Can we just say thank you for Squad's latest attention to performance tweaks?

[Guest]

Still getting that massive does of lag over oceans. I've noticed no improvements. I've been playing with MPDT set at .01 since .21. It's the only way I can get descent framerates, and when I say descent, I mean more than 5 fps on my old laptop. I've still got all the same exact audio bugs as before, except maybe an improvement in the garbage collection hiccups, and that's from a unity upgrade, not squad tweeks. One would theorize that if squad actually included code optimization, one would see improvements in the way the game sounds.

I might be wrong about it but I think the "max physics delta time per frame" value is irrelevant to FPS.

The game already works and even before worked as this: render a frame - do physics step - render a frame - do physics step - ...

But the size of the physics step was modulated by how long time has passed between frames and by this maximum cap.

CPU consumed on doing the physics step is the same regardless how big the delta t is, it's just that the physics is more accurate when the step is smaller.

With value of 0.1, the game started to slow down the game time versus real time below 10 FPS

With value of 0.04, the same starts to happen below 25 FPS

That is wrong. MPDT settings have a direct relation to FPS. The reports of improved framerates are an exact description of what happens when one lowers the MPDT settings.

[LameLefty]

I'm seeing performance improvements on my 2.0 gHz i7 MacBook Pro in space or when keeping the camera pointed away from the ground. However, terrain lag is the no better and ocean lag is noticeably worse. Alas.

[Libra00]

Yes, definitely thank you Squad for the attention to performance. I've gotten a noticeable FPS increase on large vehicles and loading times are a good bit shorter. Thanks!

Edit: i5 quad 3.3GHz / 8GB RAM.

[Themohawkninja]

I've actually been getting performance drops in the VAB, as there have been some random major lag spikes.

That being said, I have yet to seriously test the new tweaks. I have noticed that the SRB sounds seem to be fixed, which I am quite happy about.

[Chrisd857]

I can say that water/horizon lag is still annoying, but when in space or in the VAB/SPH, my FPS has decidedly gone up. And this is on a crappy computer that almost has no right to be running this game (I really ought to get more than 1GB of RAM, or a newer-than-2005-computer).

[_Aramchek_]

Not seeing that on my i7 3930K @ 4Ghz and GTX570. Actually no version sense 0.18.2 have had any measurable improvement in performance loading screens aside but thats pretty irrelevant In my opinion.

With the physics delta at .22 defaults I am getting a more solid framerate, and slightly higher fps sometimes on my [email protected].

It's not a night and day difference, but a difference does exist.

[spectrhz]

I'll definitely say thank you to squad! My part count capability has increased by a couple hundred by this update! Must have been some junk floating around in the code that they removed

[Exeveros]

Before had problems even looking at Laythe and its ocean... now my 120 part heavy transporter is searing through the skies of Laythe with not even a single hiccup.

Developers take note how a company is run and a product improved! Congratulations Squad! Extraordinary update. Now its time to set up a science station on Eve... Thank you a million times!

[Technical Ben]

It has been stated that the performance increases would be most noticable with large craft and lower end machines, and that's where I think your confusion comes from: the people who are noticing very little impact all seem to have high end i7 processors.

My 8l 16 cylinder Veyron is not doing any better mileage with that nice "economy" fuel.

[Whackjob]

When you're measuring framerate in seconds per frame, instead of the reverse, it's easy to see an improvement.

But really, yes, it's a tremendous improvement if you're using huge constructions. I'm not Whackjobian, but I've got some big rockets, and they're much easier to control now. Once we get Unity multithreading, things'll be even better, but who knows when that'll happen.

I've rarely flown at seconds-per-frame levels. That one, the picture I had up, I was getting maybe 5 fps. I'm running almost 30 now.

[HarvesteR]

I changed the default max delta-time per frame to 0.04, as you all noticed. It was already possible to tweak manually, of course, but the reason I changed it was because I noticed most players didn't know this tweak existed, and at it's old default of 0.1, you needed to be lagging pretty horribly already (10 fps is nigh on unplayable) before the game would start to slow down. That made the whole feature a lot less useful than it had the potential to be.

Setting the new default at 0.04 means it'll start to kick in at around 25 fps, which means it'll always try to keep performance to a reasonable level. But more importantly, this will now happen to everyone regardless of whether they know of the setting's existence.

Oh, and thank you for the thank you thread! It's always great to see that people appreciate the work we do.

Cheers

Edited December 18, 2013 by HarvesteR

[SkyRender]

The improvements are indeed nice, though I've noticed they come at a cost (at least for my system): the atmosphere on Kerbin now lags the game severely no matter what. Ostensibly dropping the values for precision on the ocean helps with that, but then the water gets horribly glitchy. I know I'm not the only one with the issue, and that SQUAD is aware of it, but it is nonetheless disheartening to see a "performance boost" result in a performance drop instead in a very common scenario.

[Lunaran]

However, terrain lag is the no better and ocean lag is noticeably worse. Alas.

I used to ride reentering craft as far as chute deployment and hit "end flight" as soon as I'd proven to myself the craft wouldn't tear apart, because that last 500m was just stuttering torture, but they took that away

I'm still playing this game way too much, of course, so, <3

[lincourtl]

I changed the default max delta-time per frame to 0.04, as you all noticed. It was already possible to tweak manually, of course, but the reason I changed it was because I noticed most players didn't know this tweak existed, and at it's old default of 0.1, you needed to be lagging pretty horribly already (10 fps is nigh on unplayable) before the game would start to slow down. That made the whole feature a lot less useful than it had the potential to be.

