KSP2 System Requirements

[MARL_Mk1]

7 hours ago, intelliCom said:

These tests have too many variable elements. You need to minimise the variables as much as possible. Here's what I propose:

Conditions:

• Settings either set to (Very) Low or High ("Medium" could be too demanding on lower-end cards, so let them see what Low looks like at least)
• All other programs other than KSP and the launcher closed. (Please record launcher used)
• A phone or paper(s) to record notes. Alternatively, a prescribed form to be printed / filled out on another computer.
  • I do not recommend recording the game unless you have a dedicated capture card. Recording software can use system resources, skewing the results.

To be recorded:

• PC Specs:
  • GPU ^{(This should be listed as "RX _____", "GTX  ______", or "RTX ______". This is NOT "Intel UHD Graphics" or "Vega" (unless you are limited to an iGPU))}
  • CPU
  • RAM
• Any overclocking used
• Screen resolution (1080p, 1440p, 4k, etc.)
  • If you are running a lower end card, feel free to set this to below 1080p if you need to achieve a 60 fps experience. Just record the resolution change.

Benchmarking tests:
For these, read all the steps for the test prior to starting that test. This is to minimise mistakes in the testing process. Keep the steps on your phone/paper(s). If a prescribed form is being used, a copy of the steps will be provided.

"Kerbal X to the Heavens"

  Reveal hidden contents

Demonstrates an average case scenario. The Kerbal X rocket goes straight up. That's it.

1. Load the Kerbal X onto the launch site. Note if any FPS spikes occur. Do not launch yet.
2. Record FPS at launch site once there are no lag spikes.
   • Could be after 5 seconds, or 10.
   • If 30 seconds have passed and lag spikes are still occurring, write notes on this.
3. Set throttle to max, launch the Kerbal X. Do not touch buttons for the rest of the test unless specified.
   • If it strays off course, record notes. Do not attempt to correct it.
4. Record FPS at the following mission times:
   • T+20s
   • T+40s
   • T+60s
5. Stage the Kerbal X when necessary. Ideally the instant that the previous stage runs out of fuel.
6. Once all FPS has been recorded, close the game and send results to the online form.
   • Either I can write this, or @MARL_Mk1 can.

 

"Engine Cluster from Hell"

  Reveal hidden contents

Demonstrates a worst case scenario. At least one copy of every Methalox engine in KSP fires at full throttle. Basically the mother of all static fires.

1. Load prescribed craft onto the launch site. Note if any FPS spikes occur. Do not launch yet.
   • Either I can design this, or @MARL_Mk1can.
2. Record FPS at launch site once there are no lag spikes.
   • Could be after 5 seconds, or 10.
   • If 30 seconds have passed and lag spikes are still occurring, write notes on this.
3. Set throttle to max, activate engines. Do not touch buttons for the rest of the test unless specified.
4. Record FPS at the following mission times:
   • T+20s
   • T+40s
   • T+60s
5. Once all FPS has been recorded, close the game and send results to the online form.

 

"MOAR BOOSTERS: Ragnarok" (For lower specs, I recommend Low settings only for this one. High settings may be unable to complete this test.)

  Reveal hidden contents

This is a true stress test. A behemoth mass of solid rocket boosters is launched into the air.

1. Load prescribed craft onto the launch site. Note if any FPS spikes occur. Do not launch yet.
   • Either I can design this, or @MARL_Mk1can.
2. Record FPS at launch site once there are no lag spikes. (FPS is probably pretty low anyway)
   • Could be after 5 seconds, or 10.
   • If 30 seconds have passed and lag spikes are still occurring, write notes on this.
3. Activate SRBs. Do not touch buttons for the rest of the test unless specified.
4. Record FPS at the following mission times:
   • T+20s
   • T+40s
   • T+60s
5. Once SRBs are out of fuel, wait 5 seconds and record FPS again.
6. Wait until SRBs fall back down. Timewarp if necessary. Do not timewarp within atmosphere at all.
7. The SRB mass should crash into the ground. When it does, it will likely create a massive lag spike. Record the number of seconds you must wait with a stopwatch or your phone.
8. Once all FPS has been recorded, close the game and send results to the online form.

