pa1983

i5 4670K would still be as fast in KSP, maybe a bit faster but theres nothing wrong with the i7 4820K. I dont know how it overclocks but it has a soder IHS from what I know so should be easy to cool. But make sure the Motherboard supports it out of the box. My ASUS board can be flashed from a USB memory stick even with out CPU, memory etc, only requires a PSU to flash it. MSI im not sure but if that motherboard was out before the new IVY-E well then its more then likely not to support it and even post. So one must make sure to either have a SB-E CPU available to flash or get a motherboard that supports it out of the box or can be flashed even if the board wont post. The advantages of the i7 4820K for overclockers is the Soder IHS. Easier to keep temperatures down at high power outputs. For power users the 4820K makes sens if you need a ton of ram or high memory bandwidth or a lot of PCI-E lains but not that much CPU performance so the LGA2011 is a platform for the I/O intensive user. Im no real fan of the LGA1150 or older platforms mainly because the lack of PCI-E lanes but still for most users its enough and the processors are as fast per core performance or even a bit faster so unless you need more I/O, memory or cores LGA1150 will be good enough. For a pure KSP PC an i5 4670K is the ultimate CPU. i7 4820K makes sense if you can overclock it a lot more. 2ms on a screen is usually gray to gray so its useless number and theres no standard to measure it. You only know if you have a good screen by looking at it and comparing it to other screens. I prefer IPS panel and CCFL is actually better then LED in terms of spectrum. 800W PSU is way overkill. Could run that thing on a 450W PSU easy. I have 750W PSU and never see much over 400W from the wall during synthetic stress test on bot GPU and CPU and thats with a i7 3930K at 4Ghz and GTX570 and 32Gb of ram. Idle my computer pulls 130W overclocked from the wall and rarely over 300W load in games or programs. Question is was it a good PSU? Seasonic is the best OEM and one of there 80+ gold rated or better PSU's is a good investment. Corsair sells the AX line for example that are made by Seasonic. I have stop using other brands and OEM's and Run Seasonic in 3 computers now ranging from 2-5 years old and still going strong and 5-7 year warranty is not bad either and efficient and dead silent to. Cheap PSU's will break thats just a fact so paying a lot for a good one is worth it. They last longer and keeps the system stable and they come with a long warranty, 5-7 years. 1333Mhz ram is fine but last time I checked 1600Mhz didn't cost a dime more and the i7 4820k supports up to 1866Mhz to tough a bit overkill but theres realy no down side to faster memory other then a very small gain in performance often measured in 0.1% or less. But processors tend to have there cache and pref-etch etc matched to the accepted mainstream memory standard and that has been 1600mhz for a few years now so its usually the price performance sweet spot sens it do not cost more then any other DDR3 ram and it offers good performance. And I personally would avoid those prebuilt liquid cooling system. Better to build one from quality parts one self. Pump can fail and they do make noise so quality is key there. The best heat pipe coolers are as good as those closed and ready to use kits and they require no pumps. Even satellites uses heat pipes for the same very reason, no power and a ton more reliable and heatpipes move heat very fast so there accentually very good. These closed and prebuilt 120-140mm liquid coolers dont impress me at all. Howl point of water cooling is removing heat from a small source and moving it to a big radiator. Unless the radiator is not much bigger then competitive air coolers theres realy no point. Airflow is pretty restrictive so they end up being a lot more noisy then say a good tower cooler with less densely spaced fins like a NH-D14. Water cooling needs a big radiator to make sense from a noise perspective and cooling performance advantage to justify the added cost. But prebuilt Liquid coolers have advantages in cramped chassis where they make a good option. But I find them more to be a market ploy taking advantage of the "water is cooler then air" factor. You still use air to remove the heat in the end so and heatpipes uses a liquid that is evaporating at low tempratures, boiling and its very efficient. Water is actually a poor heat conductor but it can hold a lot of energy. It also hard to say what cooler fits you bill with out knowing what chassis you will use and what cooler the GFX card has. A well designed PC all these components must match for good cooling and noise performance. I consider 120x240 rad to be minimum for water cooling to make sens. But yea your PC will play KSP just fine at top of the line performance so it wont get much faster then that other then overclocking the processor. If it was just for KSP it was not the best price/performance option sense an i5 rig could probably do the job for almost half the cost with identical performance. If its for something else why not as long as your happy thats whats important. I run the LGA2011 x79 platform my self so and Im happy with it but its probably the least cost effective platform for KSP. The latest build was an i7 3770 in an ITX case and it preforms stock as good as my i7 3930K @ 4Ghz in KSP but I did not build a 6 core machine for KSP. An i5 3570K or 4670K would preform just as good sense KSP wont use any of the advantages of the i7 CPU at all. P.S. Dont take it all to negatively. Its just my opinions from my experience. That rig is sweet on its own, optimal for KSP or not. But sense the question was, what would the best hardware for KSP be thats the answer i compiled. I would easily use that i7 4820K in my filserver just to have a ton of ram so its not a bad CPU its plenty good

