Benchmark Result Discussion

I'm in the process of overhauling my PC, and I just need a few more parts to complete it. Now, my new mobo is PCI-5, and compatible drives are slowly coming out, although the initial performance gains are quite underwhelming.

With only one PCI-5 slot available on my new mobo, I am tempted to purchase a compatible drive. Since VaM is CPU intensive and the PCI-5 slot is directly connected to it, one could assume it would greatly help, or am I misunderstanding the technology or how VaM interacts with the CPU?
 
Sans titre.jpg
 
@GordonShock
From my understanding you would benefit a lot more from faster RAM than a faster SSD. The drive speed matters when laoding a scene. But I'm no expert, could be wrong.
 
Saw some folks wondering about the 7900xtx, hope this helps! Don't mind the 1% lows, I was working in the background

1687694746877.png
 
Can someone please tell me what's wrong with my PC? My test seems very low on everything. Thank you. ???

1688186129688.png


1688186226292.png
 
Last edited:
Can someone please tell me what's wrong with my PC? My test seems very low on everything. Thank you. ???

View attachment 263553

View attachment 263556
It's not really that bad. My guess would be a bit slow DDR4 RAM as for the overall platform, and then, it's even slowed down by populating all 4 slots.
Yours GPU is also a little underperforming. Is it a bit power limited? Or air cooled? You can try to use MSI Afterburner to create some curves to optimize it.
Benchmark-20230106-162805.png
 
Last edited:
It's not really that bad. My guess would be a bit slow DDR4 RAM as for the overall platform, and then, it's even slowed down by populating all 4 slots.
Yours GPU is also a little underperforming. Is it a bit power limited? Or air cooled? You can try to use MSI Afterburner to create some curves to optimize it.
Thank you so much, yes it's an air-cooled GPU. Also, no power is limited on anything as I don't know how :). Will try Afterburner.
 
Upgraded from a 1060 3GB to a RX6800, both paired with i5-11400f

old:
1684912540.jpg


new:
1686003999.jpg


However through my headset meta quest 2 i seem to be locked around 35 FPS, i am quite sure ASW is turned off since in other since i hit higher FPS. Is there anything i am missing here?

1688227481.jpg
 
Last edited:
I've tried the Benchmark with official settings following the on screen notes and the benchmark aborted itself, without me doing anything after the hair part (or the one right after that) after throwing the errors that I've marked in red quarre. The hub page states I have all the files. How do I fix that? Is there a way to verify my files? (maybe that could also fix the rest of the error spam that I've learned to live with)

Sorry for posting errors in here.

1688749925556.png
 
I've tried the Benchmark with official settings following the on screen notes and the benchmark aborted itself, without me doing anything after the hair part (or the one right after that) after throwing the errors that I've marked in red quarre. The hub page states I have all the files. How do I fix that? Is there a way to verify my files? (maybe that could also fix the rest of the error spam that I've learned to live with)

Sorry for posting errors in here.

View attachment 266338
Dunno if you want to test yours machine, or 'custom install'...
If you want to compare yours pc build with others posted in this thread, you should use 'clean \ vanilla' install - base game with only default, benchmark and it's deps, vars...
If you want to compare 'vanilla install' vs 'current intall' on yours own machine...
You really want to fix errors first.
I never really made any proofs for it, but in my personal impression, VaM runs slower if there are errors in the log. Especially if there is a lot of them. In my experience, they're slowing down VaM more than having these missing vars installed.
It's logical actually - unity, especially old builds, really hate long lists. Thats why we got limited\shortened shortcuts tab, and... thats why VaM lags so much if there is physics explosions spam in the log.

Back to the topic. Idk about sharr var, but demi eyes are perfectly fine on hub. You really should redownload both of them to fix zip errors.
Ascorad expressions had a mistake in filenames in '9.var' version if i recall [space in one of the morphs file name, vmi or vmb files]. Update the package, get rid of the v9 var, and you will be fine with this one.
As for the other errors i see many of the missing vars are hub-hosted... Going to in-build missing deps tab might be a good start.
 
