Help Us Test the New Alpha 19 x5 Preview | Aug. 17, 2016

[Jar]

Thank you for your patience while we get Alpha 19 in top shape. We've hit a few snags in development, but it's now looking better than ever, and we're very excited to let everyone try it out.

We'd appreciate it if you could try out the new preview version of Alpha 19 and let us know what you think.

Once you've opted in, please follow the start-up instructions and run the performance test.

To opt in to the Alpha 19 preview:

• Right-click on the game title in your Steam Library
• Click on ‘Properties’
• Select the ‘Betas’ tab
• Set the dropdown menu to 'alpha19preview'
• Close the Properties window
• Steam will now update Universe Sandbox ² to the preview version
• Once updated, launch Universe Sandbox ²

New or improved in this preview build:

• Improved tidal heating & Roche fragmentation
• Improved explosion effects
• New fragment system for cold/hot, rocky/gaseous fragments
• Improved volatiles
• Many, many bug fixes

Check out the new tidal forces and explosions effects:
Home > Open > Physics tab
Home > Open > Explosions tab


Here's a video from Thomas, our physics programmer, showcasing the new effects:
https://youtu.be/OXUtZPKaDek

Thanks for testing and for your patience!

[Quizer9O8]

The long wait was it totally worth it after seeing some footage
Cant wait to see this game bloom into its full potential.

[Prometech]

*Explodes*

[Prometech]

What exactly are these effects?

[kallisti]

First, I'm somewhat surprised by the lack of response to the new preview, both here and on Steam. But oh well.

I was looking forward to this, and the video looks great. However, I experience a crippling loss in performance using the new preview. I run Win7 64bit on a three-year-old ASUS gaming laptop with an Intel i7 CPU and GeForce 765M GPU; both Windows and GPU updated with the latest updates/drivers. While perhaps not optimally set up, Alpha 18 yields ~47 fps (measured with the in-game fps counter) in the standard solar system simulation, which is smooth enough. Less content-heavy sims run much smoother, with only sproadic short fps drops during multiple collisions for example.

However, the new preview A) yields ~27 fps officially (which would be tolerable, and understandable due to the updated graphics) but B) only yields very few effective fps in the simulation of the planets' orbit. The in-game performance test showed an average of 67 fps last night, but both the test simulations and everything else points to some kind of problem displaying it properly.

Here's a video showing the result (it's unlisted so can only be accessed with this link): https://youtu.be/KdF3xQaisxs

Any input is most appreciated.

Edit: For comparison, here's how Alpha 18 runs the same sim: https://youtu.be/WYHqrvXhJSc

[Prometech]

I have noticed that this new version is continuously using 100% of my cpu even when simulating an empty window, so whats the deal? It doesn't affect my performance but it does use up system resources I often need for recording in-game.

[kallisti]

Mine also uses 100% of CPU when loading just about any sim in the new preview. Only 25% for an emtpy new sim though. Retested with Bombardment of moon; less than 60 official fps with Preview 19 (however it only displays 3-5 fps in reality), and 100% load.

[Prometech]

Seems like we are again going backwards on performance. The new Fragments kill the Simulation speed,

[kallisti]

Bombardment of moon

https://youtu.be/6JLmap1Al_I

https://youtu.be/kQ0Jj_53API

[Greenleaf]

I have noticed that this new version is continuously using 100% of my cpu even when simulating an empty window, so whats the deal? It doesn't affect my performance but it does use up system resources I often need for recording in-game.

We have identified an issue with the parallelization, which seems to be causing this. On most systems, performance is not actually affected by the reported 100% utilization (which is why it slipped under the radar), but on some systems it does negatively impact performance.


In addition to that, performance is generally down from before, which is because of a lot of new tech. Work is underway to bring it back up. Generally we add new things and tune, release it, and then tune it some more.

[Prometech]

@Greenleaf

Also Seeing an issue with fragments disappearing too early after a collision. This seems to be caused by the fragment budget which aggressively scales back the simulation to improve the frame rate. could we please see a way to disable the budget? When I do my photography I don't need 60 FPS, only an image that is appealing.
For fragments to disappear so quickly often ruins my chances to photograph the epic collisions that could be.

[Dan Dixon]

Thanks to kallisti & Prometech for the detailed report on the performance drop (and the cpu utilization issue).
(and please don't delete the videos, keeping them around is useful to us)

And to Prometech for their thoughtful critique on the need to keep particles around longer if desired.

[kallisti]

Thanks for the response! The previous optimalization update REALLY improved the performance overall on my computer, so I have hopes the current issue will be resolved. Keep up the good work!

[Prometech]

CPU usage

EDIT: I found a way around the Dynamic Budget. I just set it to zero. Tada problem solve.

EDIT 2: Doesn't seem that this brought back quite what I wanted. There is an issue where upon impact of a large object with another object, say the Earth and Moon, 2-3 Huge fragments form and end up "Absorbing" the fragments around them resulting in an unsatisfactory impact effect.

[MD2903]

I downloaded the alpha 19 preview. When i tried the performance test, my screen went black and my cursor disappeared. When running a simulation like the solar system, I'm not able to click objects and my cursor occasionaly disappears.

Guess I'll stick with regular simulator instead of the alpha 19 preview

[Prometech]

It Seems we are moving backwards on Collision Accuracy. In previous versions, At high velocities, you would end up with many small fragments, in this version we end up with massive fragments that then magically begin to shrink and then will inexplicably disappear, or turn into huge attracting bodies and collide boringly into the surface and not fragment like the parent object.

Because we don't hear much from the developers I can only assume this is a result of trying to optimize for the new collisions and fragmentation systems, but I hope to see it return to a more realistic representation.

 as you can see in my second attachment, the 100% CPU load bug does affect my machine signifigantly as it results in sustained CPU temperatures above 150 Degrees Fahrenheit

[Jar]

Because we don't hear much from the developers I can only assume this is a result of trying to optimize for the new collisions and fragmentation systems, but I hope to see it return to a more realistic representation.

You're exactly right with this. The new fragmentation system has changed a lot, and because it's very new, we're still tuning the balance between performance and accuracy. We've made progress on this since the last preview version, but there will always be more to do. Of course, we don't want to take any steps backwards in the process. 

[Angel Armageddon]

Indeed. I'm not very fond of unrealistic, massive fragment.

[Angel Armageddon]

But, if you think about it, if a Mars sized object or bigger the earth head on, the planet would shatter and these very large type of fragments would be left and they would collide with one another and make a larger planet, possibly with one of these large fragments becoming a moon.

[Greenleaf]

But, if you think about it, if a Mars sized object or bigger the earth head on, the planet would shatter and these very large type of fragments would be left and they would collide with one another and make a larger planet, possibly with one of these large fragments becoming a moon.

Planets colliding would not give a few big chunks, but rather a liquid/gaseous spray which could then coalesce. Give the limited fragment budget, the material sometimes come off in too big pieces. This is something which is being actively worked on through better budgeting, better performance and better visualizations. We want this to be even better, for sure.

[Angel Armageddon]

But, if you think about it, if a Mars sized object or bigger the earth head on, the planet would shatter and these very large type of fragments would be left and they would collide with one another and make a larger planet, possibly with one of these large fragments becoming a moon.

Planets colliding would not give a few big chunks, but rather a liquid/gaseous spray which could then coalesce. Give the limited fragment budget, the material sometimes come off in too big pieces. This is something which is being actively worked on through better budgeting, better performance and better visualizations. We want this to be even better, for sure.

Awesome to hear.
Any news about the SPH system or atmospheric scattering?