I think I just sad that didn't I? Well running in to high temperatures before overclocking potential is reached is not the same problem and the ship was not designed with overclocking in mind either. Either you get a processor with soder IHS or you remove the IHS and apply some real liquid metal. http://www.liontech.se/PartDetail.aspx?q=p:4610247 That metal will solve that problem. I have seen two test of it and I have also done the numbers, thermal past VS a liquid metal with conductivity similar to aluminum and well a stock i7 3770K would run about 10.5C cooler if I recall stock and running at about 150W overclock about 20C cooler. There tests and my math matched 99%. Can find all the information needed on wikipedia for thermodynamics and specification for common thermal pasts on many web pages selling it. Some manufacturers of thermal past even have calculators where you can add say the die area of a CPU and get the estimated Delta T between CPU and heatsink or what ever else is interfacing with it. But any way I dont use IVY or Haswell for overclocking. All my processors are soderd and those that where not back in the day I removed the IHS from. Was quite common practice to do on AMD K6-2 processors back in the day. I also do very extensive testing and so fare no processor I have owned in the last 10 years or so have run in to temperature problems before the silicon hit its limits. I usually run 24-32h prime95, two instances of intel linpack for at least two hours sense I use Hyper Threading that will stress the CPU a LOT more thats the reason an i5 clocks better then an i7 because an i7 can use its resources simultaneously more efficiently and as a results more transistors are working requiring more power so there is a bigger chance of instability with Hyper Threading on. But the gains with hyper threading is in the order of 10-60% in multi threaded programs so an extra 10% power use and a little lower overclocking potential is worth it if your using it for multi threaded programs. I also run 24h memtest86+. On should also run those test before overclocking. No point on overclocking if the stock system is unstable and that do happen. Crap motherboard, psu or ram. Most people I speak with that overlcock dont test nearly good enough. many just plays a game and then will say there system is stable. Well how can they know? I would say 9 out of 10 rigs overclocked are unstable at least and it would not be hard to make them fail. I have made a lot of rigs fail even with little overclocking. Its not hard. My goal is Factory stability when overclocked so I run extensive tests and re run them. The reason many but not all people hit 4.5Ghz etc is because they just dont test good enough. I run my processor at 4Ghz. 4.2Ghz it will require a lot more voltage to past intel linpack and its not worth it. But I can run it at 4.2 and run prime95 for as long as I want or one instant of intel linpack. But two and it will spit out errors randomly. And theres nothing like its stable if you can run your programs on it. If a test program fails its not stable, theres no in-between. Either a system will computer correctly or it wont. The question is not if its unstable but how unstable if it fails a test. A bit error can go unnoticed or it could crash the howl system. All depends on where it happens. I have had it up to 4.4Ghz playing KSP no problems. Just wont pass stability tests for a longer period. Thing is eventually even KSP would cause it to fail. Just statistics.

Its pretty irrelevant that KSP is 32bit and by it self limited to 2Gb. The fact that the OS will use ram for other stuff. You dont put 2Gb of ram in a machine because your program dont use more then 2Gb. First off you have a ton of processes running all requiring ram. In a modern OS the ram serves as a Level 4 cache more or less under the OS control. If the the hard drive was the ram instead then the Ram would be the equivalent of the processors cache. What all modern OS will do and have done for some time is caching pages in ram. Be it disk space or ram memory is divided in to pages. If you access a file on your hard drive if there is space the OS will cache it in ram even if you close it. Next time you open it if its still in ram the OS will retrieve it from there and not the hard drive. Thats whey programs and files open quicker on machines with a ton of ram compared to low end machines with 4-6Gb of ram that more often have to flush pages to the hard drive. I have 32Gb of ram, I can open a program the next day and it starts quicker to day then yesterday because its still cached in ram. Even the new build I made for my parents have 16Gb of ram, 1Gb dedicated to the HD4000 GPU in the i7 3770 processor. On could argue 8Gb would be enough but they will have that PC for 5 years at least and well 8Gb is minimum this days and easy to use up editing photos or doing anything medium heavy. But that machine is extremely responsive with its SSD and RAM combo. Advantage of disk cache or page cache, what ever you want to call it is to speed up the machine and hide the slow hard drives, network attached storage, optical media, usb drives etc etc as much as possible. If you dont have ram for this and you dont have an OS for this you would basically be back at windows 95 days of computer responsiveness with the hour glass staring you in the face. Doing anything be it starting a program, opening a file would take ages EVERY time you did it just because the CPU sits around waiting for that slow hard drive with a access time of 15 000 000 CPU clock cycles. Even an SSD cant compare with RAM in access time so an SSD wont make ram irrelevant as page cache. There is a big misconception that if you have enough ram for your program more ram wont make the computer faster. Well thats not true. Your program can load that file faster your working on or make changes to it faster and continue on with out delay. Any modern OS will in many cases do the changes in ram to a file first then write the changes in the background. If the user requests the file before its written to the hard drive etc the copy will be retrieved from ram from the disk cache/page cache. A big reasons for this misconception is windows. Windows shows ALL ram that is not used by programs but as page cache as free ram. Well its free in the sens that page cache dont have priority over programs when it comes to reserving ram. If a program needs ram it needs it so it can accentually run. But still that page cache needs to be flushed to the hard drive and that takes some time and will happen more often if you have little free ram so more ram is better or you just end up swapping and we all know how fun disk swap is right? Can read more about it here. http://en.wikipedia.org/wiki/Page_cache

