Other CPU Performance Patch (Up to 30% faster physics, up to 60% more FPS)

[cbkstudios]

I just tried with my intel 12700H and 3070Ti Mobile, I have no difference. I think I set it wrong. I left 1,3,5,... because I didn't disable HT. I also left engine affinity. Can someone guide me?

Original :
View attachment 403310

Patched V13beta1
View attachment 403311

settings :
[threads]
computeColliders=4
skinmeshPart=4
skinmeshPartMaxPerChar=4
applyMorphs=4
applyMorphMaxPerChar=4
#affinity=1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16
affinity=1,3,5,7,9,11
#engineAffinity=1,3

[threadsVR]
computeColliders=4
skinmeshPart=4
skinmeshPartMaxPerChar=4
applyMorphs=4
applyMorphMaxPerChar=4
affinity=1,3,5,7,9,11
#engineAffinity=1,3
[profiler]
enabled=0

And the boot.config file :
gfx-enable-gfx-jobs=1
gfx-enable-native-gfx-jobs=1
gfx-disable-mt-rendering=0
wait-for-native-debugger=0
gc-max-time-slice=3
job-worker-count=12
single-instance=1

The command : Get-Process | select VaM.exe -expand Modules -ea 0 | where {$_.ModuleName -like 'skinmeshpartdll.dll'} :
result : 1136 SkinMeshPartDLL.dll

I've been messing with this for the past 4 hours. Here is the setup I tested it on:

Gigabyte Aero 16XE5
i7-12700H
RTX 3070 Ti Laptop GPU
32GB DDR5 RAM (4800)

The i7-12700H has 6 Performance Cores and 8 Efficiency (total 14). The Performance cores are Hyperthreaded and I'm pretty sure it is not disabled due to the task manager showing 14 cores but 20 logical processors. According to all of what I've read in this thread the idea is to only use the P cores and skip every other thread. However, with about 2 hours of testing, the best setup I found was this. Anything less than using 12 resulted in less performance increase.

[threads]
computeColliders=12
skinmeshPart=12
skinmeshPartMaxPerChar=12
applyMorphs=12
applyMorphMaxPerChar=6
#affinity=1,2,3,4,5,6,7,8,9,10,11,12
affinity=1,2,3,4,5,6,7,8,9,10,11,12
engineAffinity=1,2,3,4

[threadsVR]
computeColliders=12
skinmeshPart=12
skinmeshPartMaxPerChar=12
applyMorphs=12
applyMorphMaxPerChar=6
#affinity=1,2,3,4,5,6,7,8,9,10,11,12
affinity=1,2,3,4,5,6,7,8,9,10,11,12
engineAffinity=1,2,3,4

[profiler]
enabled=0

I have minimal FPS increase however, the "wait" time is pegged at 0 and the FPS are much more stable. I think that is the best I can get from this at the moment but, an improvement is an improvement. I've not tried that rig's setup in VR yet though to see if there is any change. The scene I tested it on was a work in progress scene I have with a bunch of complex CUAs and 2 people. One of them wearing 4 pieces of SIM clothing. Closer I get to a figures the FPS drops to around 60fps. When I'm further away from a figure it is upwards of 100FPS. However, just reducing the hair physics bumped both near and far up by about 20FPS each. Physics update settings are set at "auto".

Additionally, I ran a test at 18 cores and using all of them. Task manager showed usage but, there was no improvement beyond what I had gotten with just using P cores. Give this setup a shot and let me know if it does anything for you.


***Update***
I removed (1) piece of SIM clothing, cut hair physics in half (not much of noticeable difference), and removed (2) of the CUAs (that were animated - there is some sort of render issues with them causing frames to drag). Consistent 120-130 FPS on desktop and 60-90 FPS in VR. I've never had 90 FPS in VR before on that laptop. For VR I was using a Quest 2 with virtual desktop and hard wired with link cable to laptop. @Makimax - try the settings I mentioned and let me know what it does for you.

 

[Makimax]

I've been messing with this for the past 4 hours. Here is the setup I tested it on:

Gigabyte Aero 16XE5
i7-12700H
RTX 3070 Ti Laptop GPU
32GB DDR5 RAM (4800)

The i7-12700H has 6 Performance Cores and 8 Efficiency (total 14). The Performance cores are Hyperthreaded and I'm pretty sure it is not disabled due to the task manager showing 14 cores but 20 logical processors. According to all of what I've read in this thread the idea is to only use the P cores and skip every other thread. However, with about 2 hours of testing, the best setup I found was this. Anything less than using 12 resulted in less performance increase.