 

That being said, I would absolutely like to contribute to these benchmarks. Just make sure serious thought is put into the benchmarking tests to minimise error.

 

3 hours ago, purpleivan said:

If this infomation is going to be useful and allow for for replicable tests, then a set of standardised  tests should be agreed on, to remove as much as possible, differences in how the game is being played and viewed at the time the framerates are being noted.

Here's some suggested scenarios.

Static (no engine firing) tests.

1. On the pad with the Mk1 capsule as the only part in your vehicle (to minimise the physics impact on FPS), with the camera unchanged from it's orientation when the launch pad scene loads. Wait 3 seconds after the scene becomes visible to allow for any processes related to spawning to finalise, before noting the FPS. This test should give a good baseline test of the game's performance in a very low intensity situation, in terms of both physic and graphics, and is quick and easy to replicate.
2. In an 80-81km orbit of Kerbin in the Kerbal X1, with the camera pointed towards the center of the planet, after doing a scene game save/reload, to reset the positioning of the camera . Not

Dynamic (engines running) tests.

1. In the VAB adjust the the Slim Shuttle (or similar stock vehicle if it's not in KSP2), by moving the launch clamps out out the stage to be triggered at launch (so the vehicle does not lift of the pad when fired). Send the vehicle to the pad, trigger the 1st stage and let the engines run for 10 seconds to allow for maximum emission of plume effects) and note the FPS. This to be done in the camera view of the pad when the scene is loaded.

The rocket plume effects are fill rather than vertex heavy, so it's critical to ensure as much as possible that the distance of the camera, the angle of it to the vehicle, amount of time the engines have been running, speed of vehicle etc. is consistent, for FPS measurements to have any usefulness. Therefore test of this are best on with the vehicle clamped to the launch pad and using the default camera view, to standardise that.

A measurement of the FPS of stock vehicles in-flight, with their engines running would be useful, but more tricky to standardise. The same is true of re-entry effects.

If gamesaves are easliy accessible in KSP2 (I'd hope that they are), then a set of standard tests could be created that way. e..g. a save of a vehicle that is mid re-entry could be loaded, the camera controls not touched and an FPS noted after 5 seconds, to allow for any build up of visual effects and settling of other state.

Of course any visual or potentially physics affecting settings would need to be standardised around a minimum and maximum settings, pair of test. For those with less powerful rigs, just post the minimum settings tests, or even just 1 test if that's all you have time for.

 

If people want to post other tests that's fine and will bring up some interesting edge cases, but having some standardised tests (with agreed names for tests to avoid confusion) will be the most useful.

Standard names should be used for reporting of the agreed tests, as this would allow for a simple post of easy to understand/refer to results, such as that below, rather than relying on people repeatedly describing the same tests, with an FPS figure at the end.

KSP2 Standardised test results.

Standard Test Static 1 = 93fps

Standard Test Static 2 =78fps

Standard Test Dynamic 1 = 36fps

You could even add it as a little graphic to your sig

Both of you brought up great points and considerations that I'll add into the main template shortly

2 hours ago, intelliCom said:

I was considering having "aim the camera straight down" as part of the test. However, that wouldn't resemble the usual gameplay that a player has.
That being said, if  @MARL_Mk1 is up to discussing how to prepare some benchmarking data collection (Google Forms, test cases, craft files, etc.), then him, you and I could discuss the test cases more heavily.

Also the VFX point is why I had the "Engine Cluster from Hell" test. Every single engine brings out every single kind of engine plume. If there are huge performance dips when there shouldn't be, one of the engines could be causing performance issues for some reason.  So it doubles as bug testing.

We can discuss it further yeah. Google Forms might not be the best one if we want the results to be instantly reflected in a thread comment.

Establishing the best and simplest test scenarios for people to follow would be the ideal way. If we ended up overcomplicating it, most people would just ignore it.

1 hour ago, JoeSchmuckatelli said:

If you are going to do this - you need to hammer into people the requirement of including their screen resolution.  Then you must separate results into the categories of 1080p, 1440p and 4k. 

