FPS, PHYS rate, accuracy [answered]

[cavok84]

Firstly, thanks for the continued support with such a large update. I'm still trying to unpack all the changes made and better understand them. Hopefully,someone will be able to assist me in optimizing the simulation.
-----------------

A while back, I asked a question regarding the interrelationship of FPS, PHYS rate, Step, and accuracy-- it was explained to me why Vsync OFF would result in a more accurate simulation at the cost of utilizing more processing power.
http://universesandbox.com/forum/index.php/topic,17193.0.html

-With the update, I've noticed a few strange issues regarding the same topic.

Here is my original data pre version 20

Solar System Sim (default)
Default time step of 15 days/second

Vsync OFF: FPS- approx. 240 fps PHYS- approx 930, 1 step
Simulation Error: .05 m/s- .25 m/s
Tolerance: 11.4 million km

Vsync ON: FPS- 60     PHYS- 623, 1 step
Simulation Error:  .9 m/s- 4.0 m/s
Tolerance: same as above
-----------------------------------------------------
Update 20 data

Vsync OFF: FPS- 200fps   PHYS: 1090, 1  step
Simulation Error: .08 m/s-0.4 m/s
Tolerance:  11.3 million km- 11.6km

Note:High Performance enabled in Windows-- no noticeable change

Vsync ON: FPS:60 PHYS: 500-1000 (oscillating), 1 step
Simulation Error:  0.9-4.0 m/s
Tolerance:  11 million km - 11.8 million km

Note:when I enable "high performance" in windows power options PHYS stabilizes at approx 1140
*this never affected my sim in this way before, and the PHYS rate is double what it was before at 60fps
Now, the PHYS rate is  unaffected by Vsync, yet sim error is... ?

So, given this info:
1. Did something change with the code or logic behind n-body calcs that would explain why at 60FPS my PHYS rate is now double what it was before?

2. Any idea why the standard Windows 10 'balanced mode' (cpu is allowed to declock) would cause such fluctuations in the PHYS rate?

3. I noticed that 'sub stepping' is now an option, and it is disabled by default. I recall this automatically occurring when it was needed. Why the change?

unrelated:

4. Do the yields given when one mouses over a star with "explode" have any true effect now? (i.e., transfer a greater amount of heat to nearby bodies, greater velocity or magnetic effects? Different physical effects of a type Ia and type II?

5. Does a planet's magnetic field have any protective effects against solar winds yet, or any climate/atmospheric effects?

6. I noticed my FPS and other numbers are a bit lower with this update. Just running the basic solar system sim without making changes, has the game become a bit more demanding? Any more reliance on GPU?

Thanks in advance

cavok84

[cavok84]

bump. Any mods/engineers have any ideas?

[Jar]

Thanks for your patience, cavok84. I checked in with our physics programmer, and he's a bit confused by your data here. Not to say that your data is wrong, but that he is confused as to why this would have changed at all in recent updates, as the code has not really changed. Right now, however, Thomas (Greenleaf) is working on implementing some rewritten physics code, so it would not be productive to look into these current differences since it will all be changing soon.

I'll try to answer your other questions:

3. Disabling sub-stepping was a shortcut solution (temporary until the new physics) to fix the issue with trails becoming very polygonal. Once the new physics is implemented, we will be able to fix this issue while keeping sub-stepping enabled.

4. I believe there are some bugs with the effects of supernova right now, but it should affect at least the temperatures of nearby bodies.

5. Yes, planetary magnetic fields currently affect the erosion from solar winds (mass loss).

6. I believe our new user interface is a bit more demanding. Our new physics should introduce a small performance boost, though. And we're always working on optimizing other areas of the game as well.

[cavok84]

Thanks for the reply! All in all, great update. I know my issues aren't necessarily bugs, but figured it might point to changes which could alter the optimal configuration for sim quality and accuracy.

One last question,

Regarding the mag field of planets:

for the mag field to reduce mass loss (or atmospheric loss?) does the parent have to have a magnetic field value > 0? For example, the sun, in the default solar system sim, has a magnetic field of '0'.

[SyzygyΣE]

Thanks for the reply! All in all, great update. I know my issues aren't necessarily bugs, but figured it might point to changes which could alter the optimal configuration for sim quality and accuracy.

One last question,

Regarding the mag field of planets:

for the mag field to reduce mass loss (or atmospheric loss?) does the parent have to have a magnetic field value > 0? For example, the sun, in the default solar system sim, has a magnetic field of '0'.

I don't believe stellar magnetic fields have any effect on planets' mass loss at the moment. For the planets, however, their magnetic field should be greater than 1 times their radius to reduce the mass loss rate. Something else I find problematic is that moons are not shielded if they are within a planet's magnetic field. For example, the Galilean moons are within Jupiter's magnetosphere and are therefore should be shielded from the solar wind, but they are not. In US2, the moons have to have their own magnetic fields to be shielded, even if they are within another magnetic field.

[cavok84]

4. I believe there are some bugs with the effects of supernova right now, but it should affect at least the temperatures of nearby bodies.

5. Yes, planetary magnetic fields currently affect the erosion from solar winds (mass loss).

Thanks. I don't know why I wrote"mag field of star affecting mass loss", as primarily solar winds would be the culprit I imagine, (i.e. strong mag fields or direct EM radiation can destroy the ozone layer, but typically solar wind particles remove atmosphere)

*One thing I'm a bit confused about though, the Explode option, when applied to stars, gives a yield based on the escape velocity, but doesn't seem to be completely consistent with a supernova. For example, a white dwarf will type 1a supernova when adding hydrogen to its out layer, or you can just use the explode button-- but the explode button gives a fraction of a type 1a yield.

--If we want to see the most realistic depictions of supernovas, whether type 1a or type II, how should we go about it?

--Lastly, what affects the intensity of the solar wind? Any way to change this?
moving these last questions (with the addition of another) to a new thread since it's departed a bit from original topic