Kerbal Space Program Update 1.5 Grand Discussion thread

[Johnny Wishbone]

First, I'm happy to report that 1.5.1 has addressed the aero issues I was having. My planes actually glide again as they did in previous releases.

 

2 hours ago, Deddly said:

I think what he means is that you should compare the performance between the two stock versions. That way, you will know for sure whether or not 1.5 really is at fault for your performance loss. 

You don't have to play in stock apart from testing the performance, but it's the first step in finding what is causing the problem. 

Just throwing my 2 cents into this discussion: I play 100% stock. No mods whatsoever; pure as Colombian snow. Thus far, I have not noticed any performance issues. That said, I also am not using the new burn calculator (which apparently causes some performance issues if I'm understanding other posts in this thread correctly) and I haven't really started using big ships with hundreds of parts, etc. I'm still in the early stages of my career game, so my ships are generally small with low part counts. Everything seems to be running smooth for me thus far, especially with the aero fix now in place.

[NoobTool]

1 hour ago, NoobTool said:

I've noticed a pretty substantial performance hit in a clean stock install. What's more, the performance changes seem related to how many/what type of engines are activated (NERV's seem to be the biggest offender). This, along with the fact that burn-time is being calculated incorrectly for NERV-powered craft, leads me to believe that the issue is related to the new extended burn-time calculator.

To clarify, the performance issues I'm reporting are occurring with a relatively high-part-count craft. Also, they occur whether or not the extended burn-time indicator is turned on or off in the settings (I imagine the calculations still run whether it is turned on or not). Even though this is occurring with a somewhat high part count craft (~460 parts?), the difference is quite pronounced. As soon as NERV's are activated, the game begins to freeze for 1/3 of a second every second. As soon as the NERV's are deactivated, this behavior stops. This did not happen in 1.4.5.

Edited October 19, 2018 by NoobTool

[Gaba_Kerman]

the pool ones

[nasa legolas]

ok so I LOVE THIS UPDATE ITS NEEEDED SO NEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEDED X5

[DMagic]

3 hours ago, NoobTool said:

As soon as NERV's are activated, the game begins to freeze for 1/3 of a second every second. As soon as the NERV's are deactivated, this behavior stops. This did not happen in 1.4.5.

Have you tried running the engines at different throttle levels? There seems to be a massive impact of running engines at very low throttle vs full throttle, both in terms of garbage allocation and overall performance. For a real fun time set the engines's throttle limiters down to about 5%, then throttle up just a tiny amount... If you have MemGraph open you can see the garbage allocation go through the roof.

Edited October 19, 2018 by DMagic

[Shadowmage]

4 hours ago, NoobTool said:

To clarify, the performance issues I'm reporting are occurring with a relatively high-part-count craft. Also, they occur whether or not the extended burn-time indicator is turned on or off in the settings (I imagine the calculations still run whether it is turned on or not). Even though this is occurring with a somewhat high part count craft (~460 parts?), the difference is quite pronounced. As soon as NERV's are activated, the game begins to freeze for 1/3 of a second every second. As soon as the NERV's are deactivated, this behavior stops. This did not happen in 1.4.5.

 

51 minutes ago, DMagic said:

Have you tried running the engines at different throttle levels? There seems to be a massive impact of running engines at very low throttle vs full throttle, both in terms of garbage allocation and overall performance. For a real fun time set the engines's throttle limiters down to about 5%, then throttle up just a tiny amount... If you have MemGraph open you can see the garbage allocation go through the roof.

Interesting information -- wonder if the issue is being caused by something engine specific (particles?), by the stock delta-v calcs, or something else entirely?    That information at least should be enough for the devs to know where to start looking for the cause, assuming the issue is replicable(sp?) on other arbitrary installations.

Does activating converters or other constant resource drain cause the same sort of performance impact?  If it was simply vessel/part mass updates triggering the delta-v calculation, any changes to vessel mass should trigger the same kind of performance issues.

[NoobTool]

3 hours ago, DMagic said:

Have you tried running the engines at different throttle levels? There seems to be a massive impact of running engines at very low throttle vs full throttle, both in terms of garbage allocation and overall performance. For a real fun time set the engines's throttle limiters down to about 5%, then throttle up just a tiny amount... If you have MemGraph open you can see the garbage allocation go through the roof.

 

2 hours ago, Shadowmage said:

 

Interesting information -- wonder if the issue is being caused by something engine specific (particles?), by the stock delta-v calcs, or something else entirely?    That information at least should be enough for the devs to know where to start looking for the cause, assuming the issue is replicable(sp?) on other arbitrary installations.

Does activating converters or other constant resource drain cause the same sort of performance impact?  If it was simply vessel/part mass updates triggering the delta-v calculation, any changes to vessel mass should trigger the same kind of performance issues.

I haven't yet tested various throttle settings, but the performance drops happen whether throttle is set to 0, or 100%. Even with throttle at 0, simply activating the NERV's with action groups causes the performance to drop, deactivating them remedies it.