Thanks for confirming this HarvesteR. For some of us on less beefy machines though, it resulted in horrible performance. Adjusting it back to 0.1 still gives me water lag, but it's back to being acceptable -- as it was in 0.22. I said this to C7 earlier, but it would really be helpful if the change log listed all the little changes. It would reduce a lot of the hand wringing and guesswork on each new release.

That being said, I hope you, and the whole dev team realize how special KSP is, and how thoroughly appreciated all your hard work and vision are. Happy holidays.

[The Jedi Master]

It would seem some people have it run slower, and some faster. Me, I think I'm in the slower group. I definitely noticed the flag near the launch center waves slower.

And the throttle control... oh my God...

[Pds314]

Did anyone else notice that the

default: max physics delta-time per frame

is now 0.04 and the slider only can do so much in comparison to older versions?

Do i remember wrong?

No. You are correct. The old default was 0.10. I noticed that as well. I would think that this would make things more laggy, but apparently not by much.

[cerberusti]

Hmm... I hesitate in posting this (especially since I am on my 5th beer for tonight!)

I see a few complaints about multithreading, but they are misguided. There are very few nontrivial programs which are limited by CPU speed, they are mostly limited my memory access time (you can calculate PI, or encode some video or audio and see a speedup, but little else.)

Cache memory is mostly the deciding factor in modern execution speed, and all the cores on your processor share most levels of cache. Attempting to multithread a memory bandwidth limited program leads to significantly slower execution than a single threaded solution (it thrashes the cache at an exponential rate as you start extra threads, and keeping data consistent across cores requires that you invalidate the cache more often.)

I do not program video games for a living, so take that as you will. I do however have more than 20 years of experience as systems programmer, and my day job is very much optimization of programs which have real consequences if they do no perform well (think robotics when you miss updating what it should do, and financial programs where you must be faster than the competition in order to make money.)

Many of my programs do have multithreaded versions... but you need to contact our customer support to get that version, and the first question they ask is if you have multiple processors (which have independent caches) or multiple cores (which share a cache.) If you have multiple cores, an optimal program is almost always single threaded.

You can sometimes get a speedup if you ignore cache coherency, but that means a buggy program instead.

EDIT: This also ignores the rather large extra time investment correct multithreading involves, and so most companies (correctly) make the choice to ignore it and focus on extra features in their single threaded program (I spent about ten years as a very expensive consultant for software companies.)

Edited December 19, 2013 by cerberusti

[MR4Y]

I'm not really thankful. At least for me, 0.23 is even slower than 0.22.

[Lazro]

I'm not really thankful. At least for me, 0.23 is even slower than 0.22.

If you really think so, check your settings for your older save and see if they're the same. If so, then try sliding your physics delta-t up a bit? For smaller ships particularly I've found this can increase framerates.

[ekim98]

I don't get it, my FPS is actually slower than ever before, and I have no mods or anything

[Whackjob]

I do notice one thing, however. Large constructions tend to be far more difficult to control.

I'm seeing a kind of "runaway" effect where I start tipping in one direction, and all the control in the world can't correct it. I've even put a large load of RCS on this thing, on top of the myriad of reaction wheels I normally use. Only option is more reaction wheels. I think. We'll see.

#EDIT: Ran another test after adding about 50 reaction wheels. Flies fine now. Not a problem.

Edited December 19, 2013 by Whackjob

[rosenkranz]

http://i.imgur.com/P44LhgF.png

I've gotten a five-to-six-fold increase in FPS when watching this.

Good lord Whack! What is that even for?

[Whackjob]

Good lord Whack! What is that even for?

When I get around to bolting on the other five sections to those two, I'll have one sharp looking Munar lander. Note the legs on the second segment.

[CoriW]

Something interesting that I've noticed about this update is that whenever Kerbin's atmosphere is on screen, my FPS will not budge up or down from 17. On the other side of the scale, if the atmosphere is not on screen, my FPS will go up to 60.

Does this seem odd to anyone else? Oh and I may as well add that my graphics settings are as follows.

(Max Physics Delta Time: 0.04)

Terrain Detail: Very Low (Which is basically Low, but with the Ocean Max Subdivision set to 1.)

SM3 Terrain Shaders: Enabled

Terrain Scatters (WIP): Enabled

Scatter Density: 50%

Render Quality Level: Simple

Texture Quality: Half Res

Fallback part shaders: False

Aerodynamic FX Quality: Low

Screen Resolution: 1920x1080

Full Screen: Enabled

Anti-Aliasing: Off

V Sync: Off

Frame Limit: 60 FPS

Pixel Light Count: 4

Shadow Cascades: 2

Also, my CPU is an AMD Athlon II X2 215, my GPU is a Radeon HD 5570 1GB, and my RAM is 3GB

[TNM]

There is an issue between 0-160km where the terrain (mostly ocean some say) hits hard on FPS. Probably an LOD setting in the engine that prevents incremental steps to unload stuff, otherwise there wouldn't be a sudden jump at 160km.