[threads]
computeColliders=12
skinmeshPart=12
skinmeshPartMaxPerChar=12
applyMorphs=12
applyMorphMaxPerChar=6
#affinity=1,2,3,4,5,6,7,8,9,10,11,12
affinity=1,2,3,4,5,6,7,8,9,10,11,12
engineAffinity=1,2,3,4

[threadsVR]
computeColliders=12
skinmeshPart=12
skinmeshPartMaxPerChar=12
applyMorphs=12
applyMorphMaxPerChar=6
#affinity=1,2,3,4,5,6,7,8,9,10,11,12
affinity=1,2,3,4,5,6,7,8,9,10,11,12
engineAffinity=1,2,3,4

[profiler]
enabled=0

I have minimal FPS increase however, the "wait" time is pegged at 0 and the FPS are much more stable. I think that is the best I can get from this at the moment but, an improvement is an improvement. I've not tried that rig's setup in VR yet though to see if there is any change. The scene I tested it on was a work in progress scene I have with a bunch of complex CUAs and 2 people. One of them wearing 4 pieces of SIM clothing. Closer I get to a figures the FPS drops to around 60fps. When I'm further away from a figure it is upwards of 100FPS. However, just reducing the hair physics bumped both near and far up by about 20FPS each. Physics update settings are set at "auto".

Additionally, I ran a test at 18 cores and using all of them. Task manager showed usage but, there was no improvement beyond what I had gotten with just using P cores. Give this setup a shot and let me know if it does anything for you.


***Update***
I removed (1) piece of SIM clothing, cut hair physics in half (not much of noticeable difference), and removed (2) of the CUAs (that were animated - there is some sort of render issues with them causing frames to drag). Consistent 120-130 FPS on desktop and 60-90 FPS in VR. I've never had 90 FPS in VR before on that laptop. For VR I was using a Quest 2 with virtual desktop and hard wired with link cable to laptop. @Makimax - try the settings I mentioned and let me know what it does for you.

Hello @cbkstudios. Thanks for your tests. I just tried and it still doesn't change anything for me. I've been trying for over 2 days without success.

My Lenovo Legion 5i pro gen7 12700H / 3070Ti 8G in 150 Watt, 32G RAM has profiles (eco/auto/performance), I don't know if this has an impact, I put it in performance.I updated the motherboard firmware, and the Nvidia driver to the latest versions.

"Physics update settings are set to "auto". How to reduce hair physics? Do I have to use the GiveMeFPS plugin? I put myself in the same conditions anyway to see if there is a gain or not.

Does the way VAM is started have an impact?
- in VR, I put my Quest3 in wireless, wifi6, with Virtual Desktop, then right click on VD in the W11 taskbar then I search for the executable. Steam is not launched at all.
- in Desktop, NO VR, I simply double click on vam.exe and it launches SteamVR which I immediately turn off. I tried to go through the updater then Desktop.

How is your boot.config file?
Do you know if it is possible to do only Timeline 3 and not the whole benchmark, I had to try more than 30 times with no difference.

Maybe the Patch is not made for laptops. Can @turtlebackgoofy confirm this for us?

Before

Patched with your settings (12), I haven't tried with 18.

 

[cbkstudios]

Hello @cbkstudios. Thanks for your tests. I just tried and it still doesn't change anything for me. I've been trying for over 2 days without success.

My Lenovo Legion 5i pro gen7 12700H / 3070Ti 8G in 150 Watt, 32G RAM has profiles (eco/auto/performance), I don't know if this has an impact, I put it in performance.I updated the motherboard firmware, and the Nvidia driver to the latest versions.

"Physics update settings are set to "auto". How to reduce hair physics? Do I have to use the GiveMeFPS plugin? I put myself in the same conditions anyway to see if there is a gain or not.

Does the way VAM is started have an impact?
- in VR, I put my Quest3 in wireless, wifi6, with Virtual Desktop, then right click on VD in the W11 taskbar then I search for the executable. Steam is not launched at all.
- in Desktop, NO VR, I simply double click on vam.exe and it launches SteamVR which I immediately turn off. I tried to go through the updater then Desktop.

How is your boot.config file?
Do you know if it is possible to do only Timeline 3 and not the whole benchmark, I had to try more than 30 times with no difference.