Heat is not the main killer, not voltage either but its probably second place. Current is the killer and pretty much NO enthusiast considers current as the killer. Temperature dont kill as long as its with in specifications. 80C is not a problem, most CPU's can take that 24/7. Even higher temperatures with out permanent damage. Most intel processors have a limit between 95 and 105C before thermal protection even kicks in. Higher voltage is dangerous but as long as its with in specifications and not exceeding max limits its not the main killer either. Temperature will increase leakage and increase current but just a bit. Voltage will increase current a lot especially if you increase voltage a lot since current goes up exponentially with voltage. Thats why one can use a bit higher voltage but excessive voltage will have consequences and no cooling in the world will stop the current from increasing to any significant degree. Current is the killer and what high currents causes is "Electromigration". It basicly means atoms of other materials emigrate with the electrons and the higher the current the faster they emigrate. Current in a modern CPU is about 75 to a 100 Amps. How fast materials emigrate depends on the current and the square area of the conductor/conductors and the material there made off. But in the long run soder points and components inside a IC will be destroyed from electro-migration when connections break due to material like copper or thin emigrates. So ANY overclocking is by this very definition not risk free. Not even a stock CPU will last for ever. Electro emigration is a real world problem and has been known for 100 years. When a CPU is designed its given a life span by the manufacturer trough the design and lithography process that should be enough for its intended use and life span. One side effect of electro-migration that all IC's like a CPU suffers from is that its stable frequency decreases over time due to electro-migration so thats why many new processors can be under volted. All processors have a slightly higher voltage then needed when new to offset the negative effect of electro-migration. Sure the higher voltage causes higher current and more electro-migration but a balance between voltage and frequency gives the best life span. So no there is no such thing as "safe" overclocking. There is always a risk and one have to know those risk to make a good desiccation. So just increasing the frequency is not safe. Assuming you increased a 2Ghz CPU to 3Ghz on stock voltage. Thats is at least 50% increase in current and probably as much or more in degradation due to electro migration. How long it will last, how knows only the manufacturer would know and they wont tell you. Electromigration is a real problem for manufacturers. I know because my relative designs IC's from 3Com and now Ericsson so we have talked about this subject. Voltage or temperature with in spec wont kill a CPU, Electro-migration will and "extra" cooling wont save you if you love high frequency's and voltages. Might by you a tad bit of time but what doce that matter when 99% of the killing is done by electro-migration? Not much. I have been in to overclocking since 1998 and I have long ago abandoned uber low temperatures. They might give a better OC since many but not all cpu's etc overclock some what better with lower temperatures, was more evident with some old processors I had even on air cooling unlike more modern processors. instead a very stable medium overclocking with little increase in voltage and current kept in check is what I prefer. A good cooler like the NH-D14 and a handpicked 140mm fan will give good performance and dead silent PC and it wont cost you a fortune like water cooling. I would say the more you know about overclocking and I mean know like real facts not myths and crap the safer it is. But its not risk free like some say. Its quite common to here that its safe as long as you just increase frequency and not voltage for example. Not true. As stated above current goes up continuously with frequency so electro-migration increases to so life span will be reduced. More relevant question is, will it die before I retire the CPU? Probably not. But crap do happen. 9800GTX cards died because of electro-migration where the thin moved with the current. Baking it in the oven repaid the soder points. Intels 6 series chipset with IVY-Bridge suffered from early degradation due to electro migration where some old transistors effected the current and degraded the sata portion of the controller. My relative at Ericsson sad that mistake like this has happened many times before so its not uncommon. But overclocking will push the chip beyond what it was designed for. Old Pentium 4 northwood could pretty easily be killed by increasing voltage close ore above 1.6V and frequency close to 4Ghz and that usually resulted in an unstable and soon unless and broken CPU due to electro-migration with in 6 months. It was Called Northwood Sudden Death Syndrome. I do recommend that people overclock IF they have the patience to read up, learn and stability test properly. Lot of people dont tough. But if done properly its relatively safe and the computer will last longer then you want to keep it around but theres no such thing as no RISK. Biggest risk is not knowing what your doing. You can read more about electromigration here. http://en.wikipedia.org/wiki/Electromigration

You want the best per core performance possible so thats an i5 or i7. You dont need Hyper Threading and the extra cache on the i7 wont do much either in games. i7 usually comes at a slightly high clock but its marginal at best. So from a price performance view an i5 will give 99.9% identical performance to an i7. Rigth now the i5 4670K is your best option. Any modern i5 or i7 is fast out of the box even none K versions due to Turboboost giving them speeds up to 3.8-4Ghz depending on model. I tested an i7 3770 3.4Ghz with max turbo of 3.9Ghz and it was as fast in KSP as an i7 3930K overclocked to 4Ghz on all cores. KSP just dont use more then 1 core plus 20-80% of another core at the most so turbo boost should kick in pretty well and give very good performance on none overclocked i5 or i7 to. An overclocked i5 or i7 will be faster but not that much since most overclock all the cores to the same speed so KSP dont realy benefit from that since turbo boost can go almost as high stock on few active cores. But yea a i5 4670K is a good start, should give you the best possible stock performance in KSP and if you want to overclock it will give you good overclocked performance to. And dont be cheap on the ram. 8Gb is bare minimum this days. 16Gb is well normal now. Ram is used by to OS to speed up a lot more then just allocate ram for your programs. Windows tends to show this as free ram but theres no such thing as free ram in a modern OS this days. More is better. I ran my CPU at 4.4Ghz in KSP instead of 4Ghz but realy didnt make much difference. Im CPU limited with my i7 3930K and I have tweaked the game. But the way I figure it if physics runs at half real time at 4Ghz well one would basically have to run the CPU at twice the speed to get back up to real time. So well 10% overclocking when your pushing part count wont realy do that much. Basically need double the CPU performance to speed up physics twice as much so if your running 0.5x real time a 10% OC at best would give you 0.55x real time. Not realy worth the trouble. Not saying your should not overclock but its easy to overclock all cores to say 4Ghz or something but the CPU might run at 3.7Ghz minimum with all cores in turbo boost mode any way stock with out overclocking. I know the i7 3770 that runs 3.4Ghz stock can all 4 cores at 3.7Ghz under prime95 and intel linpack continuously since it was within its TDP still and it can hit 3.9Ghz with 1-2 cores if I recall and the KSP test I ran seems to indicate high turboo boost frequency's. If the rig is Purely for KSP you could try disabling two cores in the UEFI/BIOS and see if that increases overclocking potential. Should make KSP run faster since more then two cores dont do any good any way.

