Benchmark Result Discussion

[Motoyo]

7950X with 32GB ram, and 4080

 

[Matt_]

 

[keycode]

and now ??? how can I get in the future a fresh bench. ❤ result upon wich gpu/cpu can manifest the biggest (dick) power without a benchmark update (after installing the fantastic last vam century release) ?

 

[Bob Nothing]

With my new 4090. I couldn't get it to do the 'official' benchmark, my guess is that is because I'm running at 4k. (I don't have a 1080p monitor anymore.) In any case, here are my results in 4k with my MSI 4090 liquid cooled and my 7900x liquid cooled.

 

[keycode]

With my new 4090. I couldn't get it to do the 'official' benchmark, my guess is that is because I'm running at 4k. (I don't have a 1080p monitor anymore.) In any case, here are my results in 4k with my MSI 4090 liquid cooled and my 7900x liquid cooled.

Probably it has not to do with your monitor res. (and you can set it manually at the "official res." or even to 60Hz) ps: I had an half intention to change my poor old 1440p monitor for a 4k ... but considering the "limited" power of my (new!) 3080.. I guess I should resist to the temptation.

 

[trety]

With my new 4090. I couldn't get it to do the 'official' benchmark, my guess is that is because I'm running at 4k. (I don't have a 1080p monitor anymore.) In any case, here are my results in 4k with my MSI 4090 liquid cooled and my 7900x liquid cooled.

Seems like Ur'e the very first Ryzen user paired with DDR5 + 4090 in this thread.
If U would like to try to test CPU 'alone', without bottleneckin, by GPU as U were with 20x0 card if i remember correctly, U can simply setup Urs windows desktop monitor resolution to 1080p, and then do the same for Vam. It will force it to go 1080p.
Also, seems like 1.21 update broken benchmark a bit, it doesn't allow to change physics update cap to 20 anymore, blocking official test to run.
In 4K there wasn't much difference between two tests i made, 3 vs 20, but it actually went a bit better with cap at 3 for 1080p

Out of curiousity, my system using Urs settings

Spoiler: 4K

Spoiler: 1080p

Another out of curiosity test - 1080p using my main, bloated, ~ 560Gb install, with my basic reshade.

Spoiler

 

[Seraphim]

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.

Next month, get the 7800X3d. It will be better than 13900k in single core performance, and do not have to worry about core scheduling, since all cores are the same. Or get 7950X3d if you also need more cores for other types of applications; But then you will need to make sure the cpu scheduler uses only the 8 3d cache cores fore your games.

 

[trety]

Next month, get the 7800X3d. It will be better than 13900k in single core performance, and do not have to worry about core scheduling, since all cores are the same. Or get 7950X3d if you also need more cores for other types of applications; But then you will need to make sure the cpu scheduler uses only the 8 3d cache cores fore your games.

As i do agree with 'wait to see what AMD would bring', i'm not sure about theirs single core scores tho.
AMD still haven't show ANY single core perfomance graphs for incoming x3d chips.
And last gen x3d were actually worse in that matter than 'base' [non 3d cache] versions.

Sure they're were better with multi-thread and newer games in general, thanks for that bigger cache, but even low-end Intel [for example i3 12300T beats up 5800x3D] crushed them up in single core.

PassMark CPU Benchmarks - Single Thread Performance

Benchmarks of the single thread performance of CPUs. This chart comparing CPUs single thread performance is made using thousands of PerformanceTest benchmark results and is updated daily.

www.cpubenchmark.net

 

[Bob Nothing]

As i do agree with 'wait to see what AMD would bring', i'm not sure about theirs single core scores tho.
AMD still haven't show ANY single core perfomance graphs for incoming x3d chips.
And last gen x3d were actually worse in that matter than 'base' [non 3d cache] versions.

Sure they're were better with multi-thread and newer games in general, thanks for that bigger cache, but even low-end Intel [for example i3 12300T beats up 5800x3D] crushed them up in single core.

PassMark CPU Benchmarks - Single Thread Performance

Benchmarks of the single thread performance of CPUs. This chart comparing CPUs single thread performance is made using thousands of PerformanceTest benchmark results and is updated daily.

www.cpubenchmark.net

I'm planning on getting a 7900x3D once they launch. I currently have the 7900x, but I've got a buddy who wants it, so, gives me the excuse

 

[Seraphim]

As i do agree with 'wait to see what AMD would bring', i'm not sure about theirs single core scores tho.
AMD still haven't show ANY single core perfomance graphs for incoming x3d chips.
And last gen x3d were actually worse in that matter than 'base' [non 3d cache] versions.

Sure they're were better with multi-thread and newer games in general, thanks for that bigger cache, but even low-end Intel [for example i3 12300T beats up 5800x3D] crushed them up in single core.