Maybe the Patch is not made for laptops. Can @turtlebackgoofy confirm this for us?

Before
View attachment 403680

Patched with your settings (12), I haven't tried with 18.
View attachment 403681

So, unfortunately all of this is a bit more complicated than pushing a button. For example, your resolution is 2560x1600. The ASUS I'm playing with has a native resolution of 3840x2400. That alone drags the ASUS FPS down vs your setup when I run it in desktop mode. Not to mention that not all CPUs and GPUs are created equal. Some are just bad quality, some win the silicone lottery during manufacturing. Having a high performance laptop means you can do cool shit anywhere you want on it vs needing to sit at a desk. The drawback is performance you get from the laptop. High end laptops just can't compete with high end desktops. Most of it comes down to power source. I think max I've seen for laptops is like 140w when high end PCs often start at 650w power supplies. Everything is just smaller and heat has to be taken into account.

Additionally, for VAM, depending upon what scenes you use will drastically affect your FPS. Are you running one character, multiple, SIM clothing, hair physics to the max? I've found even using certain CUAs will destroy the FPS in a scene. Best thing to do is find a typical scene you like and run your own tests.

I found that the first test scene I was using gave me drastically different results vs another scene. The original test scene I was using was 2 characters and not a ton of physics. I found another scene posted on the hub by a user that has 3 characters that had a pretty repetitive physics animation. I re-ran everything and found that the following is what worked best on that scene vs the other settings I had posted.

[threads]
computeColliders=6
skinmeshPart=6
skinmeshPartMaxPerChar=6
applyMorphs=6
applyMorphMaxPerChar=6
affinity=1,3,5,7,9,11
engineAffinity=1,3

[threadsVR]
computeColliders=6
skinmeshPart=6
skinmeshPartMaxPerChar=6
applyMorphs=6
applyMorphMaxPerChar=6
affinity=1,3,5,7,9,11
engineAffinity=1,3

enabled=0

Laptop Performance Mode
Your laptop should always be in "performance mode" for any VAM stuff. The performance mode allows for an increase in the input power on these gaming laptops. This increased power is needed for better performance.

Reducing Hair Physics in a scene
- go into edit mode
- select a character that is wearing hair
- click on the hair tab
- click on the physics tab under the menu that opened
- there are (2) categories that say (eats GPUs)
- typically, you can reduce those 2 settings in half without much of an impact to quality. Just play with the settings to get them as low as you can and still look good
Note - you do not need the GiveMeFPS plugin. You just do it manually, where the plugin does this automatically.

Launching VAM in Desktop
- When you open up the VAM_Updater folder scroll down until you see a .bat file called VaM (Desktop Mode)
- Click this file and VAM will load in desktop mode and skip trying to open in VR.

Launching VAM in VR
- The way I launch VAM in VR is I connect the Quest 2 to the computer.
- I make sure that the virtual desktop is enabled on my computer (it is by default).
- I put my headset on and open the virtual desktop app in the Oculus apps
- Virtual Desktop starts up and I open VAM through the virtual desktop

My boot.config file looks like this:

gfx-enable-gfx-jobs=1
gfx-enable-native-gfx-jobs=1
gfx-disable-mt-rendering=0
wait-for-native-debugger=0
gc-max-time-slice=3
job-worker-count=12
single-instance=1


Overall, the results you posted even the "before" are better than what I was able to get on the ASUS laptop (after I matched my resolution to yours) before the patch AND after the patch. That being said, you are probably running the best you can at this point. I tried running the benchmark program but, the file must be corrupt because it closes half way through no matter what settings I use. So, I'm unable to post results but, the 5 scenes it ran through were showing about 10-12 FPS lower than what your scores are.

I believe the most important thing with performance laptops is heat. If your CPU or GPU gets too hot they will throttle or decrease their performance to make sure the laptop doesn't overheat. If you start to encounter this, get a nice laptop cooler to sit your laptop on. If you want to see if your CPU is being throttled go to your windows search bar and open Resource Monitor. It will show you the CPU usage in green and the maximum frequency in blue. Your maximum frequency should always be 100%. If it drops below 100% that means your CPU is decreasing performance because it is likely too hot.

I'm not a programmer or computer tech but, like to learn and research stuff to teach myself. Hope some of this was helpful.

