paul_c

Just checking in....Played KSP a bit more too, after having moved on and a bunch of other games, then remembering KSP2 might be coming soon. Personally for me, and I hope my caveman status adds a little relevance: 1. Having to do the levels in sequence has merits but on balance isn't a great idea since it would simply be too boring. The harder levels (especially NCD) are pretty grindy as-is. 2. It is not possible to differentiate between "I just happen to have a bunch of experience from playing KSP in previous saves" and "I am running a "test save" alongside my Caveman attempt. If anything, its a disadvantage because it shows you're pretty green to it. IF NO DATA TRANSFER were done, eg not opening craft files of a different save, instead stuff were 'transferred' from your own memory or paper notes alongside playing the game, fair enough. 3. As I recall, you need to do many flights within NCD anyway, to build up the cash. I think this means anyone would inevitably be using the Caveman save to do "testing" (as in, other flights with useful info/data). Other info/data/knowledge outside of these are likely to be quirks with spaceflight, or more accurately, KSP's gameplay implementation of it. Outside testing to find these out is fair enough. TLDR: If there's no data/file transfers going on, then a "test save" is fine in my eyes.

My NCD was detailed back on page 12 (and its own thread too: ) I remember, the key to being able to earn money was taking on the "test part ......." contracts and recovering 100% of the craft, so the only cost was the fuel. Thus, so long as you keep on and on at those, they earn money. And I built up a 2 dimensional table of TWR & deltaV and their "performance" in terms of height gained and speed achieved. So I was able to reliably hit those contracts 1st or 2nd time (and sometimes do 2 in 1). I never ever took on tourists - the rocket cost more than the earnings. After every 10 or so money-earning contracts, I'd do a bit more 'exploring' and very gradually earned the science. I also used contracts strategically, for example every "satellite in orbit of Mun" contract was also coincidentally a relay in part of a constellation......or part of a trip there etc.

Aerodynamics (or, lack of them) are critical for re-entry from high altitudes or high speeds. You need to have a fairly blunt capsule to shave off a lot of the speed and when the chutes deploy, they won't break. I remember a phase of somewhat marginal chute operation. I lowered the altitude and raised the pressure they opened at, to achieve later operation (the altitude is an above-ground-level one, not above-sea-level so it will cope with hilly terrain etc okay). You can often survive even with chute deployment around 800-1500m.

Kerbin's is by definition 70km* *Although you can't timewarp above 1x at that altitude, so 70001 isn't that 'useful'. I consider the practical working minimum, 70250m for that reason. And then there's the consideration if you're docking.........etc

Thanks for the tips, once I have a chance to play KSP again I'll try them out!

It means it needs to be assembled in the VAB/SPH but it can be two or more parts which are docked in space.

I've been getting back into KSP for a few days/weeks now and a bunch of missions are going to Jool and its moons. The delta V for maneouvring around Jool is quite high and I'm always clashing with the moons in a fairly random way. But is there any way to sort out the intercepts to the benefit of saving the dV needed to carry (ie more fuel, bigger rocket(s) underneath etc etc) I am guessing, you want to get the intercept with Jool as close to 250km Pe, but direct in (ie no moon intercepts beforehand) but then on the way out, get an intercept which curls you into another Jool orbit with as low an Ap as possible, but then the Pe is probably not going to stay at ~250km? If you intercept Jool then do a manoeuvre of some kind to intercept your intended destination moon, have you saved anything/much anyway, or will whatever you haven't burned at the Jool Pe just be needed at the moon's Pe to be captured by its orbit?

The contract system is 'flawed' or 'works' in that if you land on a body, it will give you contracts "to" and "from" that body (ie tourist trips, where to be successful they need to safely return to Kerbin). And also it will come up with "extract ore from...." contracts where its a huge amount of ore to get into orbit or somewhere else. Another indirectly related one is where a space station is specified, but with a huge amount of liquid fuel or ore on it, and/or a piece of equipment which will never be used (resource scanners not in polar orbit, etc).

If its post-game then just upgrade stuff and use a better aerial?

Doesn't weight come into it at some stage? You're launching 100+ aerials at 70kg each, that's 7000kg you're sending to a far-away planet. Compare with a piloted spaceship to the far away place, at about 900kg for the CM and astronaut. No aerials whatsoever needed. Even if you take into consideration you HAVE to do a return trip to get the science points back home, 1) returned points are greater than transmitted ones 2) if you can get there (with the Caveman transfer), you can get back 3) the weight saving can be used towards the extra fuel for the return trip 4) piloted trips mean you can gather crew report science too. I would have thought some maths shows that (1) alone, means a piloted return trip gets you over the line with less trip(s) than a remote trip. Obviously for Mun and Minmus, remote probes make sense. Its fairly easy to set up a relay constellation and/or put enough comms power onto a craft to do it. But you'll need to send a pilot for the full points from crew reports.

