Enthalpy

Welcome to Chang Zhen 7 (or CZ-7 or Long March 7), a new 2.5 stages Chinese launcher with good oxygen and "kerosene" engines everywhere. https://fr.wikipedia.org/wiki/Longue_Marche_7 http://www.b14643.de/Spacerockets_1/China/CZ-NGLV/Description/Frame.htm It's reported to put 13.5t or 7.5t, with 4 or 2 side boosters, on a 200km x 400km x 42° Leo and may get some day a solid or hydrogen upper stage for higher orbits - here is instead an electrically pumped kerosene upper stage for it. The per-mission sized Li battery weighs 20.8kg per ton of propellants for 60bar in the chambers. 23kN thrust let reach a transfer to Jupiter from an elliptical orbit. Four D=1m niobium nozzles expand to 178Pa to achieve isp=3894m/s=397s . The engine accounts for 120kg. The adapter (98kg plus 20kg separation) and the frame (132kg) are hexagonal welded trusses of AA7022 tubes machined to typical L=854mm, Ri=30mm, Ro=31.5mm - stronger at the payload, lighter at the engine. Rolls guide the adapter like at Zenit. The tank (53kg) for maximum 2542kg Rg-1 is of 1mm aluminium welded on the frame. The tank (35kg) for maximum 8264kg oxygen comprises 100µm brazed Maraging steel, 15mm foam, multilayer insulation, and polymer belts to hold it at the frame. A polymer net at the frame shall stop objects falling on the tank. The payload belt shall weigh 5kg, controls 100kg, unaccounted items 50kg, totalling 495kg dry mass without the battery and after adapter separation. Pleasant 66.6kg/t of propellants including the battery. The fairing covers this upper stage too. It should be longer and also wider, because the mass capability to Gto now rivals Ariane, Falcon, Atlas and Delta that host D=4.572m, and also to ease my solar thermal engine. The Gso capability makes payloads sooner profitable than ion engines can and is efficient. Marc Schaefer, aka Enthalpy

News report that Launcher One is to grow to put ~300kg on sun-synchronous orbit, and also that the French Onera targets the same kind of payloads - not only because Americans do it , but because new satellite operators want to put huge constellations in orbit. And, mamma mia, Onera too wants to start from an aeroplane despite Europe has the perfect spaceport in Guyana towards East and North. Worse: they want to use hydrogen peroxide. Porca miseria, do they have anything agaist nitroglycerine? And then, of course, turbopumps. Well, my pressure-fed equivalent of Tronador can, without modification, also put 290kg on a 800km sun-synchronous orbit. Suggested there http://forum.nasaspaceflight.com/index.php?topic=26645.0as reply #23 on 07/18/2012 but it's very close to my pressure-fed equivalent of Launcher One suggested there http://saposjoint.net/Forum/viewtopic.php?f=66&t=2554&start=50#p38614on Jul 16, 2012 and anyway, here's the drawing again: Two stages, starts from the ground, made of cheap materials, burns safe propellants, and decently efficient. I don't really find an excuse to have pumps at that launcher size, but at least, Rocket Lab http://www.rocketlabusa.com/index.html does it in two stages from the ground with batteries and electric pumps as I suggested there http://www.scienceforums.net/topic/73571-rocket-engine-with-electric-pumps/ not very costly, and it does improve the performance.