If you want to benchmark your computer as a whole to see how it compares to others with your same setup try this benchmark program:

UserBenchmark: PC Speed Test Tool - Compare Your PC

This will also show you that not all PC hardware is created equal. I ran this on the ASUS and the GPU in this ranks in the top 10% of other PCs with this same GPU. The CPU ranks slightly below the average mark.

 

[Da_Pig]

Hello @cbkstudios. Thanks for your tests. I just tried and it still doesn't change anything for me. I've been trying for over 2 days without success.

My Lenovo Legion 5i pro gen7 12700H / 3070Ti 8G in 150 Watt, 32G RAM has profiles (eco/auto/performance), I don't know if this has an impact, I put it in performance.I updated the motherboard firmware, and the Nvidia driver to the latest versions.

"Physics update settings are set to "auto". How to reduce hair physics? Do I have to use the GiveMeFPS plugin? I put myself in the same conditions anyway to see if there is a gain or not.

Does the way VAM is started have an impact?
- in VR, I put my Quest3 in wireless, wifi6, with Virtual Desktop, then right click on VD in the W11 taskbar then I search for the executable. Steam is not launched at all.
- in Desktop, NO VR, I simply double click on vam.exe and it launches SteamVR which I immediately turn off. I tried to go through the updater then Desktop.

How is your boot.config file?
Do you know if it is possible to do only Timeline 3 and not the whole benchmark, I had to try more than 30 times with no difference.

Maybe the Patch is not made for laptops. Can @turtlebackgoofy confirm this for us?

Hey Makimak.

From what I see, the patch HAS made a difference, as all the processing times are down, and so is your total time (ms) which is their sum. What I don't know is why that hasn't translated in better FPS as it should. Possible reasons: Vsync or some other frame limiter on.

AFAIK there is a version of the benchmark that only includes baseline 3, but can't find it right now - I think the link was in the Benchmark results discussion

 

[cbkstudios]

Hey Makimak.

From what I see, the patch HAS made a difference, as all the processing times are down, and so is your total time (ms) which is their sum. What I don't know is why that hasn't translated in better FPS as it should. Possible reasons: Vsync or some other frame limiter on.

AFAIK there is a version of the benchmark that only includes baseline 3, but can't find it right now - I think the link was in the Benchmark results discussion

Good point. I noticed that and was thinking it would at least make his FPS more stable and have less of a gap with the min and max but, that doesn't appear to be true either.

I've also noticed that in some scenes putting MSAA to x8 increases FPS for me in both desktop and vr. While other scenes putting it down to x2 greatly increases the FPS.

Bottom line, mess with settings for each scene until you get what you want. Wish there was an easy button for this shit...

 

[Makimax]

Good point. I noticed that and was thinking it would at least make his FPS more stable and have less of a gap with the min and max but, that doesn't appear to be true either.

I've also noticed that in some scenes putting MSAA to x8 increases FPS for me in both desktop and vr. While other scenes putting it down to x2 greatly increases the FPS.

Bottom line, mess with settings for each scene until you get what you want. Wish there was an easy button for this shit...

My laptop has a 150W version of the 3070Ti Mobile (Alimentation 300W), I know that there are versions that are a little more limited in consumption, this may be the reason for your slightly lower FPS. I sometimes have big image freezes when working a lot, it's very unpleasant and it ruins the experience. You have to simplify and then it's better but it's a shame.

For the Benchmark results with Patch, the processing times have all decreased but strangely the FPS has not changed. I set the general vsynch of the nvidia panel to Fast. It's strange.

I tried the hair trick and it improves the FPS, reduce the AntiAliasing too, the result remains acceptable but the VR experience is so much better when everything is higher and there is no limit. In VR the defects are much more visible.

With the patch, the gains are so huge for some Desktop users, it's a shame that we don't have such gains but it is understandable by the reduced space of the laptop and the already high heat even without the patch. So I'll make do with what I have and I'll probably get a desktop PC later because it's a big budget.

For the Benchmark I did 2 different clean installations, I just added some vars of PrimeMocap and CuddleMocap in addition to the benchmark. The two installations are identical one with patch, the other without to compare the same thing.

We have 6 core performaces, some modify the parameters
skinmeshPart,
skinmeshPartMaxPerChar
applyMorphs
applyMorphMaxPerChar

And for some the configuration [threads] and [threadsVR] is different. To see if it is a way to improve. I don't believe it anymore. I can only dream of the next generation RTX5xxx Desktop.

