Zeiss Ikon

Members
  • Content count

    1,048
  • Joined

  • Last visited

Community Reputation

720 Excellent

2 Followers

About Zeiss Ikon

  • Rank
    Senior Rocket Scientist

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. Zeiss Ikon

    What did you do in KSP today?

    With the recent startup of my "Take Six" career in RSS/RO/RP-1/Principia, I bypassed the whole "burn the fins off WAC Corporals" stage by starting with A-4 rockets. The first two launches were unaltered A-4s (aside from ballast replacing the mass of warheads that hadn't been installed). With thermometers and barometers mounted to the guidance unit, they collected science while their batteries lasted (not much longer than the powered ascent), with the results returned by radio. The data gathered, or course, was applied to improving the technology, and after but a few launches (which, none the less, took some three years), upgrades were available for both the booster and by then, the upper stage (based on the original WAC Corporal). Then the failures started. Lulled by the first couple launches of the A-9 configuration upgrade engine with XASR-1 powered upper stage (but still in a sounding rocket form -- long and narrow, with fins and a long ogive nose, although stability in the thin air at 30+ km was augmented by spinning up the booster before staging), a version with a short, fat tank and spin-inducing ullage motors that allowed for much larger sounding payloads (for contracts) inside a smaller fairing saw four engine failures (ranging from simple failure to ignite the A-9 engine on the ground, to a combination of reduced thrust from the A-9 and failure to ignite the upper stage) in four launches. Fortunately, the penalty for contract failure (in terms of funds) is only the same as the advance, so all I was losing was the cost of the launchers and their rollout (and considerable prestige), but after the fourth failure (with construction taking too long to permit a second try on contracts accepted before construction started), the program was down to barely 30,000 in December of 1954. Finally, I managed to complete a couple contracts, and with accumulating flight time, the A-9 engine started to settle down. Now, in January 1955, it's time to send the astronauts to the edge of space (mainly, after the first contracts for crewed supersonic and crewed suborbital are paid, in order to keep them from retiring in 1957).
  2. Zeiss Ikon

    BOINC

    In fact, some BOINC projects do recognize the owner of a computer (or the team, since users can organize into groups) that actually found something. Einstein, for instance, has given co-author credit to BOINC participants in their papers -- SETI has said they'll do so, if they ever find a signal. And some projects exist with more immediate human benefits than Einstein or Milkyway -- Folding, for instance, seeks to "solve" the protein folding problem by using massively parallel computing to predict folding of proteins with known final shapes, in order to find an algorithm that can predict folding without knowing the final shape via x-ray crystallography or similar methods. Solve that, and it'll be possible (for instance) to create designer drugs to fit a particular receptor site without lengthy trial and error -- just write a stretch of DNA that codes for the required amino acid sequence, and insert it into a strain of E. Coli (as was done with and actual human gene decades ago for insulin -- but in this case, the gene would be written from scratch) -- and do it at reasonable cost (more like the cost for a single treatment procedure than that to set up a pharmaceutical production process). If you don't want to participate in BOINC or other distributed computing projects, then don't -- but relevance is a value judgment anyway. I consider mapping the Milky Way's stellar populations to be relevant, because I expect mankind to need those maps eventually; I hope my mind, at least, will survive long enough to need them myself.
  3. Zeiss Ikon

    Post your Para Flyers here !

    Okay. I created this one for a challenge a few months ago, when Jeb was sneaking parts out of the bin to get as far from KSP with as few parts of as little mass as possible. This one is eight parts, not counting the stuff that got left on the runway (Mk. 1 Command pod, extra fuel tank so the engine could run up without depleting the actual flight fuel, and the other half of the radial decoupler), Jeb and his EVA parachute: There was a simpler version, too, but I never completed the whole "Far and Light" flight with that one -- I just deleted the landing gear, so it was small circular intake, Okto 2 mini probe core, .625 m stack battery, Mk. 0 Liquid Fuel tank, Juno, and a Command Seat on top, plus the launching rig. If you try to fly one of these, keep the throttle low -- it only takes about 1.2 kN to climb slowly at 20+ m/s when the tank is full, and that requirement will drop as the fuel burns off. Punch it harder than that, and the jet/seat assembly will spin out of control under the parafoil even despite the reaction wheel in the Okto 2. I could probably upload the .craft files for both versions if anyone cares... BTW, that land you see ahead is the next peninsula east of KSC -- Jeb flew due east for about three hours on the 50 units of LF in that little tank, and then landed deadstick with dry feet (barely). The no-gear version, flown as high as possible, could probably add at least 10% to that distance.
  4. Zeiss Ikon