Example: Knowing that player JS has a 3070 but is only getting 30fps with a 200 part-count craft is disingenuous to the average player unless you also know he's rocking a 32 inch 4k monitor. The average person only knows FPS as a number and does not think past that. 

It will be incredibly unfair to the Dev team for people to start shouting about how 'badly optimized' KSP2 is because players with 3k series GPUs are getting low frames, where the misinformed presumption is that affected everyone, equally.  However once you start scratching the surface and you discover that a 3070 at 1080p is fine - even optimal - the damage will already be done. 

The industry is in a transition state from the legacy of 24 inch monitors at 1080p to larger monitors at 1440p and 4k.  Most people don't recognize that. 

Please collate results by the number of pixels that the GPU is being asked to push! 

Yeah, completely forgot about resolutions, which is one of the biggest factors. Thanks for pointing it out!

50 minutes ago, JoeSchmuckatelli said:

Edit - I started the relevant threads. 

 

Folks participating... Please keep your results in the appropriate threads! 

 

 

There's no reason to open any threads about this yet. The methods, case scenarios and users organizing it are still being discussed (hence the existence of this thread).

 

Said final thread would be opened the 24th alongside the release of the game.

[JoeSchmuckatelli]

33 minutes ago, GoldForest said:

All screens today are wide screen, so, you're asking for all performance metrics 

No.  I get what you are saying from a technical standpoint - but the conventional usage of wide is... Okay fine I'll change it. 

[SSTO Crasher]

8 hours ago, intelliCom said:

Demonstrates an average case scenario. The Kerbal X rocket goes straight up. That's it.

1. Load the Kerbal X onto the launch site. Note if any FPS spikes occur. Do not launch yet.
2. Record FPS at launch site once there are no lag spikes.
   • Could be after 5 seconds, or 10.
   • If 30 seconds have passed and lag spikes are still occurring, write notes on this.
3. Set throttle to max, launch the Kerbal X. Do not touch buttons for the rest of the test unless specified.
   • If it strays off course, record notes. Do not attempt to correct it.
4. Record FPS at the following mission times:
   • T+20s
   • T+40s
   • T+60s
5. Stage the Kerbal X when necessary. Ideally the instant that the previous stage runs out of fuel.
6. Once all FPS has been recorded, close the game and send results to the online form.
   • Either I can write this, or @MARL_Mk1 can.

This assumes they keep the kerbalx

[Hanuman]

The way they added the older card, but specified a card that also has 6 GB of VRAM makes me wonder if that's the actual requirement.

I mean, I would hope that it's not going to require 6 GB of GPU to start up, but the thought won't go away.

[SSTO Crasher]

Since I’m pretty sure the clouds and other graphics are what will destroy the computer is there an option to disable them, that would make the game like ksp1 but cooler.

[Astr0Guy5]

6 minutes ago, SSTO Crasher said:

Since I’m pretty sure the clouds and other graphics are what will destroy the computer is there an option to disable them, that would make the game like ksp1 but cooler.

One can only hope. In 5 days all our questions will be answered.

[purpleivan]

3 hours ago, intelliCom said:

No one knows because it's not out yet lmao

I was hoping that either one of the mods might have some info on that, or better still one of the developers.

It might have come up in conversation at some point.

1 hour ago, JoeSchmuckatelli said:

If you are going to do this - you need to hammer into people the requirement of including their screen resolution.  Then you must separate results into the categories of 1080p, 1440p and 4k. 

Example: Knowing that player JS has a 3070 but is only getting 30fps with a 200 part-count craft is disingenuous to the average player unless you also know he's rocking a 32 inch 4k monitor. The average person only knows FPS as a number and does not think past that. 

It will be incredibly unfair to the Dev team for people to start shouting about how 'badly optimized' KSP2 is because players with 3k series GPUs are getting low frames, where the misinformed presumption is that affected everyone, equally.  However once you start scratching the surface and you discover that a 3070 at 1080p is fine - even optimal - the damage will already be done. 