Last bench with your last settings (no change)

 

[mostvanvege]

Good point. I noticed that and was thinking it would at least make his FPS more stable and have less of a gap with the min and max but, that doesn't appear to be true either.

I've also noticed that in some scenes putting MSAA to x8 increases FPS for me in both desktop and vr. While other scenes putting it down to x2 greatly increases the FPS.

Bottom line, mess with settings for each scene until you get what you want. Wish there was an easy button for this shit...

Did you try the “givemefps” plugin? Thats quite handy to gather the performance relevant settings in one place. It even has recommended tier levels to speed your hw up. CPU is bogged down mostly by butt physics, GPU is by many things but mostly lights or unoptimized cloth models. If you rearrange a bit your favourite scene you most likely can find additional perfomance. VAM is a sandbox and as such by nature it can hit the limits of any hardware. You need the mindset of a game developer a bit, try find the balance with details vs realities.

 

[henshin]

I want to try this but I'm completely lost how to configure it.
Is there any guide?

We have 2 config files right?
Can somebody explain what is or how to configure every option?


boot.config
----------------------------------------------
gfx-enable-native-gfx-jobs=
wait-for-native-debugger=0

//
I have that, but it seems there are more options available for that file (I have no idea what they do):

gfx-enable-gfx-jobs=1
gfx-enable-native-gfx-jobs=1
gfx-disable-mt-rendering=0
wait-for-native-debugger=0
gx-max-time-slice=3
job-worker-count=12
single-instance=1

//


SkinMeshPartDLL.ini (default settings)
----------------------------------------------
[threads]
computeColliders=6
skinmeshPart=6
skinmeshPartMaxPerChar=6
applyMorphs=6
applyMorphMaxPerChar=6
#affinity=1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16
affinity=1,3,5,7,9,11,13,15
engineAffinity=1,3

[threadsVR]
computeColliders=6
skinmeshPart=6
skinmeshPartMaxPerChar=6
applyMorphs=6
applyMorphMaxPerChar=6
affinity=1,3,5,7,9,11,13,15
engineAffinity=1,3
[profiler]
enabled=0


----------------------------------------------------

I'm using a laptop:
CPU: AMD Ryzen 9 8945HS
GPU: NVIDIA GeForce RTX 4070

 

[turtlebackgoofy]

Porting mono to the newest runtime is kind of a shitshow and many plugins break, I think I'm just gonna wait for vam2 beta and then ask @meshedvr if I should commit improvements I find to the source code. Unity 2022.3 should give a lot more room for optimization, especialy openxr and the burst system.

• (Character Model System) Completed GPU acceleration of skinning engine which enables high performance on high poly-count models.

• (Character Model System) Added lightweight blend shape (morph) system which removes calculation of normal and tangents for improved performance and memory usage. Must be used in conjunction with the normal/tangent recalculation system.

This will be a game changer tbh. Unity 2022.3 should also have better physx calculations.

 

[Seraphim]

I want to try this but I'm completely lost how to configure it.
Is there any guide?

We have 2 config files right?
Can somebody explain what is or how to configure every option?


boot.config
----------------------------------------------
gfx-enable-native-gfx-jobs=
wait-for-native-debugger=0

//
I have that, but it seems there are more options available for that file (I have no idea what they do):

gfx-enable-gfx-jobs=1
gfx-enable-native-gfx-jobs=1
gfx-disable-mt-rendering=0
wait-for-native-debugger=0
gx-max-time-slice=3
job-worker-count=12
single-instance=1

//


SkinMeshPartDLL.ini (default settings)
----------------------------------------------
[threads]
computeColliders=6
skinmeshPart=6
skinmeshPartMaxPerChar=6
applyMorphs=6
applyMorphMaxPerChar=6
#affinity=1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16
affinity=1,3,5,7,9,11,13,15
engineAffinity=1,3

[threadsVR]
computeColliders=6
skinmeshPart=6
skinmeshPartMaxPerChar=6
applyMorphs=6
applyMorphMaxPerChar=6
affinity=1,3,5,7,9,11,13,15
engineAffinity=1,3
[profiler]
enabled=0


----------------------------------------------------

I'm using a laptop:
CPU: AMD Ryzen 9 8945HS
GPU: NVIDIA GeForce RTX 4070

8945HS

[threads]
computeColliders=8
skinmeshPart=8
skinmeshPartMaxPerChar=8
applyMorphs=8
applyMorphMaxPerChar=8
#affinity=1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16
affinity=1,3,5,7,9,11,13,15