Just get up to 2000m/s at 32-34km and then increase vertical climb to 100m/s or more if possible and throttle down the jets as needed. Takes practice sure. Give it a day. Having fewer jets usually makes it easier to. Less temperamental on the throttle. Also shutting jets down if possible helps. EDIT: And yes having control surfaces on the inside would not work IRL. I could most of the time put them on the outside. One reason I have control surfaces on the inside is because there protected and well looks like crap on the outside. I would not realy have much need of them IF and I say IF KSP's control surfaces where able to be assigned to specific functions. But they are not. Putting control surfaces in the center of an axis is the only way to limit control and well ASAS is crap at being smart so while you might be able to have good pitch control manually asas will oscillate and use all your control surfaces as ailerons or as pitch at once, what ever makes the most sense to it. Problem is that one often losses pitch control or get excessive roll movements. So thats in the end why I have control surfaces inside the craft in the center of the nose and tail. To control pitch and Yaw separately and give them just the right amount of authority to get good response but no oscillation from ASAS. If you could assign etch control surface a specific function, yaw,roll or pitch it would not be needed realy. Also bigger control surfaces would help big crafts a lot so we dont need so many small. But more esthetically problem and a part count problem. And some times I put wings inside the fuselage to balance the craft out a bit. Some might think its "wrong" but then any Wing or Fuselage with more then one side facing the same direction, basically any fuselage made of wings or any wing having more then a single layer would be cheating by the same definition. After all the game trys to "imitate" reality, it is "NOT" reality just an illusion of it. But yes try to put some rudders in the CENTER line of the pod like I did or some Elivators in the center line and you will see that you cant use them for anything else then ruder or pitch manually and ASAS cant either. That way you can give the craft the right authority on etch axis, either its pitch, yaw or roll. Its important for a good craft and you will see as you develop your craft how important for stability it is.

I dont think any one is saying your cheating. You your self where talking about "not" cheating. That statement assumes you think there is a way of cheating. I personally would say infinite fuel is cheating but only for a challenge just like breaking any other ruel in a challenge is cheating. I use it to test my crafts on the runway. I can try an emergency landing just after take off at maximum tonnage to see how the craft takes that by not consuming any fuel. That way I can take my time and not worry about going to low on the wight to be sure it can take it. That would also help if I refueled before landing some where also. Different reasons to use the developers tools menu. But in the end its simple. If the game allows it with out hacking the game (changing the code) its not cheating. You cant cheat in a game since the rules are sett by the code and the code was written by the programmer. If the game dose something the devs didn't intend its up to them to fix it, after all it was most likely there fault, not intentionally but what ever they tried to make it probably worked but also did something else. If intake spaming was not intended well having etch intake add to the total was a pretty dumb idea. Even I could figure that out asap and altitude limit or a combination of number of intakes and altitude. Once you have enough intakes more would not mater. Code for that would not be very hard to write. But if you as a player change the game code then your cheating by definition compared to others. But if that matters is another matter. Multi-player games yes that would be cheating. Single player, well not so much because how cares? In the end people saying "your cheating" by doing this or that are just dumb. If you cant understand that the game allows for more then two intakes per engine or that landing gears has no mass etc then thats just there "idea" of whats correct and not that they are trying to force up on other's. An Idea is not right nor wrong, even right or wrong is an idea made by humans so in the end its irrelevant. The game dont care, the game runs a defined set of mathematical and logical rules and thats all it cares about not any once personal preference. Most people whining about what is cheating and not seems to have some idea that the computer is bound to reality. But no anything goose in the virtual world and its up to the devs to define the rules with proper coding. People playing KSP in general have a big problem distinguishing between there personal preferences off what to do and not to do and what the game actually allows one to do. I have my personal preferences like every part should have its own physical space as close to a 100% as possible with some minor exceptions. But thats my personal preference. If you cant realize that your personal preferences might not be the same as others then its you how has the problem calling people cheats not the one being called a cheater. Calling some one a cheater would only be right if they are breaking the rules of a challenge.

Are the airospikes for landing or getting in to orbit? Two NV-1 should get it to orbit just fine or just add a third one in the rear. I would go for 4 jets and 2 NV-1 and maybe two 909 engines for STOL assisted landing. I land with two arospikes with my 200+ ton spaceplane on Duna (zsure its less when landing there but you know) so two should be more then enough for yours. Dont need 1:1 ratio just enough to lower the decent rate and also remember that you will be lighter and Duna has less gravity so it wont take much to get 1:1 lift to wight ratio. Remember your still using the wings even at lower then normal stall speeds.

