Performance Impact

[sfjuocekr]

This is the title of the ship and this is what it claims to achieve as well.

 

I play KSP2 on a 5800X3D, 3080FE and 64Gb of RAM yet this ship manages to crumble KSP2! This is my usual rocket design that -usually- works just fine, you launch it, tilt it just a tad, turn on SAS and make it follow prograde. It -should- more or less launch itself into a 100+ AP from where you can circularize.

 

Is it efficient? Probably not, I just wanted to hand-in a possible less than performant situation for the development team.

Download

edit: I seem to have no clue how to add the damn image

EDIT1001: I gave up, I have no clue how this image adding is supposed to work.

Edited March 6, 2023 by Starhawk
Corrected link

[Dmitr]

a very large ship, but it feels like it's almost suicide to assemble from more than 100 parts for weaker computers. (dude I have the same problem with the image)

[Initar]

For the image, I advise you to use imgur, then copy the link under the BBCode section.

Paste on your post and voila.

Edited March 5, 2023 by Initar

[Dmitr]

1 hour ago, Initar said:

Для изображения я советую вам использовать imgur, затем скопируйте ссылку в раздел BBCode.

Вставьте свой пост и вуаля.

Hmmm can we use google drive ? it's just that the url from google drive doesn't work for me.

[sfjuocekr]

2 hours ago, Initar said:

For the image, I advise you to use imgur, then copy the link under the BBCode section.

Paste on your post and voila.

I uploaded the image to my own server, the "insert image from URL" button to the lower right will not take it.

3 hours ago, Dmitr said:

a very large ship, but it feels like it's almost suicide to assemble from more than 100 parts for weaker computers. (dude I have the same problem with the image)

This is not a large ship, I make the same design from smaller parts and the result is the same. It was also mentioned that the performance regression caused by multiple engines on a single source was to be expected, basically all my rockets are of this design.

[RayneCloud]

In the most recent video from @ShadowZone that he posted today, Nate sai during the interview that their own internal testing vehicle part size is around a 150 parts as they think that's the "sweet spot" for vehicle design and creativity. So, I imagine they're trying to balance performance impact around that part count. Even KSP 1 struggles with large complex vehicles, but that limit is closer to around a 1,000 parts from what I've seen of certain creators on youtube and twitch.

[Starhawk]

44 minutes ago, sfjuocekr said:

I uploaded the image to my own server, the "insert image from URL" button to the lower right will not take it.

I believe that you need the 'http://' part of the URL in order to trigger the 'Insert Image' functionality.

Happy landings!

[sfjuocekr]

I have flown "simple" rockets like this for over a decade, this is not a large rocket at all and I have in-fact flown much more elaborate contraptions that do not suffer from this performance impact.

 

Going from 60 to <5 by just turning engines from zero to >1.

 

Quote

1. Fuel flow/Resource System optimization. Some of you may have noticed that adding a high number of engines noticeably impacts framerate. This has to do with CPU-intensive fuel flow and Delta-V update calculations that are exacerbated when multiple engines are pulling from a common fuel source. The current system is both working and stable, but there is clearly room for performance improvement. We are re-evaluating this system to improve its scalability.

 

So this is why I came here, this type of rocket is what I usually build and could hopefully be used to profile this performance regression. I can also supply a simpler version, but the result is the same.

7 hours ago, Starhawk said:

I believe that you need the 'http://' part of the URL in order to trigger the 'Insert Image' functionality.

Happy landings!

Trust me, I tried with and without.

[Starhawk]

3 minutes ago, sfjuocekr said:

Trust me, I tried with and without.

I took the liberty of just using the copy link function of my web browser and pasting it back into your post.  It automatically adds the prefix.  It seems to have worked.
At least, I can see the picture now.

Happy landings!

[ShadowZone]

11 hours ago, RayneCloud said:

In the most recent video from @ShadowZone that he posted today, Nate sai during the interview that their own internal testing vehicle part size is around a 150 parts as they think that's the "sweet spot" for vehicle design and creativity. So, I imagine they're trying to balance performance impact around that part count. Even KSP 1 struggles with large complex vehicles, but that limit is closer to around a 1,000 parts from what I've seen of certain creators on youtube and twitch.

That's not really what Nate said, at least that's not how I understood it. He said that 150 parts is the "sweet spot" where you can start to have some fun with the game. That was in reply to the minimum spec discussion. So what this means is that minimum spec should enable people to use at least a 150 part vessel without too much lag. It doesn't say that they don't test with larger vehicles.

 

[RayneCloud]

6 hours ago, ShadowZone said:

That's not really what Nate said, at least that's not how I understood it. He said that 150 parts is the "sweet spot" where you can start to have some fun with the game. That was in reply to the minimum spec discussion. So what this means is that minimum spec should enable people to use at least a 150 part vessel without too much lag. It doesn't say that they don't test with larger vehicles.

 

Ah, I must have misunderstood then, fair point Shadow.

[sfjuocekr]

On 3/6/2023 at 3:47 AM, Starhawk said:

I took the liberty of just using the copy link function of my web browser and pasting it back into your post.  It automatically adds the prefix.  It seems to have worked.
At least, I can see the picture now.

Happy landings!

Thanks, though for some reason... the image does not work for me