fps just adjust your scene.

 

[turtlebackgoofy]

If VR requires 90 fps for a playable experience, does this mean 4090 is the only option for VR?

Can anyone share how the VR experience is with a card below 4090?

90fps means 1000/90=11.1ms per frame. If your CPU finishes the physics in 3ms, your GPU has 8.1ms to render the scene to not exceed the budget for 90fps. If you have a slow CPU that finishes in 6ms, your GPU must finish all it's work in 5.1ms. So if you have a 7800x3d or 9800x3d with 6000MT/s RAM, a 4070 might suffice. 14900k is also good. VAM is not a regular game, but a physics simulation where CPU and GPU must work in serial. So a better CPU/GPU will always increase your FPS until FPS==physics rate or FPS==vsync.
There is no such thing as "bottlenecked by CPU" or "bottlenecked by GPU", but the most cost effective way to increase FPS is upgrading your CPU/RAM speeds.
Your best bet would be to look through benchmarks and check the "physics time" and "scripts time" of different CPU. This will give you an estimate how many milliseconds that CPU upgrade would give you. My experience is that you get most FPS by upgrading the CPU and RAM in any case, no matter what. If you want best FPS/$ you should upgrade to the most expensive CPU and RAM first, meaning an x3d ryzen or a highend intel with atleast 6000mhz RAM. After that buy the best GPU thats left in your budget.

 

[mostvanvege]

90fps means 1000/90=11.1ms per frame. If your CPU finishes the physics in 3ms, your GPU has 8.1ms to render the scene to not exceed the budget for 90fps. If you have a slow CPU that finishes in 6ms, your GPU must finish all it's work in 5.1ms. So if you have a 7800x3d or 9800x3d with 6000MT/s RAM, a 4070 might suffice. 14900k is also good. VAM is not a regular game, but a physics simulation where CPU and GPU must work in serial. So a better CPU/GPU will always increase your FPS until FPS==physics rate or FPS==vsync.
There is no such thing as "bottlenecked by CPU" or "bottlenecked by GPU", but the most cost effective way to increase FPS is upgrading your CPU/RAM speeds.
Your best bet would be to look through benchmarks and check the "physics time" and "scripts time" of different CPU. This will give you an estimate how many milliseconds that CPU upgrade would give you. My experience is that you get most FPS by upgrading the CPU and RAM in any case, no matter what. If you want best FPS/$ you should upgrade to the most expensive CPU and RAM first, meaning an x3d ryzen or a highend intel with atleast 6000mhz RAM. After that buy the best GPU thats left in your budget.

Exactly this. Tough I may add it might depend on what you prefer between: visual or motion fidelity more? Lights, a crowded scene, sharper resolution taxes the gpu more on visual, while scripts which set your scene in motion taxes the cpu. You can dial them to your hw level and plan your upgrade according to preference. Just be mindful of your choices.

 

[Shortstack Enjoyer]

Is it a known issue that having a lot of .var packages and other content installed causes problems with the benchmark? It gets stuck trying to load "Janie" after the 2nd hair sim sequence, wouldn't load even after over 15 minutes. Was getting around 70 FPS on average up until that point (at 1440p Ultra, 1.0 Render Scale).

 

[turtlebackgoofy]

Is it a known issue that having a lot of .var packages and other content installed causes problems with the benchmark? It gets stuck trying to load "Janie" after the 2nd hair sim sequence, wouldn't load even after over 15 minutes. Was getting around 70 FPS on average up until that point (at 1440p Ultra, 1.0 Render Scale).

janie fails to load because of a race condition somewhere in the scene loading, but only occurs if a morph fails to load while the character is loading. You should get some error like "file could not be found .... xxx.vmi" or something. For me it was some weird unrelated .var that contained morphs and which had a space in filename inside the .var. Fixed it by opening it as a zip file in 7z and renaming it correctly.

 

[Seraphim]

 

[lolmao500]

30 fps on a 4090 + 9800X3D at 8K?? Damn... we really need VAM 2.0

 

[suprtt]

Any idea what I could do to get my fps up? Just upgraded the processor from a 5 and I've seen the same processor with a 3060 getting nearly double the fps? Is the 3050 that bad!?

 

[lolmao500]

Is the 3050 that bad!?

Yes. Yes its pretty bad. Like 40% slower than the 3060.

 

[SortaLewd]

Does anyone know if upgrading to a 5700x3d from a 5600x would be worth it? Would there be decent gains? I have a 6700xt and only use VaM in Desktop.

 

[Bucketor]

Hello,

i5 13600 clean install and cpu patched.
pretty happy with this little thing

 

[brasileirinho]

Does anyone know if upgrading to a 5700x3d from a 5600x would be worth it? Would there be decent gains? I have a 6700xt and only use VaM in Desktop.

I have the same setup 5600 and 6700xt (5600x and 5600 performance difference is negligible). I also wondered if it would make any difference, and here is what I found out:

In some scenes yes. The 5700x3D is closer to the 5800x3D than it is to the 5600x. The 5600x gets 80FPS in baseline 3, while the 5800x3D gets 130FPS. So I would say you could get a 50% performance boost in some scenarios, if you are not GPU bottlenecked.