The industry is in a transition state from the legacy of 24 inch monitors at 1080p to larger monitors at 1440p and 4k.  Most people don't recognize that. 

Please collate results by the number of pixels that the GPU is being asked to push! 

Agreed that anyone providing test results needs to state what resolution they are running the game at, however the we don't need the physical screen size as it isn't relevant.

Well aware of the presence of 4k monitors... have a couple of them sitting in front of me

30 minutes ago, SSTO Crasher said:

This assumes they keep the kerbalx

These were just examples of a specification, the final ones would of course require the use of the available stock vehicles.

If we manage to go the route of using a set of game saves to allow for custom vehicles to use, then the use of stock vehicles would not be necessary of course.

Edited February 19, 2023 by purpleivan

[Bej Kerman]

5 hours ago, Vegatoxi said:

For example - Elite Dangerous Odyssey. It was POOOORLY optimized on alpha. A bit better on release, and since 2 years it still not match it's own requirements and cant give stable 60 FPS.

If you nitpick examples from developers with poor reputations? Sure. If you consider how development tends to go? KSP 2 is probably going to have much lower requirements as EA progresses to a more consumer-ready state.

[Stephensan]

The template is a little long, i think we should only need post the Template once with full system specs. and then we just repost ontop of our original until we change graphics settings/etc.

 

then it would be a simplified.

• parts/amount (can estimate)
• location/alt
• fps/feel of the game (anything interesting the user wants to add)
• Photos/Videos.

finish

not the entire length of the "where during this flight" just the immediate form, and then if the user WANTS to add what is like.

 

 

[TLTay]

I don't see why the developer can't just include a built-in benchmark like many other games have.

Given the high spec requirements, it would end up getting added to a lot of YouTube hardware reviews and benchmark videos. Nearly free publicity over the course of the years. 

[Starhawk]

Overlapping topics have been merged.

Forum Moderation Team

[regex]

Quote

Is that actually what's considered a "high-end" PC? I thought I was buying something mid-range because I spent less than $2000 USD...

[GoldForest]

Just now, regex said:

Is that actually what's considered a "high-end" PC? I thought I was buying something mid-range because I spent less than $2000 USD...

If we go by nVidia and go off the technical standpoint: 3080 is high-end, 2060 is mid-range. 

I think @Stephensan is going off statistics, so like most people own a 1060 to 1080, so this would be considered "Mid-range". 

[purpleivan]

40 minutes ago, MARL_Mk1 said:

Some modifications to the template were just made.

I'd be great if we could come together and think about at least 2 or 3 more test scenarios that vary in part count, location and stress testing.

Tagging @intelliCom @purpleivan

What do you think of the formatting on the current "KerbalX to the Heavens" test?

I'd try to keep it as brief as possible for the reader to maximize the amount of users willing to get into it without making it feel like a chore.

For craft file sharing and save games we'd have to wait for the 24th either way.

I think my original reply to this was lost when I hit Submit Reply, but it was to a thread that by then didn't exist thanks to the merge

I'd say that the "Kerbal X to the heavens test" is a fine one at the lighterweight end of the spectrum, as the vehicle has a pretty low part count.

I'd add something to say "Don't move the camera until at over 70km" just in case people think it's ok to move it once over the pad and in the ascent. In addition I think a little more detail for the +70km test items would be good.

For the "+70km" I'd add "make sure that none of Kerbin is in view" if that's the view we want for that test item.

For "+70km, looking towards Kerbin." I'd add something like "make sure Kerbin fills the screen".

 

As for other tests, I think something like the one proposed by @intelliCom that has a bunch of large boosters raging all at once would be good, especially if the vehicle is clamped to the ground, so not moving away from it's exhaust plume.

As for snacks?... Aways

19 minutes ago, JoeSchmuckatelli said:

I can understand why you did this... But unless you get people collating performance reports by screen resolution... You are likely to have a bunch of confused players out there that first week. 

Most people don't seem to understand the relationship between resolutions and hardware and will only look for FPS numbers... Which will be all over the place starting Friday. 