Well I guess Im cheating then I use no-clip, not always but during some parts of the build. I started using it after the first Falcon X Jumbo so its not like I need it. But I realized that I knew almost every trick to "clip" with out the no-clip option. Most users have parts clipping even if they dont know it and even if they hate clipping. Squad made clipping harder in 0.19 or something, dont remember the version but before that it was super easy clipping any way. Most tricks still works with some tweaks but why fiddle with that when you cn use no-clip and save 4 hours of frustration? I only have rules that part should have there own space like engines or intakes etc. Very little overlap allowed. But every one has there own rules. KissSh0t, have you tried 4 jet engines as I suggested? I think you could get airborne with as little as 4, 6 at the most. There is a fine balance between adding more fuel and more engines. But if you can climb at 45 degrees from the runway at 50m/s or more you should have enough jets. Make sure you have enough lift to. Remember jets pick up power higher up and also if you cant run them at full trust where they are needed, at 32km altitude the less efficient they are since there trust to wight ratio is poorer compared to fewer once working at max trust. You want to be able to run all of them at full power at 32-33km no throttling down until you get to your target speed and start climbing. Im pretty sure 4-6 engines can get that bird to 2000m/s. One turbojet can hull 16-17 tons from the runway and hit 2000m/s at 33km altitude so thats a good estimate on max take off wight. Every tone less then that will give a huge performance boost but that boost is not needed if you cant do anything with it. Remember if your going to other planets, jets are dead wight so less is more as they say. But there is a point where two few jets will burn more fuel in tonnage then adding two more jets instead or more fuel. But you will easily notice this. You want to climb at 100m/s max and no lower then 20m/s the first 32km. After that when going for space vertical speed is good. Other then that your craft looks good and promising. More lift that previously and thats good. Less engines might seem harder to fly with but its more controllable and in the long run the only way to make the craft efficient enough and one have to get use to horizontal ascent profiles. Right mix of engines is crucial and often overlooked.

You want to land here. Was the old dry lake that was almost flat. Its hilly now but thats where I land in all my videos be it 0.21 or older. Still relatively flat compared to most places and its about 600m altitude so a lot denser air making lower speeds possible. You do not want to land HIGH up with a spaceplane on Duna. Decent rate and speed will kill you if you try that. And yes making corrective burns on the way to Duna or any other planet saves fuel from what I can see. I dont go for a perfect burn and I can later often correct using RCS only if you do it at the optimal point.

Who are you referring to? People need to learn to quote if there not talking about the post above them.

Restart have not worked well for me in some versions. Better go back to SPH/VAB and launch from there works fine. Never seen it go red either so.

I got a small one. 130 parts or so. Can try that one out. Problem with "medium" is that BIGGER is better. Bigger holds more cargo with fewer parts per ton and m³. Its the same law that apply to real aircrafts and ships. So even a medium SSTO hulling 1/4 the payload would still require 1/2-2/3 the parts if I build it the way I do to get the efficiency needed. Bigger it is the more efficient the engines are to since you have more intakes to run the last few jets on before going rockets. I could do a medium 500 part on but it would probably be crappy in terms of capability. What we need is more aerospace parts like bigger wings or some way of resizing wings with out mods to get part count down. But Squad quite frankly dont seem to care the slightest about spaceplanes. Baned suggestions on parts is just a way to say we dont care what you think. The new "what not to suggest list" makes me sick realy. Whats the point off the forum if every thing is baned that could be suggested? List would be sorter if they wrote what could be suggested. Some suggestions makes sense to ban but that was before they changed it and made it a mile long. If they dont like a suggestion just ignore it. There more interested in carer mode and that makes sense but still rockets get new parts here and there with every release but spaceplanes haven't gotten a new wing or a new landing gear not even a bug fix for the old on yet. A few wings cant take long. Even I know enough about Blender to spit one out in a few minutes. But I want a stock part for a game that I paid for not do it my self.

Using 1-4x time warps seems to almost reset Physics Delta requiring a restart or the game will crawl to a halt and show still frames. Tough not sure its related to altering the delta physics becuse I didnt have that problem with 0.18.2 or older running Physics Delta at 0.01. I never use 1-4x warp for that reason. had that problem ever since the switch to unity 4. I can run my 1100-1200 spaceplane on a stock i7 3770 with the integrated HD4000 graphics on most graphic settings on the lowest and some on medium and 1366x768 and its as fast as my i7 3930K and GTX570 so if you dont care about graphics HD4000 will do fine.

I was under the impression that the physic engine dot "drop" calculations. I mean you must finish it or it just wont make sense. I was under the impression physics works similar to frames where frame rate is a resolution. Lower the Physics Delta and fewer irritations (calculating the interactions of the parts, stress, collisions etc) are made etch second meaning theres less resolution on the physics part of things. If physics where not completed for etch physics frame if thats how it works odd things would happen I would assume. I personally cant see any difference in the games behavior (bugs etc) with 0.1 or 0.01. Yes one can set 0.01 manually in the settings.cfg and I have used that for the past 6 months. Runs a lot faster but so fare no bugs or odd things that would not happen with 0.1 either. I realy dont see the point of using 0.1 or even higher. Physics Delta seems to be the least used but most effective option to gain performance. i dont know what CPU your running but a stock i7 3770 or a i7 3930K @ 4Ghz and the difference with 1100-1600 parts is still frames and super low real time VS playable frame rate (Still slow tough) and acceptable real time preformance. There is also the question of Real time rendering not just frame rate and the game tends to run a lot slower with high Delta Physics when it comes to real time with many parts. EDIT: you should also write out what the Axis on your graph is, fps, time, parts etc so we dont have to assume. Especially if where not native English speaking.