    BOINC

    You're not quite correct on resource usage, @Gargamel. If you run one project on GPU only and another on CPU only, they'll run side by side (though Einstein GPU work now requires a full CPU in addition -- it used to be about 0.2 CPU). I've just changed my BOINC settings back to shutting down when CPU usage exceeds 10% (which would be around 80% on any one core, indicating a likely game session), but when I'm not playing KSP, I get seven MilkyWay tasks and one Einstein task running in parallel. And @kerbiloid, the point of BOINC is it runs on whatever you have. For a short time a few years ago, I was running BOINC tasks on a 300 MHz Pentium II (they completed just about fast enough not to expire, if nothing caused a delay) on an old Win98 computer (revived with a lightweight Linux distro), alongside my desktop machine (then a Core2Duo 2.4 GHz). If I have a machine I leave powered up all the time anyway (either because it's acting as a server, or just so I don't have to wait 2-5 minutes for a startup when I want to use it), it ought to be running some kind of background task to do something useful in its idle loop. There are multiple distributed computing controllers now; BOINC is just the biggest one (as far as I know). If you don't find a BOINC project you want to support, you might Google for other projects. I think there's even one that does collective blockchain mining -- a single PC is too slow to be worth the power consumed as a dedicated Bitcoin rig (never mind Ethereum), but once again, if it's going to be left running anyway, there's no good reason not to let it do something in the background, and getting a share of the blockchain production of a cluster of a few thousand of those, for energy you'd use anyway, is like getting a discount on your electric bill.
  5. Zeiss Ikon

    What did you do in KSP today?

    I have those in RO, too. Cheaper than a WAC Corporal (because captured parts, I presume), and lifts a lot more (and a WAC as second stage goes really high).
  6. Zeiss Ikon

    What did you do in KSP today?

    I thought I recalled someone having built some basic logics (AND, OR, NOT) with solid parts in LKO? Nice! Is that just the historical parts added to a stock game? Burning Lf/O?
  7. Zeiss Ikon

    BOINC

    My opinion: most of what [email protected] is searching for, LIGO will never see. They're mapping things like new discoveries of neutron stars and binary neutron stars; LIGO apparently can only detect merging black holes. Very different things. Where Einstein might find gravity waves (if it ever does) is in short-term changes in period of binary neutron stars (potentially with sub-one-second orbits, detectable by watching the Doppler shift of emitted radiation). I don't have other needs to soak up my CPU cycles. I gave up SETI some years ago, before BOINC provided a common control for these distributed computing tasks. MilkyWay, however, I feel produces more usable results than SETI (especially with recent reexaminations of the Drake equation suggesting we might well be the only advanced civilization in the Milky Way at this time) -- presuming the Drake factors like "average life of a civilization" don't get us, we'll be using that information for the rest of time. Even if we never leave our own solar system, we'll use the information on which stars belong to what populations (i.e. originated from which merged dwarf galaxy or full size galaxy, leading to where they're headed and how fast) to inform astronomy and cosmology for as long as we keep looking at the stars with a scientific eye. If we once become starfarers, we'll need to know that information whether we take a million years to cross the galaxy, or do it in a single lifetime (after inventing a workable warp drive).
  8. Zeiss Ikon

    Who's playing Realism Overhaul etc.?

    Exactly. The more parts mods you have, the more RAM you need. I was just looking at the empty nodes on the right hand end of the tech tree last night; it looks a lot like either I didn't install some optional parts mods, or there's a plan for stuff that hasn't yet been selected to be part of RO. Either way, I'm not too worried; by the time I can unlock enough nodes to care about those, they'll be filled in.
  9. Zeiss Ikon

    Who's playing Realism Overhaul etc.?

    I will guarantee you that both Muroc Dry Lake and Delamar Dry Lake (both used as landing sites for various air-dropped X-planes at various times) are flatter and smoother than the grass north of the runway in my RSS/RO/RP-1 install. I've now tried launching even fairly light-loaded airplanes from the grass, and landing them there. Both are prone to issues. Sure seems to me that even in 1951 the capability existed to pave a 3+ kilometer runway smoothly enough that it won't throw your airplane into the air if you hit a seam at 10 m/s.
  10. Zeiss Ikon

    Lets play KSP, or wel just forget it.