Reassigned concentrators help to transmit data from Martian orbit to Earth. The probe can only target one ground station. At 1.524AU (Sun-Earth distances), seven D=4.572 dishes concentrate 64kW light on solar cells of varied bandgaps. Collective 50% efficiency and 50% eclipse time by Mars leave mean 16kW supply to feed the transmitter half of the time with 7000We for each polarisation and phase. Again a 4D constellation, cubic for simplicity, with amplifiers 40% efficient at full power, transmits 5600W=+67.5dBm per polarisation and phase at the strongest symbols. Seven D=4.572*80% in-phase dishes radiate at 50GHz with +74dBi gain. At mean 1.823AU=273Gm, free-space losses are 284dB. A D=24m*80% reception dish gains +80dBi. After 3dB losses, the ground station receives -66dBm per polarisation and phase at the strongest symbols. 100MHz available band permit 40MBaud and the noise bandwidth shall be 40MHz. 30K noise temperature make -108dBm for each polarization and phase. With 6dB margin, the full-power symbols are +36.4dB stronger than the noise, so each polarization and phase swings from -66 to +66 times the noise voltage. 64 points per polarization and phase are separated by 2.10 noise voltages, so a symbol has 0.30% chances per polarization and phase of being wrongly received. A RS(4095,3935) code on 12b symbols, which have 0.60% chances of being wrong, corrects 80 positions. A few sockets of 20-core Xeon decode it using Log and Exp tables, or better one Xeon Phi. 4*6b per symbol at 40MBaud and the 3935/4095 efficient code transmit, if Mars eclipses the beam half of the time, mean 58MB/s. Of R=3390km Mars, 14PB for a D=0.2m resolution full bicolour map is transmitted in 7.5 years. Minimum data compression improves that. Or if 200MHz are available, 80MBaud of 4*5b transmit mean 96MB/s. Faster, or more colours, or better resolution. Marc Schaefer, aka Enthalpy

Service Pack 1 for the Keyhole-11 to the Moon... ---------- The orbit's inclination must be changed before raising the perigee: 1120m/s. Sinking the aposelene costs 605m/s. This leaves 17274kg on Lunar orbit, or 627kg less. More hydrogen also increases the tank and insulations by 111kg, so the ~3200kg equipment bay loses 738kg. Still easy. The module but fits in the Delta long 5m fairing now. Though, the equipment bay is still heavier than needed, so the module doesn't need to exploit the full capacity of the Delta V 541 (even a 531 should suffice) so the tank shrinks. If needed, the Delta operator claims it can expand the fairing. ---------- Transmissions to several Earth stations simultaneously, less oversized and hopefully less wrong now... Two 2m*5m solar panels suffice and are easier to orient. 30% efficient, eclipsed 50% of the time, they provide mean 4kW, of which 2kW feed the transmitters. That's pessimistic because when eclipsed by the Moon, the craft doesn't transmit. Here 13 small ground stations at sites with usually clear weather are separated by >2400km to cumulate the throughput of several beams - better than exaggerating a constellation modulation: 6 stations are generally visible from the Moon, sometimes 5, briefly 4. Allowing for maintenance and cyclones, the craft sends 5 beams to chosen stations. Hard disk drives shipped by boat can replace high-speed cables. Each of the craft's D=4.572m 50GHz antennas has its first zero excentered by 600km at Earth's equator, so optimized sidelobes are weak at 2400km, and the beams well separated. Aperture synthesis could replace five reflectors but is uneasy at 6mm wavelength; distinct amplifiers for the beams, possibly to distinct and interleaved antenna elements, would then preserve the amplifiers' efficiency. I take five reflectors here. Each beam modulates two polarizations and phases in a 4D (not 3) constellation, 4D-cubic for simplicity, hence equivalent to 4 independent amplitude-modulated channels. Each of the 5*4 channels gets mean 100W supply, so 40% efficiency at full power lets it radiate 80W=+49dBm for the full-power symbols. Free-space loss is 227dB at 384Mm distance and 6mm wavelength, emission by 80%*D4.572m gains 66dBi, reception by 80%*D4m gains 65dBi, 3dB are lost, so the receiver gets -51dBm for full-power symbols of each polarization and phase. 100MHz available band permit 40MBaud; smooth symbol transition put the noise bandwidth at 40MHz. 30K noise temperature make -108dBm for each polarization and phase. With 7dB margin, the full-power symbols are 50dB stronger than the noise, or 316* the voltage. 256 points per polarization and phase between -316 et +316 noise voltage are separated by 2.48 noise voltages, so a symbol has 453ppm chances per polarization and phase of being wrongly received (no Viterbi nor treillis here, provided it exists in 4D). An RS(255,243) code to correct 6 symbols has then 5e-11 chances of being wrong, or one 100Mb image in 0.4 million. 5 beams at 40MBaud of 4 bytes per symbol and a 243/255 efficient code transmit, if the Moon eclipses the craft half of the time, 0.38GB/s - almost the sensor's throughput as its swathes overlap. From our R=1737km Moon, 26PB for a D=75mm resolution full map is transmitted in 2.2 years. Marc Schaefer, aka Enthalpy