I try to use one big engine instead of a small one to save parts if possible and wight. Also RCS is good to have when landing since authority will suffer as the craft moves slower and slower. Also land at the lowest possible area on Duna. That would be the old lakes. The dark low patches around the equator. About 500-600m altitude that is. Air is MUCH thicker there then say at 1800m. I can probably loose another 1/3 of the required landing speed that way and also decent rate will be much lower at 600m then say 1200-1800m.

I dont realy like to clip in to parts so I would not put intakes inside a fuel tank since that wold never work IRL at all and it makes it to easy, more challenging to find space to place intakes if they have there own space. I stack them with spacers of some sort to avoid excessive clipping to at least keep it some what real. I dont mind intake a abuse for a number of reasons. The current system is unrealistic in so many ways any way. But I do consider them some sort of compressor stages so an extension of the jet engines. I also have nacelles with 3m intakes give or take to imitate skylon like crafts. I would say landing performance on Duna is at least half as bad. I have landed close to 20m/s STOL style on kerbin with all fuel remaining and the closes I landed on Duna was 38m/s but one have also burned off about half the fuel at that point making the craft lighter. 40m/s is good for Duna after that it gets dangerous. I would recommend Vertical engines because if your heading for a hill and that happens a lot you cant pull up at all or slow the decent rate enough to avoid slamming in to it so hard that it wont explode. I have less then 1:1 trust to wight ration on Duna now. Before I had more then 1:1 and it was easier to land because I could avoid hills easier. Now If I am to slow one can only hop to reduce the vertical speed enough to compensate for the ground rising. But sometimes even that is not enough.

Asparagus staged ship I would go for. Will drop of tanks as it goes. I did that once worked quite well. I need to keep it small and efficient for my spaceplane heavy lifters. It do not dump engines like traditional asparagus just tanks. You can ignore the fact that I launch my craft from a Spaceplane. Can use a rocket to get it up if you want or build it in space. But what it dose its that it has a 1m core then 4 stacks of 1m fuel tanks that can be jettison in pairs during a burn with out losing any engines. I went to Vall with it and back with plenty of fuel to spare in the center fuel tanks so It should go anywhere I would say. It works similar to the appolo missions. Command moduels stay in space. Lander and rover descends to the planet or moon then returns and crew transfers back to command module and burns for home.

0/10 First time so Cant say I have seen that user.

Well I suck at navigating to I use pictures on the forum to get to laythe the first time. After that I took some screenshots of Duna intercepts. But I always manages to loss my pictures But still it gets easier with time. I spend so much time building stuff that I get so rusty navigating at times I have to check my notes

Did you aim for Jool first? Jool should be at 3 o clock when kerbin is at 6 o clock if I recall. That should get you to Jool. Once you have en intercept with Jool and have left kerbin's SOI then try to adjust for laythe intercept. But I find it harder to get back from laythe. Most efficient way I have found seems to be to have Jool at 6 o clock then Laythe at jools 3 o clock then slingshot around Jool by plotting the course from about 3-4 o clock on laythe. Saved me about 100m/s DeltaV compared to a direct burn for Kerbin. Kerbins optimal position is harder to predict since it rotates so much faster around the sun then Jool so one have to figure that out with the closest intercept and target position markers. BTW do you have Scatter Dencity set high in settings? Dont know how much it will hit performance but it determents how many rocks, trees and stuff you will see randomly. I run 20% just to keep performance good just in case. Think it will mostly effect low end GPU's tough.

In a thread I read some one found out its a bug. Why SQUAD have not fixed it ever is well beyond me. When they do I hope it will be like 100Kg not 500Kg. Lets hope there will be medium and heavy landing gears to. I realy need taler once. Adding a belly sure looks cool but well its part heavy as any other work around so more landing gear options is needed. I cant extend my belly forward on my latest SSTO because CoL is to sensitive in the nose to such changes so I had to use the rover body as a spacer. Looks bad but was the only way. A medium twice as long landing gear would do wounders on many spaceplanes. Just look at real high speed crafts or SSTO spaceplane concepts. LONG landing gears. From what I understand about the physics in KSP is that it works like frames. Just like you have a frame rate, number of still frames that you perceived as smooth motion assuming there is enough of them the physics works in frames to. At what number of frames per second I dont know but since I have seen no ill effect, like one part flying trough another with out collision detected I assume its enough. One could theorize that if to few physics frames where calculated very second parts could pass trough one and another with out the game detecting it. You could in theory the fly trough Kerbins ground. But since I never had things like that happen with 0.01 and cant see a difference in game behavior be it 0.03 or 0.10 I figure its a good trade off for better performance. So unlike what some theorize that you cant reduce physics load because that would mean not all the physics is calculated. Well thats not true if physics is completed for etch frame but you just have less frames per second of physics for a total of calculations being less in total then default and more CPU time can be spent to get more frames of rendering instead giving higher real time rendering and better FPS. Makes more sense and since you apparently can change the physics load and stuff still works this makes more sense.