Setting the settings isn't the issue, its what the settings should be. When I did my NCD attempt, I never did hardly any EVA in flight stuff. I tried it a few times, building a "flying bedstead" style contraption. The Kerbal never held onto the ladder long enough to be useful for me, so I went down a rabbithole of trying to build some kind of cage/enclosure to ensure they didn't fall off and go splat (and possibly die). Personally for me it failed the "would NASA do this in real life?" test. (I know about Neil Armstrong's somewhat haphazard attempts at flying the moon landing simulator thing, where he nearly died doing some test flight). So I never really pursued it, and forewent the extra points. I know others have had success though - not sure how? - so its difficult for me to impose my own personal views and say it should be banned etc. But if its greatly facilitated by a setting, it doesn't seem right - Caveman should be about the basic KSP, no upgrades, and with (harsh) fixed group of settings which are the same for everyone. So whatever the setting is that makes it easy/possible, it would be a "no" from me. But I respect I don't really have a vote! And its up to others to actually decide!

I use https://alexmoon.github.io/ksp/#/Eve/18000/Kerbin/100/false/optimalPlaneChange/false/37/227 to work out the transfer window. Simply read off the UT (universal time?), then enter this in the "earliest departure", then it will give you the best time/date to do your burn to leave Eve and return to Kerbin. A big elliptical (or inclined in Eve) orbit isn't such an issue, since so long as you're roughly within a few days at the transfer window, the burn won't cost too much dV. You don't need to-the-second precision here. Similarly, if you're inclined, it will take a bit more dV but its not really worth* aligning to 0deg inclination just for the sake of the transfer burn. The "set as target" then use the indications to see the closest approach distance is handy, especially if the two bodies are inclined (such as Eve and Kerbin) because it doesn't require you to be precise enough to have the encounter within the Kerbin SOI. But its clear from your screenshot that you're miles away in your encounter. I suspect you're on the edge of what KSP actually say is/isn't an encounter (it doesn't give you it if its so far away) so I think if you get the basic position much closer, it will also be more stable when you make (small) corrections to optimise the encounter to get you home. For interplanetary stuff, its not massively time critical (unlike, say a Mun encounter where you want to burn at a precise point around the orbit of Kerbin) so you can afford the time to use the manoeuvre planning tool to tweak it so its good, then time warp there and do the relevant burn. Makes sense to do the prograde/retrograde as near to the Ap or Pe (or while still in Eve SOI) and the normal/antinormal as near to or at the AN/DN.

You'll need a relay constellation to avoid blackout on "the dark side of the Mun". in Caveman challenge, there is only the HG-55 available. If you position 3x satellites equally in a circular orbit which is altitude at least the radius of the body, you'll have 100% comms except maybe wisps of the edges (eg if the satellites are equatorial, then you might have a momentary dropout exactly at a pole). Pragmatically, put them much higher than the radius and you can afford a bit of leeway in their exact phase and orbit. I think I stuck 3x HG55 onto the relays (they are quite heavy and draggy though) and it was marginal when orbiting at 2000km but otherwise, no worries on comms.

I do the same (for the Challenge). There are 2 times of the day when Kerbin and Minmus align. Every dV saved due to an efficiency is good, imagine being 4m/s short at some later point in the flight.... There's 2 routes (or choose both): - Orbital docking, ie send 2 things up, join together a "normal" spaceship and a "mainly fuel tank" ship - Get into remote probe stuff. You'll need a comms network (which can be as simple as 2x satellites, if you do it on the cheap - but I did 3) ideally; and if you use the OCTO, you also want some kind of reaction wheel too otherwise you'll have no control. If you're into the manned flights thing, and joining 2 or more, then 1 manned and 1 remote makes sense otherwise you need 2x command modules and there isn't much performance advantage once that 2nd CM is there etc, its just bigger. ETA I've done Mun land-return within the 1 craft, 30 parts limit. Its VERY tight for performance but possible. Its so marginal, that I only did it once or twice then reverted to docking stuff, mainly because I could then visit many biomes but only need 1 'return to Kerbin' and all the heatshield & parachute gubbins.

