Benchmark Result Discussion

[Ged366]

Hi Folks, I'm trying to get my head around these results. I've heard people say you should expect a round a 50% fps drop between desktop and VR, but I've got a massive drop. In fact it seemd to sit at 30fps rock solid for the first few tests almost like it was limited.

The drop depend also of the HMD resolution, with your 50% is a bit optimistic, but yeah I think there a problem with your second result


For what it's worth this the drop I get with 5800x 3D / 6800 XT on Reverb G2 :

1080P :

Reverb G2 :

 

[trety]

Hi Folks, I'm trying to get my head around these results. I've heard people say you should expect a round a 50% fps drop between desktop and VR, but I've got a massive drop. In fact it seemd to sit at 30fps rock solid for the first few tests almost like it was limited.
what am I missing?!
Thanks in advance.
View attachment 197856View attachment 197858

People says performance is usually like a half cuz they comparing 1080p to NATIVE 'common' HMD res.
Vive Elite native res is 2880 x 1700, seems like U trying to use it at x2 scale?

Ged366, Yours settings are way too high for G2 too, unless it was intentend for comparision with Bob.harris12

1080 [repost from earlier]

Spoiler

Desktop 4K

Spoiler

VR G2 +- native res

Spoiler

 

[Ged366]

Ged366, Yours settings are way too high for G2 too, unless it was intentend for comparision with Bob.harris12

I'm not sure I'm the best to explain this with my broken english but it's normal to see higher res to have a native res, I will try to find the post on reddit it will be simpler

Unless someone explain it before I find it I will back update it.

 

[trety]

I mean i'm not sure if Vive needs to run VaM with steam vr, or it has some native system, but if it's via steam VR... U might have render scale in settings on 'auto' or even 200%

 

[Bob.harris12]

Thanks for the replies.
I had a bit of a play about with the settings in steam VR and seem to have improved things. Haven’t run the full test, but the first warmup runs around 45fps average so that apparent 30fps limit has been lifted. I’ll have another play in the morning then do a full benchmark and post my findings.

 

[Bob.harris12]

I'm not sure I'm the best to explain this with my broken english but it's normal to see higher res to have a native res, I will try to find the post on reddit it will be simpler

Unless someone explain it before I find it I will back update it.

I know what you mean. I did wonder about the high resolution, but that’s it set to 100%. If I lower it menus start to get quite blocky.

 

[Ged366]

I mean i'm not sure if Vive needs to run VaM with steam vr, or it has some native system, but if it's via steam VR... U might have render scale in settings on 'auto' or even 200%

There the link, in short it have to do with how fresnel lenses work and yes my result was with resolution on "auto" in SteamVR, like Bob said If I lower it even by a notch text start to get aliasing.

https://www.reddit.com/r/HPReverb/comments/k82pnv

 

[keycode]

I'm not sure I'm the best to explain this with my broken english but it's normal to see higher res to have a native res, I will try to find the post on reddit it will be simpler

Unless someone explain it before I find it I will back update it.

I keep my reverb (g2) res. in the crapware SteamVR "advanced" (so called -->) settings at around 45% to match the reverb g2 per eye resolution... and: I see no big gap in visual quality even compared to an average session in my desktop (vam mode) with a good Asus 144Hz 1440p display. ps: I am happy user too of a "modest" MSI ventus 3080 12Gb (and i7 13700k).

 

[Tomb]

Hi Mates!

I'm wondering what to expect from upgrade a 3080 to 4070 ti, 4080 or 4090 with meta quest 2.

Tom's hardware says in 4k:

GeForce RTX 4090 100.0% (116.0fps)
GeForce RTX 4080 78.7% (91.2fps)
GeForce RTX 4070 Ti 62.7% (72.8fps)
GeForce RTX 3080 54.1% (62.8fps)

Can't find enough 4070 and 4080 vam benchmark resultss to check, but the 3080 and 4090 results are seems to be near 50/100% ratio. Is it correct to expect the other cards perform the same in vam as tom's article says?

 

[keycode]

Hi Mates!

I'm wondering what to expect from upgrade a 3080 to 4070 ti, 4080 or 4090 with meta quest 2.

Tom's hardware says in 4k:

GeForce RTX 4090 100.0% (116.0fps)
GeForce RTX 4080 78.7% (91.2fps)
GeForce RTX 4070 Ti 62.7% (72.8fps)
GeForce RTX 3080 54.1% (62.8fps)

Can't find enough 4070 and 4080 vam benchmark resultss to check, but the 3080 and 4090 results are seems to be near 50/100% ratio. Is it correct to expect the other cards perform the same in vam as tom's article says?

seriously you expect some true real sensible etc. "upgrade" [!?] in vam visual VRquality (besides some unrelevant slightly better faps per second) stepping to a 40xx from a still excellent 3080 .... with a zukkiniberg metarest-toy??