Yea I cant understand why people dont use that trick. First thing I do in a new game is test all the settings. But I guess Im old school back when you hade to do it in MS DOS Back then tuning your settings was key to even play a game. But later years now game settings seems fewer and fewer so people seems to run what ever the game sets for them. Even when I tell people about it most of the time they think Im nuts and cant find the setting or file or they tell me they run higher physics time because they like accurate physics. Not like physics get more realistic just more accuret in theory at least. Real world benefit in the game tough that can be discussed. Well I prefer if I can even play the game first. Cant say I noticed any accuracy problems but I dont try to dock with a relative speed of mach 1 any way so. Not useful slamming in to things. EDIT: Highst speed I landed on duna Was 60m/s but it took 4 attempts and I broke 4 of 8 engines when I landed. Lowest safe landing speed was 50m/s but I would say 40m/s or less is good. Lots of wings, rocket engines and chutes is required in some combo. You just dont have time enough to set down safely and break if your doing 80m/s. 90m/s was perfectly doable before 0.21 but is no longer unless you can find 1-2km of perfectly flat ground. So yes you need a plan for landing taken in to the design. I prefer the STOL approach. Less cost in tonnage,fuel and parts. Having the rear higher mounted landing gear in the rear also helps to save the engines from damage on Duna. So thats an added benefit. And dont be afraid of adding landing gears. They actually have no weight. Tough SPH/VAB will alter CoM as if they did so add landing gears LAST because CoM will be where it is with out them. If you add them to soon and adjust CoL accordingly you will not have CoM where you think it is relative to CoL. So yea you can have 100 landing gears they wont slow you down since drag is proportional to weight and they have 0 weight so no added tonnage or drag. I dont know if tonnage is added or not to the map view of the landing gear. I would believe they are not but I assume they are just in case

Well you can reduce part count by preferring delta wings. They have the most lift. Also less engines and need for fuel means less parts. I prefer as big and few tanks as possible. Also changing Physics delta time to 0.03 in settings, the first page that comes up or edit the settings.cfg and search for 0.03 after you changed it in the settings menu and lower it to the lowest the game will except, 0.01 that is. Will significantly increase performance sens less physics frames are calculated. Devs seems to think 0.03 is the lowest reliable value. I have used 0.01 for 6 months easy and no difference in game stability etc just a lot better performance. I run an i7 3930K @ 4Ghz on all cores via the bios. I tested KSP on an i7 3770 that runs at 3.4Ghz and up to 3.9Ghz with intels turbo boost and from my testing it could continuously run 3.7Ghz with turbo boost with all cores so I would assume it ran close or at 3.9Ghz stock with no Overclocking in KSP since KSP very rarely taxes more then two cores and windows 7 or newer should know to tax fewer cores if possible to keep frequency up in programs using few threads. The i7 3930K @ 4Ghz used a GTX570 and 32Gb ram and ran 1920x1080 with most things on the highest and no AA and I rarely use reentry effects for my own playing because its to slow. The i7 3770 that ran at its factory speeds of 3.4Ghz and up to 3.9Ghz with one core in use used the Intergrated HD4000 GPU and ran at 1366x768 with most but not all settings on the lowest. Took some fideling to find out what was GPU hogging and what was CPU hogging but I got it to the pint where the HD4000 was not a limiting factor but rater the CPU. Both these CPU's benchmarked the same in KSP on my 1200+ part SSTO's. So basically any realy fast i5 would do. A 2500K would do just fine with its base clock of 3.2GHhz and it can easily be overclocked. I have used a lot of AMD processors and realy didn't start using intel until the i7 came out, well I had a C2D E8400 @ 4Ghz for gaming but never realy used that rig so sold it. But yea any modern i5 would do like 2500K, 3570K, 4670K would do. Older once can probably be found on ebay with MB and all. Theres no benefit from more then two cores atm so Hyper Treading (SMT technology) is also not useful. But there are no unlocked or realy fast dual cores from intel atm and there not much cheaper ithere counting the cost for motherboard, psu ram etc and they cant be overclocked. I would avoid AMD FX processors. Not that they are bad there just not very good in programs requiring high IPC per core. Bulldozer and piledriver where suited for servers where IPC per thread is low so theres less need for high IPC per thread and integer performance is the important part thats whey they split that in to two separate cores. Can be argued if its two cores. I dont think so sens theres no duplicates of fetch,decode and other minimum parts required for a CPU to work. AMD will include duplicate Fetch and Decoder in the successor to Piledriver. Intels single thread performance is just way above AMD atm and thats sad. I have used many AMD processors trough the years. K6, Athlon, Athlon XP, Athlon64, Athlon64 X2, Opteron etc. I still use two Athlon II X2 240E processors with DDR3 ram so there about Phenom X2 performance at the same clock. 2.8Ghz that is. But yea find this line in setttings.cfg in your KPS directory. Defualt its 0.10 but I run 0.01 the lowest that the game accepts. Lower has no effect. MAX_PHYSICS_DT_PER_FRAME = 0.01 should be 15-16 lines down from the top or so. Should give you a nice boost if you dont have to high graphical settings for your GPU. If so lower those settings but I suppose you have already made sure your not GPU limited so should not be a problem. I could not run over 800 parts with out this.