[threadsVR]
computeColliders=8
skinmeshPart=8
skinmeshPartMaxPerChar=8
applyMorphs=8
applyMorphMaxPerChar=8
affinity=1,3,5,7,9,11,13,15
[profiler]
enabled=0

 

[KyloMan34]

Porting mono to the newest runtime is kind of a shitshow and many plugins break, I think I'm just gonna wait for vam2 beta and then ask @meshedvr if I should commit improvements I find to the source code. Unity 2022.3 should give a lot more room for optimization, especialy openxr and the burst system.




This will be a game changer tbh. Unity 2022.3 should also have better physx calculations.

so, version 13b1 will be the last or there will be another that improves performance just a little bit?

 

[turtlebackgoofy]

so, version 13b1 will be the last or there will be another that improves performance just a little bit?

well I manged to improve performance in a 5 character scene from 80fps to 200fps, but it only helps ryzen cpus with 32 threads and 2 CCDs like the 5950x. The trick was to park all unity threads to CCD1 while putting all character related threads to CCD2 to use more threads without the inter-CCD cache penalty. This improvement is only noticeable in a scene with lots of characters that need to be skinned and morphed.
Just wait half a year and I will give you more performance in the vam2 beta. No reason to further improve soon to be outdated software.

 

[henshin]

8945HS

[threads]
computeColliders=8
skinmeshPart=8
skinmeshPartMaxPerChar=8
applyMorphs=8
applyMorphMaxPerChar=8
#affinity=1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16
affinity=1,3,5,7,9,11,13,15

[threadsVR]
computeColliders=8
skinmeshPart=8
skinmeshPartMaxPerChar=8
applyMorphs=8
applyMorphMaxPerChar=8
affinity=1,3,5,7,9,11,13,15
[profiler]
enabled=0

Thank you so much !!!
I need to test it further, but I think is giving me 10 or 15% extra fps

I'm using this boot.config, is this ok?

gfx-enable-gfx-jobs=1
gfx-enable-native-gfx-jobs=1
gfx-disable-mt-rendering=0
wait-for-native-debugger=0
gx-max-time-slice=3
job-worker-count=12
single-instance=1

 

[turtlebackgoofy]

Thank you so much !!!
I need to test it further, but I think is giving me 10 or 15% extra fps

I'm using this boot.config, is this ok?

gfx-enable-gfx-jobs=1
gfx-enable-native-gfx-jobs=1
gfx-disable-mt-rendering=0
wait-for-native-debugger=0
gx-max-time-slice=3
job-worker-count=12
single-instance=1

its

gc-max-time-slice

not

gx-max-time-slice

otherwise looks fine, no idea what single-instance means though

 

[Makimax]

Good point. I noticed that and was thinking it would at least make his FPS more stable and have less of a gap with the min and max but, that doesn't appear to be true either.

I've also noticed that in some scenes putting MSAA to x8 increases FPS for me in both desktop and vr. While other scenes putting it down to x2 greatly increases the FPS.

Bottom line, mess with settings for each scene until you get what you want. Wish there was an easy button for this shit...

Hi cbkstudios (and turtlebackgoofy ^^).
Do you find another setup for our i7 12700H mobile. total time (ms) is better for me but i don't know is why that hasn't translated in better FPS as it should.

 

[cbkstudios]

Hi cbkstudios (and turtlebackgoofy ^^).
Do you find another setup for our i7 12700H mobile. total time (ms) is better for me but i don't know is why that hasn't translated in better FPS as it should.

Hi Makimax,

I've not tried any more settings or tweaks with this since my posts. I spent a significant amount of time tinkering with the settings and did not see a benefit in continuing to tinker.

You may be getting the best performance you can at this point, on your laptop, if you're not seeing any more improvements.

 

[Seraphim]

its

not


otherwise looks fine, no idea what single-instance means though

Have you gotten more performance using the extra hyper-threading threads as well, since you mentioned 32 threads? I thought we were just to use every other thread (main core threads).