I also noticed something interesting on a recent ascent. NERV's are the last engine type to shutdown before apoapsis on the craft in question, so were the only engines active (throttle at 0). I set a maneuver node to circularize and the burn-time indicator displayed a figure that seemed reasonable. The craft was still in the atmosphere at the time. As soon as the craft reached 70km altitude, the burn-time indicator changed to N/A. The indicator didn't show any values thereafter even after burning w/ the NERV's again. Not sure what that means (maybe the burn-time indicator is checking for the presence of intake air for NERV's?), but I thought it might be useful information.

[THEdragon]

I really like how the kerbals in the menu screens are more expressive now! The blank expressions were okay, but now they have a bit more personality.

[BlinAndVodka238]

So, using DirectX11 makes the VAB/SPH part selection menu's Texture goes missing (just a blank silhouette), in KSP 1.5 there was a mod that fixes this by changing the VAB/SPH texture with another one, here it : https://drive.google.com/file/d/1XdydhpOOuo3yvzvwm8KDL7jI8-ikx8xf/view

The bug: 

is there an equivalent of this for KSP 1.5?

 

[Shadowmage]

24 minutes ago, BlinAndVodka238 said:

is there an equivalent of this for KSP 1.5?

Check @Electrocutor's thread, as I think he has the patch for that still available (should still work for 1.5), as well as a quick patch to fix the lack of reflections in the flight scene for stock shaders.

 

[FinalFan]

I just noticed that docking ports can now be given staging events in the VAB.  Nice!  Sure, doing it by hand was never THAT much of a hassle, but it's a much appreciated subtle-but-significant quality of life improvement.  (If you have to stage via decoupler under strict time pressures you're very much an edge case IMO, and quite possibly Doing it Wrong™.) 

[edit:  Nope, it was there all along in Advanced Tweakables, which I recently activated.]

Edited October 20, 2018 by FinalFan
edit

[JPLRepo]

17 hours ago, NoobTool said:

To clarify, the performance issues I'm reporting are occurring with a relatively high-part-count craft. Also, they occur whether or not the extended burn-time indicator is turned on or off in the settings (I imagine the calculations still run whether it is turned on or not). Even though this is occurring with a somewhat high part count craft (~460 parts?), the difference is quite pronounced. As soon as NERV's are activated, the game begins to freeze for 1/3 of a second every second. As soon as the NERV's are deactivated, this behavior stops. This did not happen in 1.4.5.

For very large part count vessels - with very long burning engines. What you can do is edit your settings.cfg file.
Change DELTAV_CALCULATIONS_BIGTIMESTEP to a higher value and see how that goes... the down-side is the DV shown on man nodes for such vessels will be less accurate.

[edit] thought I’d better clarify that a bit more. The DV required  for the manuever will be accurate but the burn time and stage bars may be less accurate the higher you make this value.

[RealKerbal3x]

5 hours ago, FinalFan said:

I just noticed that docking ports can now be given staging events in the VAB.  Nice!  Sure, doing it by hand was never THAT much of a hassle, but it's a much appreciated subtle-but-significant quality of life improvement.  (If you have to stage via decoupler under strict time pressures you're very much an edge case IMO, and quite possibly Doing it Wrong™.) 

Hasn't that been a thing for ages? I've only been here since 1.3.1, but there's always been a button to 'enable staging' on docking ports. I've done it many times before 1.5.

[FinalFan]

36 minutes ago, RealKerbal3x said:

Hasn't that been a thing for ages? I've only been here since 1.3.1, but there's always been a button to 'enable staging' on docking ports. I've done it many times before 1.5.

Well, this is embarrassing!  I guess I just hadn't done it often enough to learn it.  There were a couple launches in particular, though, where I know it would have come in handy but I recall having to do it manually... 

Maybe it's part of advanced tweakables?  I just turned that on a week or two ago.  

[Wallygator]

MacOS here --> Strangely, I had to open the 1.5.1 via opening the package contents, whilst in 1.5 I could open the app directly.  Did you guys not register this hotfix app?

[RealKerbal3x]

22 minutes ago, FinalFan said:

Maybe it's part of advanced tweakables?  I just turned that on a week or two ago.

I believe it is. Well, we learn something new every day  

[Atkara]

1 hour ago, FinalFan said:

Maybe it's part of advanced tweakables?  I just turned that on a week or two ago.

Maybe not so. I could've sworn the first time I ever staged a docking port, was in 1.0.5. No advanced tweakables back then.

[NoobTool]

9 hours ago, JPLRepo said:

For very large part count vessels - with very long burning engines. What you can do is edit your settings.cfg file.
Change DELTAV_CALCULATIONS_BIGTIMESTEP to a higher value and see how that goes... the down-side is the DV shown on man nodes for such vessels will be less accurate.

[edit] thought I’d better clarify that a bit more. The DV required  for the manuever will be accurate but the burn time and stage bars may be less accurate the higher you make this value.