I had to go vertical until 15K then turn and I leveled of at 34K, about 2K to high but I got up to 2040kms I think surface speed. Then I went 25-30 degree nose up and throttled down accordingly until I hit 44km and basically had them running on idle. The trick to an Efficient SSTO space plane is flying it like a plane not a rocket. Going Vertically Requires over 1:1 trust to weight ratio and aircraft dont have that. Also the jets in KSP dont realy give full trust until 27-28km altitude. So what you want is less then 1:1 ratio and to be able to fly you need wings. Lots of wings. Going Verticaly means you have more trust then you need later on so its not a good thing needing to lower the trust before you hit your target speed say 2000ms because you cant feed them with air. Also more engines means you burne more fuel in space hulling them around. If I would use the same principal on my latest crafts that would mean 24 jet engines and thats 1.2T per engine. That would mean going to Laythe with that craft would completely remove the ability to hull a payload there. You should aim for a 45 degree angle of attack when climbing the first 15-16km and you should have a smoke trail going 35 degree or more fallowing the aircraft the first few km of climbing. If your climbing to slow and the smoke trail is to flat you need more wings not more engines. The smoke trail will drop of closer to 10 degrees around 10Km+ so it looks a bit ridiculous when the craft is pointing 45 degrees up but later at about 15-16km if you have enough lift and right engine trust you will see climb rates increase when engines pick up power and then you should drop of to 35 degree angle of attack at 16km+ altitude and when hitting 20km it should be more like 30 degrees. By now your vertical speed should be quite good, maybe close to a 100m/s. Then at 24km I go 20 degrees and then at 30km you would go 15 degrees and finally 10 degrees or less at 32km to level off and keep climb at about 5-10ms. Thats just a rough ascent profile, etch craft is unique but thats a good design target from my experience when you build the craft. Some math geeks might have better suggestions, mine are from try and error. But you basically want to efficiently climb to 32-33km and level of there because at that altitude you have low drag but still enough air to power all engines at full trust with the right balance of intakes. Pick up as much speed as you can on jets, 1900ms is good more is better. I try to hit 2000ms, 2050ms if possible then I pitch up to 20-30 degrees slowly not to flame out. Then you will soon get 100ms vertical speed, throttle down and if you can shut down engine pares as you go that gives better control on the remaining engines throttle and you can then throttle them up to compensate. Once you get to 40-44km altitude go Nukes and if theres air to power 1-2 or more jets at full trust do that. I even run 1 jet asymmetrically on my big crafts since its close enough to the center of mass. But dont go to much nose up when you run rockets and jets because that reduces the time you can run jets. Try to get periapsis up to a few km to. You can then offset the drag by running 1-2 jets on low trust, 10-2kN etch to avoid losing to much altitude on apoapsis as you are costing up in to space and orbit. You can also run nukes at partial trust if your jets still provide decent trust since they are more efficient on fuel use. A good tip is that the same engine will always flame out so first time you fly, you flame out intentionally to find out what engine that is. If that engine is in the first action group of jets you shut down you do the same for remaining engine groups in your action groups. What you then do in your real flight attempt is to right lick that engine and watch the engine status to see if it flames out. You can predict flame outs by watching the value in resources that is with in the parentheses of intakeair. Positive number means more air is taken in then used. 0.0 means you use as much as you take in. Negative number means your using more then you take in so the small buffer you have will soon run out and you will flame out. So to fly safe you can throttle down if you see it going to 0.0 or to -0.01 so you get up to a positive or at least neutral value (0.0). if your like me and wants to cut it close you can wait until you get en engine flame out message on the engine status and throttle down then and engage RCS if needed to stop the spin. This few tips will save you a ton off fuel that can be used for payloads. But yea the big thing is to realize jets are MUCH more efficient, more then there numbers indicate in SPH. Second important thing to realize is that lots of wings is the trick to get efficiency with those jets.They go hand in hand and the right balance will give you a tone off new capability. Big payloads or just more efficient interplanetary crafts. You want lift the first 30km or so after that its more or less ballistic. But you get a lot of wings for one engine in terms of weight. Downside is high part count but engines in KSP is not very nice performance wise for your PC either, smoke takes a lot of computer power with many engines. But its still less costly to hull dead wings around then dead engines. If one prefers Vertical climb and Gravity turns wings are pretty useless realy. Basically good for landing but even that would take little fuel to do vertically so its better to either go vertically and no wings or all the way in with wings and a good ascent profile on jets that is based on horizontal flying with wings. It took me a few months to get to that conclusion but thats the way it seems to be. But any way your craft preformed very well and better then my crafts did back when I built the Falcon IV Transporter. I actually built a replica of the X-33 that took of vertically but landed horizontally that was what lead me in to the conclusion that horizontal take of with lots of wings and few jets where the best way to get efficiency. This youtube video if you have not seen it shows how I do my ascent as I described above. I edited it in such a way that you can see etch change in pitch and also engine configuration most of the time. http://youtu.be/KtFSIWnvfmk The penalty of using brute force with engines will increases the further out in the system you want to go sens it reduces your delta V hulling dead wigth. So vertical take off makes some sens for LKO missions even with payload but going to Duna or Laythe its quite costly on the payload fraction I would say. I hope this helps with your next Project. The last tip I have is to try one thing at a time and not all ideas at once. That way less will go wrong. It takes time to get all the features in to once idea of an ideal SSTO spaceplane. Thats why I build a lot of prototypes and new versions all the time. Trying to master all concepts at once is just to much. But your doing very good so you will do some awsome SSTO's Im sure of. Its all practice. I dont know how much time I spent building my crafts but its to much thats for sure