Benchmark Result Discussion

[bladerune9]

Would you mind doing a run with alternate CCDs disabled? I'm looking to build a new PC and am wondering if VAM benefits more from the 3D v-cache CCD or the higher clock CCD (and therefore whether its worth it to wait for the 7800X3D in April).

The other user's 7900X3D benches are nice, but the fewer v-cache cores make it unrepresentative of the rest of the 7000X3D lineup.

EDIT: alternatively, do you know if the core parking was working right and VAM was running on the v-cache cores?

I manually set the process affinity so only the 3d v-cache ccd was used. I also did a benchmark without doing that and it defaulted to the high frequency ccd which was performing a lot worse in the benchmark.
Sadly the machine is now in production so I will not be able to test anything on it.

 

[Magiwarriorx]

I manually set the process affinity so only the 3d v-cache ccd was used. I also did a benchmark without doing that and it defaulted to the high frequency ccd which was performing a lot worse in the benchmark.
Sadly the machine is now in production so I will not be able to test anything on it.

That is exactly what I was looking for. Ty!

 

[specdizzle]

Am I simply underpowered in the CPU department here? I've tried everything short of reinstalling Windows, but I cannot seem to get anything stable past 30 frames in the benchmark, and many scenes will fail to hit 45 and start slow-motion regardless of what I do with settings.

Looking at system metrics, it does appear to be putting my GPU to work as high as 90% load, but my render times are fine.

The thing that baffles me is the physics time. This really seems to be what's killing my performance at the moment, but CPU temps are in the 50s, and the heaviest loaded core will only spike past 30% while a scene loads, then it drops right down to where the most loaded core is hangign in the 20s while the scene runs. It is almost like it's just not using the CPU. I've tried setting up the ultimate power profile and running ThrottleStop to make sure that I'm running as wide open as I can.

Any insight as to what's going on would be greatly appreciated.

 

[Seraphim]

Am I simply underpowered in the CPU department here? I've tried everything short of reinstalling Windows, but I cannot seem to get anything stable past 30 frames in the benchmark, and many scenes will fail to hit 45 and start slow-motion regardless of what I do with settings.

Looking at system metrics, it does appear to be putting my GPU to work as high as 90% load, but my render times are fine.

The thing that baffles me is the physics time. This really seems to be what's killing my performance at the moment, but CPU temps are in the 50s, and the heaviest loaded core will only spike past 30% while a scene loads, then it drops right down to where the most loaded core is hangign in the 20s while the scene runs. It is almost like it's just not using the CPU. I've tried setting up the ultimate power profile and running ThrottleStop to make sure that I'm running as wide open as I can.

Any insight as to what's going on would be greatly appreciated.

View attachment 225985

I you do a 1080p flatscreen benchmark run, we would be able to help you better. To many different variables hitting your cpu/gpu when running the benchmark in VR: Headset base SS, Reprojection, VR runtime, etc. Plus a used bloated install folder adds to physics time.

 

[HolySchmidt]

@specdizzle
With this resolution and a 3060 I would suggest you're in a GPU bottleneck. But do as seraphim said: run a 1080p desktop benchmark and we have some more important infos.

 

[specdizzle]

Thank you very much for the help. Here is a flat 1080 run. I'm definitely interested in upgrading if it will help get to a steady 45FPS in VR, I'm just not sure where the best bang for the buck would be.

 

[specdizzle]

I got to looking into upgrade options and Dell being Dell made the power supply proprietary so any significant graphics or CPU upgrades that require a larger PSU will require a new MOBO, so....I just ordered parts for an i9-13900K and RTX 3070 ti build...

 

[HolySchmidt]

Wait a minute. 13900K and 3070Ti, in my opinion, is not a good couple and in my eyes no big upgrade from your 3060. 13900K, which I own (only because I wanted the best single core performance, not because it was a smart buy money wise), is a really nice CPU, no doubt. But you could also buy a 13700K and invest the difference in a better GPU, how about that? Makes more sense to me. 3080 or 3080Ti or maybe 6800XT/6950XT. That would be my suggestion.

 

[specdizzle]

sorry, typo.... RTX 4070 ti

From what I've found, my particular system has a proprietary power supply and motherboard. and only provides enough power for the 3060 that it came with. I considered going nuclear and getting a 4090, but the price jump between that and the 4070 was a little steep for what I'm wanting to spend

 

[specdizzle]

Much better, this is default fresh windows/VAM install...

 

[HolySchmidt]

That looks much better
Have fun with your new system! 4070Ti of course is a totally different story then 3070Ti.

 

[linkp2p]

I made a spreadsheet to help people compare various submitted results, sorry if there is any typo errors in the data or username, hope it helps

https://docs.google.com/spreadsheet...ouid=103439441323920316147&rtpof=true&sd=true

 

[HolySchmidt]

Nice one, thank you
We already had a seperate thread just for the results with an overview but I can't find it anymore, don't know the name.

 

[Magiwarriorx]

Hey guys, I'm not really matching everyone else's 4090 performance numbers. What's going on? /s

I'm mid-upgrade waiting on the 7800X3D and decided to run one of the most CPU-bound benchmarks VAM has had. The first benchmark was running on 16GB RAM with XMP off, with the 6700k at 4.2Ghz. The second one is with 3200Mhz RAM (XMP on!), with the 6700k up to 4.8Ghz.