Last edited:
Ryzen 7800X3D with RTX 4090
VR test with Virtual Desktop Streamer at God mode resolution
Benchmark-20230715-063911.png
 
Last edited:
Another VR test with Virtual Desktop Streamer at Ultra mode resolution
Ryzen 7800X3D RTX 4090
Benchmark-20230718-234338.png
 
Last edited:
So apparently a recent beta driver update for the 7900 XTX has improved VR performance for ”some” applications. Has anyone tried with a 7900 tried these latest drivers with the vam benchmark?
 
is there any consensus about performance gains going from DDR4 to DDR5? what about CPUs? does the game run better on intel/AMD or it doesn't matter at all?

i am looking to buy a rig and could use some pointers/golden rules.
 
is there any consensus about performance gains going from DDR4 to DDR5? what about CPUs? does the game run better on intel/AMD or it doesn't matter at all?

i am looking to buy a rig and could use some pointers/golden rules.
Definately DDR5 for performance on here. If your mainly only gaming and simming, AMD X3D cpu. If you also do rendering/video editing and other stuff, Intel i7/i9.
 
Want to see an example of being CPU bottlenecked? 9700k with a 3080 -> 4090
3080
9700k_3080_Benchmark-20230806-111505.png

4090
Benchmark-20230808-202727.png


Notice how only the mirror, cloth and hair benchmarks improve and it's only really the mirror that sees the expected uplift

CPU upgrade at some point - I got the 4090 for it's VRAM - for other projects - it's not even breaking sweat in VR as it's being held back by the CPU

Honestly it's only Baseline 3 that really shows what your system performance with VAM will really be
 
airgpu :
Location:Canada (Central)
vCPUs:16
RAM:64 GB
SSD:100 GB
GPU:Nvidia A10G
Stream:1080p 60 FPS, 15 Mbit/s
OS:Windows Server 2022

Price : $3.30 CAD / hour

New Project (2).png
 
So... I finally bought again 2 sticks of DDR5, instead of 4 of them i been using for a past few months...
I had that kind of setup last year, but due irl stuff i had to return them.
So here is my before [4 sticks x16gb]:
Benchmark_vanilla_4_sticks_no_xmp.png
And here's after [2x 32gb, ddr5, 6400, cl32]:
Benchmark_vanilla_2_sticks_xmp.png
But i also installed CPU Contact Frame from Thermal Grizzly.
For uknown reasons, maybe these results were just max single core performance of my setup, it didn't gave anything at all in vanilla tests. Maybe I saw lower temps at NZXT lcd, but i didn't really monitor them that much before i changed the mounting.

But the real interesting stuff starts with my main, bloated ~ 590 gb install.

Note: These tests below aren't meant to compare hardware! These are only meant to show potential upgrade in 'real life VAM usage' after changing the RAM and contact frame.
The reason for that is simple - there is no 2 identical VAM installs, so my almost 600gb of content, might be heavier for performance than someone else 800gb one, if it contains just clothing etc. And the same, since my install is heavily managed, including manually modifying, for personal use, other creators vars, it might be lighter than other people half of the size ones.

Firstly, test before any changes:
Benchmark_bloated_4_sticks.png
Changing the RAM alone
Benchmark_bloated_2_sticks.png
RAM alone gave me nice physics time reduce in both, vanilla and bloated installs.
But here is bloated with new contact frame:
Benchmark_bloated_2_sticks_ Contact_Frame.png
Physics time went even further.
It's hard to say if these contact frames are worth as for VAM. It didn't gave that much tbh, but still it's some improvement in terms of performance, and most likely in 'quality of life' too. I can't really test it, it's too freakin hot in here atm, and my small mobile AC couldn't really hold the same ambience temp during the tests.
If someone would start building theirs gen12\13 setup, it's definatelly something to consider.
If someone have it already, idk if it worth the hassle unmounting a whole motherboard from the case - unless yours cpu cooling comes with good backplate holding the mounting.
I did, and since it actually gave something, i'm happy.

/EDIT:/
Okay. So i just played some MMD motions... Holy crap. It now keeps CPU around 61-63 celsius in the scenes it was at 70ish. This thing definatelly works - maybe won't give much of performance paired with good coolers, but it's definatelly quality of life upgrade.
 
Last edited:
Back
Top Bottom