Don't forget if you aren't going down the path of remote probes, you need to get the Kerbal BACK from Minmus as well as there! You start out with 2 pilots so you can afford to lose just 1, the rehire cost will wipe you out otherwise. With no quicksaves etc....you need to be 100% sure you can get them back, somehow or other.

That's the 3 things you need to do to get anywhere in Caveman Challenge! I'm scratching my head how you could progress without at least some of those, because you will need a few tech tree squares (and some money) before you have a hope of building a spaceship which can get to Mun or beyond.

The screenshots have MechJob in them too. If you didn't use any of its extra features, it could have been uninstalled for the challenge.

When I did it, I built 2x modules with antennas (I think 8 each - they weigh quite a lot, 70kg?) but in the end, the extra antenna power wasn't needed - it only extended comms to a little further out of Kerbin's SOI and I was always planning to come back to surface of Kerbin anyway, to achieve max science points. It was a failsafe/backup which was of little value, and not needed. Fortunately, I had attached them to a fuel tank so I was able to make use of the fuel in the modules anyway, just not as much as a fuel-only module. Its worth being familiar with the power/transmission requirements to fly to Minmus (and relay comms from it) though.

How did you control the modules and station, I can't see a controller there? Also....any science? I don't see the Science Jr or Goo!

I used a separate website: https://alexmoon.github.io/ksp/ to find the best launch window (time/date) and https://ksp.olex.biz/ to help visualise the angles, but once the values were 'looked up' I manually made a note of them, wound the clock forwards using TimeWarp then estimated the ejection angle to burn at, sometimes using a protractor pressed against the monitor, caveman style. KSP program remained bog standard for this.

Regarding AIO, a handy rule of thumb is that 120mm doesn't really do much benefit, you'll see the benefits (more stable temperature, less severe peaks, because the water has a much bigger heat capacity to absorb the heat/energy when the CPU makes spikes on demand/power) with 240mm and upwards. And if you do use one, install the tubes the right way round! a 2.5GB (or faster) ethernet is only really a benefit when you have 2 or more devices connected using it, and you need to make fast comms over the network to that device, eg if you had a file server or something. As a home user I don't have much call for it. Also, a wired network connection (of any speed) has a small latency advantage over wireless (of any speed).

I think we're more or less in agreement - a CPU benchmark score (of say 28637 vs 14728) will result in a real world performance increase of less than that - around 17-25%. I am hesitant to put a definite figure on it. I think you'd notice a 20% difference. You'd notice the difference if the computer was just rebuilt anyway.

I know, I couldn't find a test which puts 8086K vs 5800X, what with the difference in age of about 2 1/2 years. I don't see how you came to 17% - its not the German, but there's about 3478 figures on that page you linked. Its main thrust is comparing a bunch of Ryzen processors to 10th gen (not 8th gen) Intel CPUs? What is a "theoretical benchmark"? A benchmark is the numbers from some or other real-world test. Whether that test is relevant to your use case, is what's important. For example, if you found a test of eg CS:GO at 1080p with a 2070 video card, and wanted to know how the two perform at CS:GO at 1080p with a 2070 video card, great! If its a different game, or another resolution, or using a different video card (note that CPU tests are typically done with a high end card to not encounter a GPU-limited situation, rather it tests the CPU at more intensive loads), or productivity tasks instead of games, or some other use, then it will always need interpreting to some degree or another. This is why I took my own linked benchmark with a pinch of salt and restricted my description to "a good improvement". Not "a 100% improvement" or "a 17% improvement" - I accept that the CPU benchmark alone doesn't directly correlate to real-world (including playing the latest games) frame rates achievable, for many reasons including if it becomes GPU-bound (which is not a big issue with a 2070 if you are sensible with resolution and settings).

I think you'll see a good improvement moving from the 8086K to a Ryzen 5800X. https://www.cpubenchmark.net/compare/Intel-i7-8086K-vs-AMD-Ryzen-7-5800X/3279vs3869 Frequency is really only one of a number of factors, don't worry too much about one vs another. Look towards real-world independent benchmarks instead. Agree that PCIe gen 4 won't really give a real-world performance improvement over gen 3 so don't let it skew your motherboard choice etc. Definitely stick with the 2070 for now, if you want to keep the old PC working (not sure if that CPU has integrated graphics?) then a 1030 will suffice to give reasonable day-to-day use. Yes DDR5 is coming but I think it will be 12-18 months before its a worthwhile upgrade. Full rebuild and "declutter" also improves performance, so you'll see a big uplift in the old vs new for that reason too - but then you could rebuild your existing PC for a lift in performance.....

Mun is tidally locked but Minmus isn't, that's the main reason.