 

[Tomb]

seriously you expect some true real sensible etc. "upgrade" [!?] in vam visual VRquality (besides some unrelevant slightly better faps per second) stepping to a 40xx from a still excellent 3080 .... with a zukkiniberg metarest-toy??

I'm just curious, and just expecting some test results.

 

[keycode]

I'm just curious, and just expecting some test results.

There are some "astonishing" fapping per second stat results, in this vam-benchmark thread, of some happy mates with 4090 and 13900k. Easy to suppose that a 4070ti should be the same or some points better than a 3090ti. The problem is... if it is worth for the actual cpu hungry vam physics to grow an "old" still very good vam-desktop with some more money vam-investment for an overdosed vam-gpu. The answer should be yes of course, if some more vam-faps per second are so much worth to get.

 

[Bob.harris12]

I dropped the in eye res to 100% (the standard in Steam is 150%). I also removed any upscaling and this is the result.

Slightly better, but not as good as I was hoping for with the upgrade to a 3080. VAM really does make your PC work hard!

Am I right in thinking the FPS hones in on 30 / 45 / 90 due to the headset displaying at 90hz?

 

[juanito06200]

salut tout le monde

mon résultat

 

[trety]

I'm not sure I'm the best to explain this with my broken english but it's normal to see higher res to have a native res, I will try to find the post on reddit it will be simpler

Unless someone explain it before I find it I will back update it.

I actually forgot to link this too:

https://www.reddit.com/r/HPReverb/comments/lo2s11

In short WMR+SteamVR have a bug which causing somewhat of double rendering frames, first it's rendered in native WMR then it copied over. Which is causing performance issues.
Thats why i was sugesting 100% resolution in steamVR settings, as it's reccomended in that reddit post.
Unfortunatelly unless VaM 1.x would have newer implementation of OpenXR we can't do anything more.

Good thing, VaM2 will run natively on WMR platform so this won't be an issue 'WhenIt'sDone'

 

[keycode]

I actually forgot to link this too:

https://www.reddit.com/r/HPReverb/comments/lo2s11

In short WMR+SteamVR have a bug which causing somewhat of double rendering frames, first it's rendered in native WMR then it copied over. Which is causing performance issues.
Thats why i was sugesting 100% resolution in steamVR settings, as it's reccomended in that reddit post.
Unfortunatelly unless VaM 1.x would have newer implementation of OpenXR we can't do anything more.

Good thing, VaM2 will run natively on WMR platform so this won't be an issue 'WhenIt'sDone'

why people insist mentioning only a confusing "theoretical" [???] "100%" setting in crapware bugged steamVR "advanced" general settings if there is [cheers!] instead a second more useful evident banal numeric value?
So what when (for instance) you select your own custom resolution for the reverb G2 ?

(a 100% symbolic value in itself it means nothing, when this, well known since years, specific-contestual-crapware-bullshit-conflict/problematic it is still unsolved by steam/windows-mixed-reality 100% drunk-moral-police-guys) ..... question marks ----> ?????

----> in my primitive beautiful latin root language this parameter it's called "risoluzione per occhio" [tr.> per eye res.]
this is my actual free custom setting (and more or less that should be the REAL default MAX. res. value for the fucking reverb):

I suppose that if a reverb g2 owner doesn't "check and correct" (as I tried to do some years ago when I bought it)

this overdosed absurd default (100%) setting in crapware steamVR general settings,

then you really get the impression that a reverb g2 resolution at 100% it is something too much difficult for a standard VR pc hardware... and this is what so many people write in the web, mostly guys who never touched in their life a reverb g2, talking just around bullshits they found in some forum. The responsability for this wrong information is on WMR and steam gang, but definitively and even more suicide-like (by commercial point of view) also on HP company. I don't pretend to be so smart, but even with my oldest gpu (gtx1070) I could play vam VR with the reverb. Is it not enough to consider "bullshit" those talking about reverb g2 being so much "demanding" , for instance as compared to a rift-s (just to mention another excellent VR headset I still have, unfortunately the last good headset made by zukkiniberg and company... before the ugly quest-toy era) ?

 

[brasileirinho]

Comparison Ryzen 1600 vs 5600 (using an 6700XT):

Marginal average performance uplift, but huge Baseline 3 difference (3x faster).

 

[vnet]

Comparison Ryzen 1600 vs 5600 (using an 6700XT)

Baseline 3 is the most cpu limited, so that makes sense... Just like the game itself, when there is more than one person atom in your scene. That's why this benchmark is far from perfect, it is close to unuseable for cpu and in game performance testing/comparison. For Vam I would always pair the same gpu with an higher end (single core performance) cpu than I would usually recommend for ordinary pc gaming.