PassMark CPU Benchmarks - Single Thread Performance

Benchmarks of the single thread performance of CPUs. This chart comparing CPUs single thread performance is made using thousands of PerformanceTest benchmark results and is updated daily.

www.cpubenchmark.net

You have to look up gaming benchmarks. 13-50% faster than non 3d cache chip. You can even see the difference on posted vam benchmarks on here. Synthetic benchmarks do not utilize the 3d cache if at all.

Ryzen 7 5800X3D vs. Ryzen 7 5800X: Zen 3 Gaming Shootout

On the menu today is another 40 game benchmark -- actually 41. This time it's the 5800X3D against its spiritual predecessor, the 5800X, to see where that...

www.techspot.com

 

[trety]

You have to look up gaming benchmarks. 13-50% faster than non 3d cache chip. You can even see the difference on posted vam benchmarks on here. Synthetic benchmarks do not utilize the 3d cache if at all.

Ryzen 7 5800X3D vs. Ryzen 7 5800X: Zen 3 Gaming Shootout

On the menu today is another 40 game benchmark -- actually 41. This time it's the 5800X3D against its spiritual predecessor, the 5800X, to see where that...

www.techspot.com

Like i said, in overall gaming x3D chips are better than theirs base versions, but we're on VaM benchmark thread. As U can see in most of the results in here, the main bottleneck for most of the people are models physics, and this happens on single core.
In terms of VaM 1.x it doesn't matter how good CPU is in multicore [if it has at least 8 cores for other, non physics related stuff], if it sucks in single core = it will be bottlenecking overall experience.
You must also note that to be able to really compare CPU alone in VaM, both systems must run the same GPU. Physics are calculated by frame, so if GPU render them slower, CPU will go worse too.
Example:

Benchmark Result Discussion

Yes maybe, it's true that all "ms" looks better everywhere, but I'm not used to check at this, more used to rely on FPS with other benchmarks. Will take a closer look to what all these data are meaning before start to investigate if there something wrong on driver/software side. There is...

hub.virtamate.com

same cpu, using 3090, then 4090. No other changes in the system. Got almost half physics time reduce - which suppoused to be 'only cpu bound'.
So if U're comparing x3d chips, with 4080\90 systems, with non x3d with like 2080, then ofc they will go better.

PS, this is not a synthetic test. It runs actual scenes U would play in the game by yourself. Ofc, there are scenarios which pushes CPU, or GPU more, but as for my experience it provides really equal data to what I can see in reality

 

[Seraphim]

Like i said, in overall gaming x3D chips are better than theirs base versions, but we're on VaM benchmark thread. As U can see in most of the results in here, the main bottleneck for most of the people are models physics, and this happens on single core.
In terms of VaM 1.x it doesn't matter how good CPU is in multicore [if it has at least 8 cores for other, non physics related stuff], if it sucks in single core = it will be bottlenecking overall experience.
You must also note that to be able to really compare CPU alone in VaM, both systems must run the same GPU. Physics are calculated by frame, so if GPU render them slower, CPU will go worse too.
Example:

Benchmark Result Discussion

Yes maybe, it's true that all "ms" looks better everywhere, but I'm not used to check at this, more used to rely on FPS with other benchmarks. Will take a closer look to what all these data are meaning before start to investigate if there something wrong on driver/software side. There is...

hub.virtamate.com

same cpu, using 3090, then 4090. No other changes in the system. Got almost half physics time reduce - which suppoused to be 'only cpu bound'.
So if U're comparing x3d chips, with 4080\90 systems, with non x3d with like 2080, then ofc they will go better.

PS, this is not a synthetic test. It runs actual scenes U would play in the game by yourself. Ofc, there are scenarios which pushes CPU, or GPU more, but as for my experience it provides really equal data to what I can see in reality

The gpu doesnt send the render frames to the cpu. The cpu sends the frame to the gpu. So if the CPU is running worse as the bottleneck, then it slows the gpu down. Not the other way around. So if you have a faster cpu render time with this chip, and are cpu bottlenecked, you are still running higher fps than the non vcached version. True your cpu does wait to send the frames to your gpu though, if your gpu is slower. on your 3090, you were gpu bottlenecked on baseline 3. But on you 4090, you were cpu bottlenecked. with a faster 3d vcache cpu, you timings would have been lower, and fps higher. VAM is a game, even if its a single threaded render. So as you said, 3d vcache runs games better. Therefore gettin the 3d vcache cpu is better for VAM. Especially since it is cpu physics heavy. Its very apparent o the vam benchmark table. Its already a well know fact. Its the main reason why those who know, are waiting for 7000x3d. Like me. If AMD Fs it up somehow on this itteration somehow, then thats on them.