Maybe the original query wasn't about statistical prediction, but about monitoring the health of an equipment to detect wear. In this case: yes, it's done - where possible. Ball and roller bearings are observed by measuring their vibrations. The remaining thickness of corroded steel vessels is observed by various means. Cracks in vessels can be detected acoustically. There are many means - too many for one person to have them ready in his head. Each means is very specialized. Some means work on the operating unit, others (aircraft engines, nuclear power plants...) need to stop the equipment and even deassemble it. Some industries (nuclear, chemical, airlines...) do it always, some (sea transport...) do it sometimes, some don't because an accident is cheaper there or because detection is impossible. ---------- But if the original post meant: "predict the reliability of a machine" - then no, sorry, this doesn't work. Beacuse failure results from poor design rather than from constraints, and no method on Earth leads to good engineering. This results from good engineers, good prototyping and testing, but most definitely not from quality practices. Even on strained parts where methods have been developed, no good reliability prediction exists. There are attempts for bearings, but for instance SKF just plainly writes "our experience tells that a bearing's lifetime is unlimited if used properly". The Hdbk-217 tries to predict electronic components' reliability but everyone who has made reliability experiments laughs at it. Woehler's curves try to predict fatigue failure due to alternate stress, but even the measurements on shafts are not repeatable. Possibly the least bad case is alloy creep at heat. And then you have software, for which the only valid prediction is that it will indeed fail, and you'd better have none.

Resupplying Mir and Saliout: up to now I believed they docked a (full) Progress or Soyuz to the stations, didn't transfer any propellant, and let the Progress or Soyuz with its own engines push the station. Is there a source that indicates a propellants transfer, in which case I'd be happy to change my belief? Moon's surface: it offers no protection against micrometeorites as there is no atmosphere. A very strong disadvantage of the Moon are its extreme ground temperatures, from >+100°C during the day to -200°C during the long night. That's only the ground temperature, from which you can insulate a craft during the day and adjust the sunlight absorption and heat emission - but night is damned difficult as you have no light. You need your own heat source, or a huge and compact tank (and already full!), very well insulated, to avoid freezing after 2 weeks without sunlight. Apollo missions stayed there only during daytime, Yutu carried a radioisotope heat generator (but failed at night I believe), and my scenario to bring lunar samples back http://www.scienceforums.net/topic/85103-mission-to-bring-back-moon-samples/ is temporarily stopped at this difficulty. There are solutions without plutonium, but they impose much of the craft's design, so they must be decided very early in the design. In orbit as opposed, you have sunlight to adjust a craft's temperature, and eclipses last for 1/2h or 1h, not 2 weeks. So in my eyes, propellants on the Moon's surface are right only if one has to go there anyway and needs them there. Even to go to Mars by producing oxygen on the Moon (provided this is cheaper, not just lighter), it's better to store the lunar propellant in lunar orbit, Earth orbit or at a Lagrange point, because the propellants to go there from the Moon's surface is already used, so the remaining tank is smaller hence lighter. Though, I've already been wrong on the tricky subject of propellant depots, so maybe some smart scenario makes use on an other advantage I haven't seen. Transferring propellants: superfluid helium was looking for trouble the big way. It's already counter-intuitive and difficult on Earth with a human operator. Sound fluids like oxygen and kerosene or spiropentane are muuuuuch easier. To separate the liquid from the gas (which isn't always the vapour! It can be nitrogen, helium...), I suppose one simple means would be a sort of centrifugal impeller that puts the liquid against the shell at the pipe's sucking end. Maybe the pipe's end can move through the tank to seek the liquid were it is, or we can make a gentle blow of the gas through the tank - that's nearly a vacuum cleaner. Or we let the whole tank rotate, if it's a mere depot, or let both craft accelerate. What about a fabric or a sponge that goes through the almost emtpy cylindrical tank to catch the liquid? It's not tight, so it can be thin metal fibres, easy at cold. I wouldn't have a membrane between the liquid and the gas, because I suppose it's impossible with hydrogen, quite difficult with oxygen, and heavy in any case.