Which is why @intelliCom @MARL_Mk1 and I have been trying to put together a set of standardised test scenearios, without which, any FPS numbers will be equally irelevant.

[Stephensan]

Just now, GoldForest said:

If we go by nVidia and go off the technical standpoint: 3080 is high-end, 2060 is mid-range. 

I think @Stephensan is going off statistics, so like most people own a 1060 to 1080, so this would be considered "Mid-range". 

pretty much this.

going based off of the community of gaming, and not the "year per year" term of what is "mid range" of X generation there is people still using gpu's from 2016 making them 7 years old... and even so the rtx 3080 is still a high end GPU, not flagship.. i hear "quite" a few people having rtx 3070s and below all the way down to gtx 1650's~ rather than anyone with a rtx 3080 and above. and 1650's are rather new to the gamers eyes, even if the preformance is not "on par with the rtx lines"

if you look at the amount of people having "worser gpu's" than rtx line, thats where most newcomers/new gamers can afford.. due to the huge crypto craze that happened, not alot of people recovered from that really. people don't buy into the newer gen gpus every year/other.

 

i can completely swallow my pride and say the last gen 2000 series is a more budget, lower range gpus but the rtx 3080 is still a high end gpu's.

[Kerbin Launch Coalition]

3 minutes ago, Stephensan said:

pretty much this.

going based off of the community of gaming, and not the "year per year" term of what is "mid range" of X generation there is people still using gpu's from 2016 making them 7 years old... and even so the rtx 3080 is still a high end GPU, not flagship.. i hear "quite" a few people having rtx 3070s and below all the way down to gtx 1650's~ rather than anyone with a rtx 3080 and above. and 1650's are rather new to the gamers eyes, even if the preformance is not "on par with the rtx lines"

if you look at the amount of people having "worser gpu's" than rtx line, thats where most newcomers/new gamers can afford.. due to the huge crypto craze that happened, not alot of people recovered from that really. people don't buy into the newer gen gpus every year/other.

 

i can completely swallow my pride and say the last gen 2000 series is a more budget, lower range gpus but the rtx 3080 is still a high end gpu's.

Sorry but no-one in their right mind thinks the £700 (or more) RTX 3080 is a mid-range card. 2060/3060 is solid mid-range, and even then I think the 3060 is still well above what most Kerbal Space Program players are using. I'm quite looking forward to seeing how it runs (or doesn't) on RTX 3070 which is by all metrics a higher end of the mid-range card. Still really looking forward to getting my hands on the game because I'm eager to get a look at the 'bones' of the game cause there looks to be loads of nice ergonomic touches.

[Alexoff]

[JoeSchmuckatelli]

13 minutes ago, purpleivan said:

Which is why @intelliCom @MARL_Mk1 and I have been trying to put together a set of standardised test scenearios, without which, any FPS numbers will be equally irelevant

I was watching the development with interest - my posts were in support of the effort. 

But you can look at the previous 10 posts in this thread and see what I'm concerned about.  People who think 'midrange' or 'high end' is a price point - because of the baked in presumption that everyone uses a 1080p screen. 

Mid range isn't a price point - it's a performance metric.  Or at least it used to be, and of course price correlated with performance - which caused the misperception. 

3080 is super high end / way overkill for most games at 1080p... But it's putting up mid to high fps numbers at 4k.

So if you have a 1080p monitor with a computer housing a 3080 - you have a high end machine. 

If you have a 4k monitor with a computer housing a 3080... You have a mid range machine. 

Both are bloody expensive. 

[Bej Kerman]

Just now, Alexoff said:

Still looks better than KSP 1, even with the intervention of modders.

[Starhawk]

Some content has been removed.

Open discussion of moderation is forbidden by the forum guidelines.

If you have an issue with forum moderation please feel free to take it up with a member of the moderation team.

Thank you for your understanding

[JoeSchmuckatelli]

Midrange / highend = meaningless. 

You must discuss resolution in relation to the hardware for anything to be relevant. 