    I don't have gkrellm installed -- and like most other things that come up in Ubuntu, there are fifteen or twenty tools to do a particular thing, several of which won't work in the DTE you use, several more of which won't tell you the exact, specific thing you need to know (even though their description in the package manager makes it seem like they do), and one or two of which actually do what you want -- if you can find them and figure out how to make them dance (at least one of those is a command line utility ported from BSD in the 1990s, and you need extremely cryptic command line arguments to get what you want). I run Ubuntu Mate 16.04, and use htop most of the time for that sort of thing -- it's preinstalled, text-mode, runs in a terminal window. It shows CPU load and RAM usage, broken down by process. I've also got a Conky on my desktop that shows overall load individually for my eight cores, plus disk I/O, and a frequency monitor in my top status bar (individual clock speed for each core). Problem is, the CPU monitors generally report all cores at 100% because I also run BOINC tasks ([email protected] and [email protected]). They're prioritized so they don't (shouldn't) get in the way of any other apps, but they soak up any unused CPU capacity. That and load balancing (switching process from one CPU to another) makes it hard to see if it's just KSP saturating its one core, or the fact I'm using SATA instead of a direct PCIE connected SSD. For what it's worth, though, my system gets yellow clock with about 1/10 the vessel Nathan Krell's does in his videos, and starts KSC at about 1/10 the rate as well -- suggesting all I need is a 40 GHz CPU.
  11. Zeiss Ikon

    Lets play KSP, or wel just forget it.

    @soulsource That's interesting -- do you know if that bug also applies to Linux? And how would I tell if that's affecting my load time? Even at 4 GHz and loading from SSD, my system takes three-plus minutes to load the stock game. Not a Steam purchase, if that matters; I bought it direct from Squad.
  12. Zeiss Ikon

    Who's playing Realism Overhaul etc.?

    My desktop machine has an 8 core AMD FX-8350 with maximum 4.1 GHz turbo, 16 GB RAM, and everything that matters is on SSD, but with full RO (including E.V.E., most of the parts mods, and Principia) I get yellow clock even when my craft are down to 8-10 parts (a WAC Corporal, for instance, is 20+ parts at launch). It takes me about twenty to thirty minutes to do an orbit launch that runs up just over five minutes of MET. The thing is, more cores/threads doesn't help in the least; like a lot of games, KSP does virtually everything in a single thread (avoids having some parts get out of synch with others). An Intel Core processor the same speed as the AMD I have would probably be adequate to keep my clock green most of the time, because Intel processors get more done on each cycle than AMD FX -- but I was on a budget when I last rebuilt my machine, and it's still a big step forward from the Core2Quad with 8 GB that I had before. I've watched in htop (a Linux utility that tracks memory and CPU by process) as KSP loads, and my RO install runs up to 12 GB RAM at peak, then settles back to about 10.5 once everything is loaded. You could probably load it in 12 GB physical RAM (the OS will swap stuff out), but that would kick my near fifteen minute load time up by another factor of two or more (swap is SLOW compared to RAM). It might be worse on Windows, or it might not (I have virtually no experience with Windows 10).
  13. Zeiss Ikon

    Who's playing Realism Overhaul etc.?

    Yeah, @CaptainHaywood it does put some demand on your machine. You need a minimum of 12 GB RAM (16 GB recommended, since I'm still not running all the parts mods), and 4 GHz isn't any too much processor speed. My laptop runs stock KSP pretty well (until it overheats due to cooling problems), but it hasn't got enough RAM and only 3.4 GHz maximum turbo.
  14. Zeiss Ikon

    Lets play KSP, or wel just forget it.

    I've got a 4 GHz CPU, 16 GB RAM, and everything that matters on SSD -- and it still takes just under fifteen minutes to start. Of course, Realism Overhaul's many mods add a lot of parts, and then there are the 55,856 patches... Speaking of which, I just launched the game to load in the background while I read the boards.
  15. Zeiss Ikon

    Logical tech tree

    Oh, I agree. Even in the basic game of 1.2.2 or 1.3.1 (the first two versions I played), you somehow have the Swivel gimballed engine before the Reliant with its fixed mount. In real history, some of the earliest engines had thrust vectoring, but it was done with jet vanes (A-4/V-2, as well as the A-6 and A-7 Redstone/Juno/Jupiter family and the Russian ethylox RD-100 through RD-103M). Aside from jet vanes, fixed engines came first, often with verniers added for control. Restartable engines were another innovation -- the X-1 could restart, sort of, and the X-15 could, but aside from these X-plane engines, restarts came with Lunar and interplanetary missions, and initially were hypergolic engines only (ignition was easy -- just start the pumps or open the valves). Throttling for boosters came in with the Space Shuttle; SpaceX may have been first to use it for uncrewed missions (okay, i just remembered, Delta IV Heavy and the original Russian boosters with twenty main chambers firing at launch throttled the cores to help them burn longer than the side boosters, but SpaceX goes to lower thrust than the others). Otherwise, "deep" throttling was down to 50% like the LEM descent engine (the XLR-11 that propelled the X-1 could throttle to 25%, but it did it by shutting down three of its four chambers). And reaction wheels are something you use to overcome solar wind and light pressure on your deep space telescope and conserve RCS propellant, not something that can turn a multi-tonne rocket in flight. The original RP-0 was created to address this -- airplanes before rockets, uncrewed before crewed, low-tech first.