Thanks for your interest! So do I wonder... About the Moon, I believe to understand that only a small fraction has been imaged with 0.5m resolution by LRO because transmissions limit the amount of data, so a complete and detailed mapping would be an improvement, be it at 0.5m or 75mm. Also, as Esa proposes to Nasa to settle a base on the Moon's remote side, having good pictures first would be advantageous. Mars, sure! This was my first intent with KH-11. What stopped me: apparently Altas V Heavy doesn't start from Vandenberg, so I couldn't get enough mass on a sun-synchronous (polar) orbit, and I redirected my effort to the Moon. This explains some features oversized for the Moon, especially the datacomms. Though, the description for a lunar mission is a good start for the same at Mars. Concentrate the resources on manned Mars: yes... I consider my sunheat engine is essential for it http://www.scienceforums.net/topic/83289-manned-mars-mission/ and the sunheat engine must be proven first on an automatic mission. Transporting an existing KH11 to the Moon needs little more than developing the engine, hence is a cheap test. This delicately megalomaniac undertaking is also a convincing illustration of the engine's possibilities. 18t is about what a Saturn-V launch put on Lunar orbit.

The sunheat engine can bring a decommissioned Keyhole-11 from Earth orbit to Lunar orbit to make better images there. As an evolving family of spy satellites, the KH11 is little known https://en.wikipedia.org/wiki/KH-11_Kennan and much data is inferred from the Hubble Space Telescope, supposed to use the same D=2.4m primary mirror and chassis https://en.wikipedia.org/wiki/Hubble_Space_Telescope so 240nrd resolution would separate 75mm at 300km distance from the Moon too, better than presently 500mm with the Lunar Reconnaissance Orbiter https://en.wikipedia.org/wiki/Lunar_Reconnaissance_Orbiter and improved datacomms would provide us clearer images from more sites. Secrecy is a hurdle, especially since the added module provides data storage and transmission, needing to know at least KH11's downlink format - but maybe it can be reprogrammed. The optical and mechanical data is a smaller worry. On the other hand, carrying the toy away avoids to deorbit it, and the saved propellant gives an operational life extension worth 100M$: incentive. And did I read that the Nro had already offered decommissioned orbital KH? Many orbservation and spy satellites orbit the Earth: Landsat, Spot, Helios, Lacrosse... I take KH11 as an extreme example, as most others are lighter. Some are also less secret. Or Nasa can use the 2.4m mirrors given by Nro to build a new, light Lunar craft carried there from Leo by the sunheat engine; though, I'd prefer to send these to Mars, an operation I may describe later. ---------- Mass, speed and propellant An Atlas V 541 puts a 13t propulsion and data module on the same Earth orbit as the chosen KH11, say 300km x 900km x 98°. The module deploys its sunheat engines, navigates to the KH11, and grasps it delicately for 40N push. I hope no electric contact is needed: the KH11 keeps its energy supply and transmissions, the module intercepts, processes, stores and retransmits the data. I estimate the empty KH-11 weighs 13t, for no good reason, and the aggregate starts with 26t. Perigee raised to 400km costs 28m/s. This action overlaps with the beginning of the next one. Apogee raised to 326Mm, the Moon's Lagrange point's distance, costs 2931m/s. The aggregate weighs now 19955kg. Perigee raised to 326Mm and inclination changed from 98° to almost Lunar orbital plane (1022m/s forward, 89m/s polar). Cost 1026m/s leaves 18373kg. The Moon is there at that moment and catches the aggregate in a 1737+300km x 58Mm polar orbit. Aposelene sinked to 300km costs 291m/s and