 

[JoeSchmuckatelli]

Look - PC players cover a wide gamut from

*really informed geeks who know what is going on beneath the hardware and system requirements

to

*I'm playing on a repurposed work laptop from last year

to 

*My mom bought me this from Best Buy for Christmas! 

It's been this way since Dell was new.  (Prior to Dell, you kinda had to build your own). We all had huge CRTs on the desk and boring boxes with no lcds and you could listen to your computer connect to the internet.  Way, way back then the geeks talked quite a bit about screen resolution.  The transition from CRT to LCD screens started in the very late 90s and only became predominant in 2007.  But once 1080p (HD) became common and cheap... It became the standard.  So much so that no one ever talked about screen resolution when talking about game or hardware performance. 

Up to the last year and a half or so, all of the reviews, all of the price /performance metrics and talk about what makes a machine mid range or high end comes from this time - when 1080p was presumed.  These conventions still bias the language today. 

1440p started popping up in the middle teens - Esp when 27 inch screens became available - and they were niche for a very long time.  4k was available on 27 inch screens and 4k OLED TVs but did not really make sense until 32 inch screens became available last year. 

And so for the first time in decades the old conventions MEAN NOTHING!  If you advertise or talk about a mid-range product or machine, that conversation is meaningful only if you specify what resolution you expect to play the game at. 

Edited February 19, 2023 by JoeSchmuckatelli

[JoeSchmuckatelli]

Let me add this: when the 3k /6k series of GPUs came out 2 years ago... They were COMPLETELY UNNECESSARY for anyone rocking 1080p.  All you got was useless frames above the performance ceiling of your monitor. 

But. 

Some people were already moving to the 27 inch 1440p and 48 inch 4k monitors as well as ultrawides.  Those people needed the extra muscle provided by the new architecture. 

The (at the time) mainstream 3070 card was /is 

* entry level / low end 4k 

* mid range 1440p 

* high end / enthusiast1080p 

You can discount the existence of 3090 /4090 as absurd and see that the 3080 / 4080 cards are for the 'enthusiast' class across the 4k/1440p boards. 

3080 is a 'more dollars than sense' card for a 1080p monitor owner. 

... And yet, now the 3080 and 6800 are two year old 'last Gen' cards.  Thus it actually makes sense to call it mid-range or 'Recommended' level of hardware in the context of a PC game - but we and they should be clear that the resolution matters. 

https://imgur.io/5sIMvjw?r

If they had system requirements much below that... We'd be screaming about wasted time and effort since 2020! 

(Also - the Dev team is not responsible for NVidia/AMD's opportunistic pricing!) 

Edited February 19, 2023 by JoeSchmuckatelli

[Kerbin Launch Coalition]

8 minutes ago, JoeSchmuckatelli said:

Let me add this: when the 3k /6k series of GPUs came out 2 years ago... They were COMPLETELY UNNECESSARY for anyone rocking 1080p.  All you got was useless frames above the performance ceiling of your monitor. 

But. 

Some people were already moving to the 27 inch 1440p and 48 inch 4k monitors as well as ultrawides.  Those people needed the extra muscle provided by the new architecture. 

The (at the time) mainstream 3070 card was /is 

* entry level / low end 4k 

* mid range 1440p 

* high end / enthusiast1080p 

You can discount the existence of 3090 /4090 as absurd and see that the 3080 / 4080 cards are for the 'enthusiast' class across the 4k/1440p boards. 

3080 is a 'more dollars than sense' card for a 1080p monitor owner. 

 

As a RTX 3070 user, on a laptop with a 1440P 144Hz screen. The GPU is absolutely considered high mid-range at that resolution and 'overkill at 1080P'. Now I've got no issue running at 1080P since 1440P on a 17.2" screen isn't all that noticeable but I expecting the game to run well with high settings at least at 1080P. 

[boolybooly]

Phew, I can probably just about playtest it ! 

I think its good they released minspec before allowing any orders.

I was recently thinking of going 1440p (currently 1080p) but am glad I didn't now. 

Saving for a card upgrade but want something more power efficient than a 2070S  and I try and I try and I try and I try but I don't get no ... satisfaction!