PS: If you recently bought the 13900k and are feeling a little fomo about the new x3d. Its ok, just enjoy it. Plus im glad that synthetic test is your go to, but doesnt mean its right for everyone elses cpu. Especially if you havnt used an x3d chip.

 

[Seraphim]

I'm planning on getting a 7900x3D once they launch. I currently have the 7900x, but I've got a buddy who wants it, so, gives me the excuse

Cool. You gonna post the benchmark results for both to compare once you get it?

 

[trety]

PS: If you recently bought the 13900k and are feeling a little fomo about the new x3d. Its ok, just enjoy it. Plus im glad that synthetic test is your go to, but doesnt mean its right for everyone elses cpu. Especially if you havnt used an x3d chip.

I went thru of many pages in here and the closest i could find was 5800X with 3090, and 5800x3D with 3080 [there were one more 5800 with 3080 but these results were super low, most likely using bloated install]. And.. U might be right actually. 3D chip had slightly better physics time in baseline 3, while both had some wait time [0.01 ms for 3090, and 0.39 for 3080]

Spoiler

So x3D chips seems to do some work even in VaMs physics. U might be right that Cpu-pass tests doesn't utylize cache thing.
We will find out next month.

Spoiler: Offtopic

You're right i'm an Intel guy, and haven't used any AMD hardware for over 10 years. Basically at this point even if Intel fastest current gen chip would loose with mid-tier AMD, i would still go intel. Reason is simple, i never had broken intel system, in my history with 8 PCs so far, while both of my AMD systems ended up in trash bin - it was over 10 years ago, idk, maybe they're more solid now. Dunno. Don't wanna try. As for GPU, i had Radeon only once, and it was still branded ATI, not AMD, so can't say much about it.
U can see it clearly at 2nd hand market, intel doesn't drop prices almost at all.
Even if i have 'fastest available' machine available atm, idc how fast, and when, it's gonna get 'obsolete'. I'm just gonna use it, and certainly enjoy it, for the next 2-3 years, with no tech problems at all. My current machine is more than enough for current gen gaming, so there would be no point [except vam lol] for upgrading it, until PS6 and new Xbox would shown up.
Idc which, Intel or AMD, are winning PC master race at this stage. I just have enough seeing AMD fanbois everywhere - people gives them a way too much trust credit.
Nvidia releases 40x0 series GPUs. People over the web: Wait for AMD RDNA3!
Nvidia [people who can't even plug an cable to theirs card...] has burning power socket. People: bad nvidia, i'm scared to buy it! [it was 3 months ago and i still see people saying it]
AMD releases new GPU: losing with NVIDIA badly. People: there will be drivers updates and they will get better! okay this one is truth as far i had read -_-
but i also read that some other drivers updates disabled thermals security for some peoples cards and... literally melted them down
New GPUs are having thermal throttling [no enough liquid]. People - It's just the one batch!
And every single CPU intel is releasing there are the most famous: Wait for AMD answer!

 

[yui_3495]

I can't run official benchmark or set physic cap to 20 even I clicked apply user preferences, I only tried to run it on physic cap 3
Here is my result (1080p,R5 5600x, RTX 3070)

 

[Bob Nothing]

Cool. You gonna post the benchmark results for both to compare once you get it?

Will do!

 

[yui_3495]

I can't run official benchmark or set physic cap to 20 even I clicked apply user preferences, I only tried to run it on physic cap 3
Here is my result (1080p,R5 5600x, RTX 3070)

Doubled the FPS with my i7-13700k, RTX4080

 

[geo_gan]

Is it true this new VAM version 1.20.77.13 broke the benchmark? Pity someone didn't tell Meshed before the 1.21 patch for it went out. Probably won't be any more patches after this now.

 

[Deleted member 357]

Spoiler: "Problems detected"

...well, I definitely did those things, and all the other things required. I guess the new version did in fact break the benchmark (EDIT: upon re-reading I guess that's not the part that's broken and my problem is actually me-specific. fug).

Pity, was really wanting to fully test my new setup against others, as I guarantee something I'm doing/have setup is holding me back (aside from the middling mobo). All the more reason to setup a separate clean VaM install, but on an older version or something I guess.

New rig is an i9-13900k/RTX 4090 (MSI Liquid Suprim X) setup, running on an ASUS Prime z790-A mobo w/2x16GB 6000MHz CL30 DDR5 RAM, in a Corsair 7000X (yeah I know I'm dumb) with an Arctic Liquid Freezer II 420 AIO cooler, powered by an MSI Meg AI1300p PSU. C Drive and VaM are both on separate 1 TB Samsung 980 Pro M.2 SSDs (w/heatsinks).