But I wouldn't upgrade. What is the difference between 80 and 120 FPS? And the scenes that get lower FPS than that you can simply delete some assets like the environment (or swap it for a more optimized one), or switch off two or three lights. You can use scene stripper plugin to help you do that.

 

[trety]

But I wouldn't upgrade. What is the difference between 80 and 120 FPS? And the scenes that get lower FPS than that you can simply delete some assets like the environment (or swap it for a more optimized one), or switch off two or three lights. You can use scene stripper plugin to help you do that.

The lights and assets [exluding it's colisions or linking them to other atoms] are compiled on GPU, not CPU, so putting them to help with fps in CPU talk isn't exactly right here.

You just said 5800x3D gives like 50% more fps in CPU heavy benchmark. Yeah, the difference between 120 vs 80 is not that noticeable. But 50% more means, in other, more balanced scenarios, ie 'realife VaM usage' scenarios, like playing actual scenes, i would say it might be huge to get 60fps instead of 40...

 

[SortaLewd]

I have the same setup 5600 and 6700xt (5600x and 5600 performance difference is negligible). I also wondered if it would make any difference, and here is what I found out:

In some scenes yes. The 5700x3D is closer to the 5800x3D than it is to the 5600x. The 5600x gets 80FPS in baseline 3, while the 5800x3D gets 130FPS. So I would say you could get a 50% performance boost in some scenarios, if you are not GPU bottlenecked.

But I wouldn't upgrade. What is the difference between 80 and 120 FPS? And the scenes that get lower FPS than that you can simply delete some assets like the environment (or swap it for a more optimized one), or switch off two or three lights. You can use scene stripper plugin to help you do that.

Hm, an additional 40~ fps definitely seems worth it to me. I tend to use plugins like Naturalis on top of using Reshade, and these things combined has a fairly heavy impact on my performance, so that's a big bump in my case. Would be interested in hearing other opinions though, especially if anyone here is using a 5700x3d with the same or similar build.

 

[brasileirinho]

The lights and assets [exluding it's colisions or linking them to other atoms] are compiled on GPU, not CPU, so putting them to help with fps in CPU talk isn't exactly right here.

You just said 5800x3D gives like 50% more fps in CPU heavy benchmark. Yeah, the difference between 120 vs 80 is not that noticeable. But 50% more means, in other, more balanced scenarios, ie 'realife VaM usage' scenarios, like playing actual scenes, i would say it might be huge to get 60fps instead of 40...

I get 80FPS in many scenes with my setup (it is the same as his). But I limit GPU usage a lot, by using GivemeFPS plugin.

 

[ShadowCreaper]

Hey
I'd like to update my PC. Based on these two Benchmark reports would you start by replacing the GPU or the CPU/MoBo/RAM?
(I've tested on my Quest 3 using Virtual Desktop Streamer, so I don't know how to turn of ASW/MS in Virtual Desktop)

 

[lolmao500]

To anyone who wants to upgrade... I would wait for VAM 2 to release... because from what I tested in the alpha, it runs WAY WAY WAY better than VAM 1 on lower specs pcs.

 

[trety]

To anyone who wants to upgrade... I would wait for VAM 2 to release... because from what I tested in the alpha, it runs WAY WAY WAY better than VAM 1 on lower specs pcs.

Yeah, since there in no physics, or strand based hairs, clothing with physics and all of goodies.
Even Beta 1 will work much better than VaM1, cuz it still gonna miss all of that.

Turn off all colisions, hairs, clothing and VaM 1 will run butter smooth too.

VaM 2 will only bring better rendering, which might be less or more demanding for GPU, depends of settings which might be more configuratable.
As far we saw lightning gonna be much less demanding in base form.
VaM 1.0 settings have close to zero impact on decent systems in desktop mode.
It gonna bring also multi thread physics, which we have no info at all atm, since Meshed didn't showed any progress with that - it's planned for Beta 2+ i think. This will be the biggest game changer. But, still we know nothing, aside Mesh gonna write his very own system for that.

Once all of the good stuff gonna be implemented, guess in like 1-2 years, VaM 2 still gonna just crush top tier PCs at the time.

 

[cbkstudios]

First pass on a new PC setup. Nothing overclocked beyond the factory tune on the ROG Strix 4090. New monitor has not arrived yet so had to do a first pass on an old work monitor.
Also, no CPU patch on this.

PC Info
i9-14900k
ROG Strix RTX 4090 OC BTF
ROG Strix Maximus Hero BTF mobo
64gb DDR5 6000MHz Corsair Vengeance RAM
Corsair RM1200x shift PSU
Thermaltake 600 case
Samsung 990 pro 4tb ssd
Artic liquid freezer III 360 AIO
3x 120 thermaltake fans
4x 140 thermaltake fans

All temps stayed below 60c during benchmarking

 

[cbkstudios]

with CPU patch