What rate do physics run at? 1000Hz? 500?

[Borsch]

Topic.
I seem to remember that AC1 runs at 500 (600?) Hz, but what about ACC? AC was awesome in every way when it came out back in 2014 and while its tyre model is still as good as it gets, its suspension/chassis just do not move as believably as some of the current competitors - iRacing or even pCars2. This is especially obvious in external view replays, but also when driving.

How is this looking in ACC right now and are there any specific plans with regards to suspension in the dev pipeline?

 

[Deleted member 197115]

Both 333hz.

 

[elloLeo Kinnunen]

external view replays

Medium chance I'm wrong about this so willing to be corrected, but I think the track that you see in external replays and when driving in AC isn't the same thing as the physics track it uses to calculate suspension travel, etc. This is so that the GPU doesn't have to render as many polygons as the physics engine computes - so in other words because the minute laser scan level details are represented physically but not graphically there is sometimes a mismatch between car and visual ground that you see in replays. Not saying this means AC is the best for suspension and chassis, just that what you see is not what you get in this case.

 

[Borsch]

Both 333hz.

Hm... Are you sure? I am quite certain that in one of his streams Stefano said that the rate was bumped up from its initial value that was around 333Hz (to around double that? I remember that it was still way below 1K) Shame if it's gone back down for ACC.

minute laser scan level details are represented physically but not graphically there is sometimes a mismatch between car and visual ground that you see in replays. Not saying this means AC is the best for suspension and chassis, just that what you see is not what you get in this case.

Good point, but difference in fidelity of suspension work is evident even on non laserscanned tracks - the number of suspension oscillations and chassis movements per each curb riding, brake press, steering input etc. AC is visibly and feelably restricted compared to the sims above.

 

[henrisal]

Good point, but difference in fidelity of suspension work is evident even on non laserscanned tracks - the number of suspension oscillations and chassis movements per each curb riding, brake press, steering input etc. AC is visibly and feelably restricted compared to the sims above.

If you watch AC replay it's maximum 64fps in "ultra" quality. That alone cuts away some movement. Also all the suspension movement needs animation and AC has some limitations with that. There is things happening even if you can see it, like tyre flexing is there but it's not visible. If for example iRacing would lower physics engine from 360hz to 333hz I'm pretty sure nobody would notice anything.

 

[Borsch]

Fps (frames per second) are a different thing to physics rates. One simple example is a video of a real life car: it could be below 60fps, but its motion is still unmistakably real.

Doesn't iRacing run at 1000Hz howadays? I know it used to be 360Hz ages ago. It's suspension kinematics have noticeably more "resolution" compared to AC. Same with Pcars2. (This LFSForums thread is ancient old, but still interesting, apparently Live For Speed runs vehicle dynamics at 2000Hz, adjusting up for lower speeds and dropping down for higher speeds. Same with Rigs of rods/BeamNG. https://www.lfs.net/forum/thread/48927-Racing-sim-physics-engine-rates )

In AC I'm missing the little quick shudders of the chassis when exiting drift and so on, like at 2:25

 

[MeNoMoney]

Doesn't iRacing run at 1000Hz howadays? I know it used to be 360Hz ages ago.

IRacing FFB runs at just 60Hz. Using another program called irffb, it can be upped to 360hz.
If you have an osw Direct Drive setup and running it with the Simucube Tool, you can boost the ffb hz signal to a much higher rate.

 

[Borsch]

I'm talking about vehicle dynamics, not ffb.

 

[Case_]

https://www.iracing.com/physics-modeling-ntm-v7-info-plus/

11) What is the sampling rate of your physics model?

360 Hz, is the short answer. But we calculate forces twice per update step, so we do player tire force calculations at least 4x2x360 = 2880 times per second. More if some of the tires are contacting multiple surfaces (i.e. curbs). We use IEEE floats.

 

[Borsch]

Excellent answer, many thanks - I will need to update my sources iRacing is not 1000Hz then... Although " But we calculate forces twice per update step" is strange too, as effectively it makes the update rate 720Hz? How many times does AC/ACC update per step? Is there any difference between AC and ACC?

I understand that there is no simple silver bullet to physics, running a crude approximation at higher rate may not give a better result than a finer model run at lower rate...
My initial opinion " AC1 = better tyre, iRacing/pcars2 = better suspension" still persists, but simple physics rate comparison may not be the answer. Basically, suspension dynamics was the biggest issue in AC for me and I wonder if there's been changes in that regard in ACC. I'll need to watch some external views replays on youtube, but also curious to hear about first hand experience.