 

[Holyplatypus7]

Hi anyone has the settings for

Alienware laptop x17 R1
11th Gen Intel 8 Core i7-11800H @ 2.30GHz
3080 Laptop GPU
32 RAM

 

[turtlebackgoofy]

Have you gotten more performance using the extra hyper-threading threads as well, since you mentioned 32 threads? I thought we were just to use every other thread (main core threads).

it depends on the scene, skin meshing can benefit from hyperthreads if there are many characters to skin. For example a scene with 5 characters can run 3 threads each on 15 threads from which 7 are hyperthreads. This is faster than running 5 characters on 8 non-hyperthreads.

I think I will try to rewrite UpdateJoints natively, since there are places in the code that might be multithreadable. This could improve single thread performance. But this would be really the last feasable improvement for vam1.

 

[gghhoosstt123]

Wish there was a way to load a person atom much faster, something like jayjaywon person cache plugin but instead of 1 person it can be 2. i actually don't mind waiting for game to load if the persons afterwards can be loaded faster.

 

[turtlebackgoofy]

Wish there was a way to load a person atom much faster, something like jayjaywon person cache plugin but instead of 1 person it can be 2. i actually don't mind waiting for game to load if the persons afterwards can be loaded faster.

the problem is that the jsons are huge and they are loaded bit by bit from a zipstream instead of unzipping them once and then loading them in one go. Unzipping the the .var first and then loading atoms would increase loading speeds 100x on a ssd, but would make every scene require a few gigs of ram at loading time.

 

[gghhoosstt123]

the problem is that the jsons are huge and they are loaded bit by bit from a zipstream instead of unzipping them once and then loading them in one go. Unzipping the the .var first and then loading atoms would increase loading speeds 100x on a ssd, but would make every scene require a few gigs of ram at loading time.

So i guess vam 2 most likely will have much faster time in terms of loading a person atoms? or will it be kinda similar to vam 1?

 

[turtlebackgoofy]

So i guess vam 2 most likely will have much faster time in terms of loading a person atoms? or will it be kinda similar to vam 1?

no idea, havent seen vam2 code yet

 

[Makimax]

This may not be the topic but morphs have a significant impact on performance, do clothing packs have one too? I don't quite understand the difference between morphs, looks and appearances.
Will I lose performance if I have a lot of vars only in clothes ?
And if I only have looks (if I understood correctly it's the person with the static scene)?
Are there morphs in the vars of looks? If so, you have to delete all the looks you don't like.
Thank you for your help

 

[StyleMavin]

I had some success with this on my i9 10900K @3.7 GHz with an Nvidia 3080. The breakthrough was in VR where I had some scenes that just killed frame rate. My cowgirl scene with Naturalis v87 (paid version) looks amazing in desktop with jigglng ass cheeks and bouncy titties. I don't know the exact frame rates but they were something like 40fps. In VR, it was a dud because the frame rate dropped to 6-8 fps and was unplayable. If I turned the mirror off, the frame rate would go up to something like 20fps, but then the right eye view would glitch after a short time and become unplayable. Another BJ scene I have would just run at 14-18fps in VR with some models l like, so it was only sort of playable. The frame rate determines the refresh rate for an SR6 sycned with ToySerialController, so low frame rates make the motion less smooth.

What I figured out was if I used Intel Extreme Tuning Utility and just turned on the Speed Optimizer, which automatically overclocks the computer, the frame rate went way up.

I also have hyper threading disabled, so that is key as well.

All of a sudden, the mirror on or off makes no difference in the cowgirl scene, and the frame rate about 35fps. In desktop the it was over 100! The BJ scene was playing at about 35fps in VR, too. By removing the model's hair, I could get 72fps in VR. The motion of the SR6 in the BJ scene is really good now. Here are the settings I used.

[threads]
computeColliders=6
skinmeshPart=6
skinmeshPartMaxPerChar=6
applyMorphs=6
applyMorphMaxPerChar=6
#affinity=1,2,3,4,5,6,7,8,9,10,11,12,13,14,15
affinity=1,3,5,7,9,11,13,15,17
engineAffinity=1,2

[threadsVR]
computeColliders=8
skinmeshPart=8
skinmeshPartMaxPerChar=8
applyMorphs=8
applyMorphMaxPerChar=8
affinity=1,3,5,7,9,11,13,15
engineAffinity=1,2
[profiler]
enabled=0

boot config:

gfx-enable-gfx-jobs=1
gfx-enable-native-gfx-jobs=1
gfx-disable-mt-rendering=0
wait-for-native-debugger=0
gc-max-time-slice=3
job-worker-count=12
single-instance=1