Thanks for this. I'll have to give that a try. The burn-time for nodes using NERV's are pretty inaccurate atm anyway (or at least, they have been for me).

[FinalFan]

5 hours ago, Atkara said:

Maybe not so. I could've sworn the first time I ever staged a docking port, was in 1.0.5. No advanced tweakables back then.

And yet, when I turn advanced tweakables off, it goes away. 

[Atkara]

1 hour ago, FinalFan said:

And yet, when I turn advanced tweakables off, it goes away. 

Ah, ok then

[AVaughan]

@JPLRepo A couple of weeks ago someone from Squad mentioned in https://forum.kerbalspaceprogram.com/index.php?/topic/178756-ksp-weekly-the-moon-race/ 

Quote

Bug fixing was included in this week’s agenda and among this final sweep, we implemented an auto spring/damper to the wheel suspension system in order to diminish involuntary bouncing. We also adjusted the wheel friction and suspension parameters to tackle this problem. We expect that this will improve the behavior of the wheels and legs in the game.

 

This seems to be the relevant changelog entry.

On 10/12/2018 at 2:48 AM, UomoCapra said:

Bug fixing was included in this week’s agenda and among this final sweep, we implemented an auto spring/damper to the wheel suspension system in order to diminish involuntary bouncing. We also adjusted the wheel friction and suspension parameters to tackle this problem. We expect that this will improve the behavior of the wheels and legs in the game.

 

After playing with 1.5.x for a while, my impression is that the bouncing of landers is actually worse.  (I don't play much with planes, so I admit haven't tested them).

That was in a completely stock version of 1.5.1 (zero mods).  The initial landing was with everything at default settings.  That is more bounces than I experienced with similar designs in 1.4.5.  (I count about 7 bounces during the first landing.  The last time I landed a lander on Minmus in 1.4.5 it had about 2 bounces.  (Admittedly a different, heavier design, with heavier landing legs).

Why is damper always set so low?  (A real world car with such poor shock absorbers would be completely and dangerously unroadworthy).  A damper should absorb a portion of the downward velocity energy during spring compression, and a portion of the spring's energy during spring extension, effectively reducing the velocity of the craft with every bounce.  This damper seems to be almost non-existent.  Also why is the damper limited to 2?    

[MikeO89]

1 hour ago, AVaughan said:

@JPLRepo A couple of weeks ago someone from Squad mentioned in https://forum.kerbalspaceprogram.com/index.php?/topic/178756-ksp-weekly-the-moon-race/ 

 

This seems to be the relevant changelog entry.

 

 

After playing with 1.5.x for a while, my impression is that the bouncing of landers is actually worse.  (I don't play much with planes, so I admit haven't tested them).

That was in a completely stock version of 1.5.1 (zero mods).  The initial landing was with everything at default settings.  That is more bounces than I experienced with similar designs in 1.4.5.  (I count about 7 bounces during the first landing.  The last time I landed a lander on Minmus in 1.4.5 it had about 2 bounces.  (Admittedly a different, heavier design, with heavier landing legs).

Why is damper always set so low?  (A real world car with such poor shock absorbers would be completely and dangerously unroadworthy).  A damper should absorb a portion of the downward velocity energy during spring compression, and a portion of the spring's energy during spring extension, effectively reducing the velocity of the craft with every bounce.  This damper seems to be almost non-existent.  Also why is the damper limited to 2?    

Man you hit a nerve. This has been a pet peeve of mine since I've been playing this game. The stock landing legs just kill me. Right now the ONLY landing legs I've found that don't do this bounce (compress down and then spring back up) are the Apollo LEM legs from the FASA mod, so I use them where ever I can get them to fit. I can't put into words how much the bouncing action of landing legs irritate the crap out of me. Why Squad does not provide an option to take out all hydraulic action of the legs (to make them fixed) is beyond me.

Edited October 21, 2018 by MikeO89

[Guest]

deleted

Edited March 29, 2020 by Guest

[Dafni]

On 10/19/2018 at 6:39 PM, DMagic said:

Have you tried running the engines at different throttle levels? There seems to be a massive impact of running engines at very low throttle vs full throttle, both in terms of garbage allocation and overall performance. For a real fun time set the engines's throttle limiters down to about 5%, then throttle up just a tiny amount... If you have MemGraph open you can see the garbage allocation go through the roof.

No really? That sounds worrying. Can anyone else confirm this?

[COL.R.Neville]

anyone else having problems with the attachment nodes on the FLT fuel tanks? 400 800 etc

like for the flt800 using node helper these are the values they should probably be set to

    node_stack_top =     0.0, 1.84375,  0.0, 0.0,  1.0, 0.0,1
    node_stack_bottom = 0.0, -1.85938, 0.0, 0.0, -1.0, 0.0,1

and this is what they are currently set to

    node_stack_top =     0.0, 1.875,  0.0, 0.0,  1.0, 0.0,1
    node_stack_bottom = 0.0, -1.8875, 0.0, 0.0, -1.0, 0.0,1

Edited October 21, 2018 by COL.R.Neville