 

[ruinousdeity]

Hi there, I'm a totally newbie just knew and bought the software from the steam.
So this is my 2nd post here.

I move my QA post from here https://hub.virtamate.com/threads/i...re-than-3-people.32581/?view=votes#post-88344
, since there are more people here.

Yes I did run the benchmark here but the results are made combining both cpu and gpu.
But unlike 3D mark, which is a software that separately evaluates the performance of the CPU and GPU and then combines them to calculate the overall performance score, the benchmark does not have that feature.

My question is, in order to handle soft body physics computation for 3-5 people (with a maximum of 15 people), I'd like to upgrade my i7-8700 CPU to an AMD R9 7950, an AMD Threadripper TR 3960X, or a WRX80 series CPU like the TR PRO 5975 WX be more suitable for my computational needs?

Here are the previous post screenshot.








VAM is a single-threaded software that supports multi-core processors and multi-processor environments, but at least in INTEL's CPU, it has poor performance optimization for logical processor technology.
Here are those YT video demos

 

[keycode]

ps: (about some videos spots in this thread) why oh why our sweet moderators
did cut the from the "like" list???

 

[Ged366]

why people insist mentioning only a confusing "theoretical" [???] "100%" setting in crapware bugged steamVR "advanced" general settings if there is [cheers!] instead a second more useful evident banal numeric value?
So what when (for instance) you select your own custom resolution for the reverb G2 ?

From what I understood (coming from WMR dev) you cant think of HMD lcd res as a flat screen, on flat screen each image has the same resolution (1/1 pixel) as the display.
On HMD the only 1/1 pixel is on dead center the more you go to the edge the more the image is stretched (this is a trick to achieve the 100° FOV on such tiny LCD panel) so 1 pixel of the native image coming out of the GPU is stretched on 1.5 pixel of the HMD, then the fish eyes lenses undistord back the image.
So to get 1/1 pixel on peripheral view we need 1.5x the actual res.

But frankly the peripheral view is so blurred and filled with chromatic aberration that one could wonder what's the point to stretch it to get better result in this area.
Everyone its taste on this one.

 

[keycode]

From what I understood (coming from WMR dev) you cant think of HMD lcd res as a flat screen, on flat screen each image has the same resolution (1/1 pixel) as the display.
On HMD the only 1/1 pixel is on dead center the more you go to the edge the more the image is stretched (this is a trick to achieve the 100° FOV on such tiny LCD panel) so 1 pixel of the native image coming out of the GPU is stretched on 1.5 pixel of the HMD, then the fish eyes lenses undistord back the image.
So to get 1/1 pixel on peripheral view we need 1.5x the actual res.

But frankly the peripheral view is so blurred and filled with chromatic aberration that one could wonder what's the point to stretch it to get better result in this area.
Everyone its taste on this one.

thanks for the linear and clear answer mate ... as you could understand from my polemic, profane talk, I am surprised about this lack (conflictual parameters value) in the crazy synergy between WMR and steamVR compromise, and of course worried with the fact that the "double" settings can cause a critical (gpu power demanding) situation when we start a VR session with reverb G2. I am sorry for the argument of the post, but (IMHO) I hope that it can be considered an interesting sub-topic even in a benchmark thread. Just to wide the view a little from only comparative details about wich gpu has the biggest dick.

Yes... that "sweet spot" when I play VR with the reverb it is nerfy sometimes ...

the two integers value (WMR vs steamVR), eventually they are not even identic (speaking of res. per eye), wmr showing the total res. of the reverb headset display, while steamVR shows the more easy to understand res. per eye, even if of course it is resulting from the same single headset monitor/display.
That's just one element because of wich I suspect that setting in steamVR "just and only" the real default res. per eye for the reverb G2 it is the best solution to get a correct, a better performance economy in graphic feedback from our poor gpu.

ps: I still love my caucasian Sparrow-navy sailor GPU big boobs:

 

[nebka94]

13700K with a 3090 Fe

Nice

 

[kioren]

That's pretty good. 5900X + 4090 + 64gram

 

[L_WOOD]

Hello everyone! I'm interested in upgrading my PC for VAM. I currently have a AMD 5800x, 64gb ram and a RTX 3070ti, I play in VR on the Quest 2 via Virtual Desktop.
I was hoping to get some information form this thread, but the results vary greatly, and sometimes people with similar machines achieve very different results.
So perhaps someone more experienced could answer some of my questions:
1. How come physics time is dependent on rendering resolution, are the physics calculated per pixel?
2. Has someone checked DDR4 vs DDR5 performance?
3. Will a 5950x processor or a RTX 3090 be a better upgrade for me? Or is only a new Intel 13000/ AMD 7000 series processor able to make any significant fps increase?