 

[mrmr32]

My results on using an AMD 3950X and RTX 3070.
3DMark reports a score of 13,218 (graphics 13,345 - cpu 12,545).
Edit: For the RAM, I'm using x4 Kingston HyperX Fury Black DDR4 at 3200Mhz (CL16)

 

[mrmr32]

To contrast my previous post, I wanted to check if my laptop (Alienware X15) could give better results. It has a much powerful single-core CPU (i9-12900H), but besides technically having a better GPU it gives worse performance (probably due to thermal issues).
The first image is the result on running the benchmark with the fans on mode "performance" (you can tune the system using an Alienware app).
To try to compensate the GPU I also ran the benchmark on mode OC1 (OC2 triggers thermal throtteling) and max fan speed; you can find that information on the second image.
The RAM that is using is x2 DDR5 6400MHz.
3DMark reports a score of 11,179 (graphics 11,381 - cpu 10,162) without overclocking, and 11,428 (12,322 - 8,100) with OC1 mode.

OC1 benchmark:

 

[Seraphim]

Nice one, thank you
We already had a seperate thread just for the results with an overview but I can't find it anymore, don't know the name.

Hardware Performance comparison tables

The following tables are sorted results from MacGrubers VAM Benchmark collected here. Updated - 22.April.2022 - I try to avoid my opinion or hardware recommendations here - there is already to much 'brand XY is better'-fanboyism on the internet. It should be clear that one can be happy with...

hub.virtamate.com

 

[Paulj]

Newly upgraded rig, so here are my benchmarks:
Is this the first benchmark with an RTX 4080 card?


I use a Meta Pro headset and it just hums along at 90FPS in VR all day long..

 

[Seraphim]

View attachment 227089View attachment 227090Hey guys, I'm not really matching everyone else's 4090 performance numbers. What's going on? /s

I'm mid-upgrade waiting on the 7800X3D and decided to run one of the most CPU-bound benchmarks VAM has had. The first benchmark was running on 16GB RAM with XMP off, with the 6700k at 4.2Ghz. The second one is with 3200Mhz RAM (XMP on!), with the 6700k up to 4.8Ghz.

Yeah your CPU is definately strangling your 4090. 7800X3d will fix that.

 

[juliosanchezcuenca@hotmai]

Hi there.

Here my results with my new 4080, and my future ex-cpu the 3600 NO X.

I'm thinking of buying the 13600k.

Unfortunately, I wanted to benchmark with my old 5700xt graphics card. But the benchmark gave me an error when loading the Janie asset. I have finally found that running the benchmark in custom does not fail.


But I can say this for whoever cares:

In a scene from Universen, one of the last. With nice quality i got with my 5700xt with this setting:

1 scale
2x msa
Shader: High
soft physics: on
quality physics: off
mirror:active and use it
pxel lig: 1-2
smoth: 3-2
glow: off
procees prio: high
phycis rate: 60
physics UpdateCap: 1-3 no idea.

I got 32-36 FPS. (PRESSING BUTTON GIVE ME FPS- No idea if only better hair or all)

Now with 4080

1 scale
8x msa
Shader: High
soft physics: on
quality physics: off
mirror:active and use it
pxel lig: 3
smoth: 4
glow: off
procees prio: high
phycis rate: 60
physics UpdateCap: 3

I got 30-35FPS. (NOT PRESSING BUTTON GIVE ME FPS-)
I got 40-47 FPS. (PRESSING BUTTON GIVE ME FPS- No idea if only better hair or all)

I did another test, i had active AWS limit 45. But for got 70-80,depends from scene and only when physycs vallues like quality or soft are off.



Be that as it may, here is the result. And waiting for VAM 2 to see if they manage to improve performance.

Think need urgently a 13600k or buy the new 5800x3d (I'm not sure about the last one). Amd user all my life i now wanna try intel+nvidea now.
Too many bugs on my old 5700xt and 380x, software sucks.

48gb are 3200mhz

 

[Paulj]

I was still within my 30 day window, so I returned the 4080 and swapped it with a 4090. Results of the 4080 on the same system are above -

Benchmark Result Discussion

I was still within my 30 day window, so I returned the 4080 and swapped it with a 4090. Results of the 4080 on the same system are above - I think your addonPackages folder is too big, you can move files to another folder, and get huge improvement.

hub.virtamate.com

 

[juliosanchezcuenca@hotmai]

VaM (OpenVR).bat to START "VaM" VaM.exe -vrmode OpenXR s

I was still within my 30 day window, so I returned the 4080 and swapped it with a 4090. Results of the 4080 on the same system are above -View attachment 231131

Really are the same result, limit for cpu i think. No idea why people hates 4080. Better what 4070ti and worst than 4090. The prices the same. 900- 1300-1800. 1 world issues.

 

[Paulj]

Yes, I cannot tell the difference visually between the 4080 and the 4090 on my system, and I've logged a LOT of hours in VAM this week.

Perhaps when I upgrade to an AM5 CPU later this year it may break the bottleneck. For now its about the same. Great card however you should see the 4090 churn through my AI models!

 

[HolySchmidt]

4090 is a great card. You are CPU bottlenecked obviously.

 

[gokusanmar]

I recently got a new laptop. These are the results of what I have been using so far ( it was version 3, so not apples to apples comparision):

And here are the results of the new laptop

And as a curiosity here are the results of the same laptop but running in silent mode ( it was running more than half as loud while retaining majority of the performance)