 

[RandomVAMUser]

1.21 broke official run, but you can still run custom one (with physics cap 3).

-apply official UserPreferences
-run custom benchmark
-hit "ignore and continue" (note: make sure you started VaM in FullScreenWindow and not Windowed)

...or if you have 2nd "clean" 1.20.77.13 client around, use that one instead (for official run).

 

[Deleted member 357]

I just created a clean 1.20.77.9 (I was under the impression that 1.20.77.13 wasn't going to work) but still couldn't do official. I did end up doing the custom though.

Spoiler: Custom/unofficial benchmark with extra high settings

Don't know how useful it is given that it won't be the most direct comparison to everyone else running the official one. It's also on a clean install so not completely representative of my normal experience.

...also just noticed it thinks I'm running Win 10, rather than Win 11, but I'm going to assume that's something minor and/or not fixable.

 

[RandomVAMUser]

I just created a clean 1.20.77.9 (I was under the impression that 1.20.77.13 wasn't going to work) but still couldn't do official. I did end up doing the custom though.

Benchmark works fine on 1.20
If this was on clean install, maybe you "rushed" it? I can get same popup error if i try runing benchmark for the first time (after downloading it).

Usually after downloading everything I load intro screne,
always allow all plugins and restart vam (or hard reset).
Than apply settings and run official benchmark, no errors (on 1.20).

EDIT: I just noticed first post on this page (with official results) confirming it works fine on 1.20.77.13

 

[Deleted member 357]

Benchmark works fine on 1.20
If this was on clean install, maybe you "rushed" it? I can get same popup error if i try runing benchmark for the first time (after downloading it).

Usually after downloading everything I load intro screne,
always allow all plugins and restart vam (or hard reset).
Than apply settings and run official benchmark, no errors (on 1.20).

EDIT: I just noticed first post on this page (with official results) confirming it works fine on 1.20.77.13

O u rite, it def seemed to be that I needed to "wait."

So yeah, here we are, all proper now:

i9-13900k (stock settings)
RTX 4090 (MSI Liquid Suprim X, stock settings)
ASUS Prime z790-A mobo
2x16GB G-Skill Trident Z5 6000MT/s DDR5 CL30 RAM (XMP enabled, stock otherwise)
MSI Meg AI1300p PSU
Arctic Liquid Freezer II 420 AIO cooler
2x1TB Samsung 980 Pro M.2 NVMe SSDs (w/heatsinks, 1 for OS/sundries and 1 for VaM)
Corsair 7000x case

...and for the record, I don't feel too bad about the ridiculous card and all. I'm upgrading from a much older 1080/i6850k rig, so the results have been pretty spectacular.

Like, running solo females at 120Hz physics, rate cap at 3, high settings everywhere else and using subsurface scattering and both Titty/BootyMagics, and still in the 70s in VR? Absolutely baller. Toning it down I'll be able to actual enjoy male atom possession, or I can watch duo dances, or I can actually do sim hair and realistic models? It's still easy to hit the limits but the ceiling is very notably raised.

EDIT: Here's one with hyperthreading disabled (recommended by trety), showcasing slight improvements:

Spoiler: Clean install, Desktop, HT disabled

...and here are some VR benchmarks:

Spoiler: Clean install, VR

Spoiler: Clean install, VR, max settings

Spoiler: In-use/"bloated" install, VR, max settings + session plugins

 

[PermissionUsernameRequire]

I'm considering to upgrade my CPU and therefore did some benchmarking to see what I can expect from it.
Currently rocking an i7 9700K (with all core boost limit set to 4,9GHz), 3080TI (FE, stock), and 32GB DDR4 at 3300MHz 15-17-35.

Official benchmark, Desktop @ 1080p:

Official benchmark, Desktop @ 4k:

My main use case is VR. I'll have to compare my benchmark results more in depth with others, but I feel like I could gain a couple of dozen percents increase in FPS by upgraing my CPU to e.g. an i5-13600K, 7700X, or maybe the upcoming 7800X3D. Opinions welcome

 

[keycode]

I'm considering to upgrade my CPU and therefore did some benchmarking to see what I can expect from it.
Currently rocking an i7 9700K (with all core boost limit set to 4,9GHz), 3080TI (FE, stock), and 32GB DDR4 at 3300MHz 15-17-35.

Official benchmark, Desktop @ 1080p:
View attachment 207850

Official benchmark, Desktop @ 4k:
View attachment 207851

My main use case is VR. I'll have to compare my benchmark results more in depth with others, but I feel like I could gain a couple of dozen percents increase in FPS by upgraing my CPU to e.g. an i5-13600K, 7700X, or maybe the upcoming 7800X3D. Opinions welcome

i7 13700k and a fitting for your case noctua cooler or equivalent, vam will after smile to you.