leaves 17901kg on the 300km x 300km Lunar polar orbit: that's 4901kg more than the KH11, for the module's structure, engines, propellant rest and data tinkering. The 15 sunheat engines push 2.7N each. This permits Hohmann transfers, making them as good as the ion engine that offers more Isp but must spiral due to its fainter thrust. As the kicks extend for long around the periapsis, I take mean 90% efficiency for the 12424m/s ejection speed at steps 1, 2 and 4. The whole transfer takes approximately 15 months. ---------- Structure The D=4.572m concentrators travel stacked horizontally. Each weighs 25kg. The 115m3 tank for 8099kg hydrogen is mean 4.4m wide and 8.5m tall. Welded AA7020 tubes machined to L=1.18m Ri=32mm Ro=33.1mm make its hexagonal truss (200kg) with 12 nodes per turn while AA7022 sheets machined to e=1.3mm except at the seams make the skin (500kg), welded at the truss for the cylindrical part and unsupported at the heads. The truss could be twice finer-grained and the skin have integral ribs. 30mm foam (180kg) let 37kW leak in. Over 15 min launch operations, the hydrogen warms by 0.6K. 50 plies of 13µm Mli (250kg) let 14W leak in, evacuated by a ~600Wm cryocooler. At each tank end, a truss of aramid or glass fibre holds to the tank. With 12 nodes per turn and 2 stages each, using L=1.18m Ri=35mm Ro=37mm tubes, they weigh 180kg together and leak <2.5W together. 1310kg insulated and supported tank and 375kg engines leave 3200kg for the equipment bay and optional hydrogen to manoeuvre in orbit. ---------- Data processing, storage and transmission Let's take 30,000 pixels wide images for KH11, with 2*12bits colours. At 75mm resolution and 1554m/s it must read 20700 lines/s, slower than at Earth: either the module averages several lines, or the KH doesn't tilt its view when observing - too little is known here. This produces 1.9GB/s, faster than can be transmitted. 5000 chips of 32GB Slc Flash store 24h worth of uncompressed sensor data. Easy to increase. On Lunar orbit, 6 concentrators are reassigned to solar cells of varied bandgaps through filters, as suggested here on Sep 20, 2014. 40% conversion provide 53kWe on dayside, of which 20kW=+73dBm transmit data. The battery for 53kW*68min weighs 470kg. 6 other concentrators work as transmission antennas. Each gains 62dBi at 30GHz - choose a frequency with muuuch available bandwidth - or 70dBi phased together. 3 own ground stations for 24h coverage have 15m collectors gaining 73dBi. After 3dB propagation losses, they receive -21dBm - the receiver can consist of a preamp, an oscillator and two diodes towards the complex baseband. If 100MHz are available and the noise temperature is 30K=-184dBm/Hz, the signal-to-noise is +83dB. 6dB margin permit 150M symbols/s of 23 bits each, for decoded 300MB/s. Phase noise isn't trivial, and the encoding shall suppress low frequencies. Can the reprogrammed KH-11 transmit so quickly? We shall modulate the polarization too, for a 3D-constellation. This was demonstrated over fibres, maybe it's already done with radio. Throughput *1.5 or 450MB/s. Still less than the sensors. 1000MHz available bandwidth would have multiplied the throughput by 8.7. Simultaneous beacons to many smaller separated ground stations would have improved too. Transmitting 450MB/s half of the time needs 4 years to map all the Moon with 75mm resolution uncompressed. Storage on Earth takes a few thousand disks. Marc Schaefer, aka Enthalpy

Half a watt is enough, observed in AM with a dumb receiver and one non-directional antenna. With proper modulation techniques, few mW suffice. That's why companies propose and experiment presently bidirectional Internet access via satellites. Radiocomm is a field for experts. Communication with satellites is much easier than on the ground because the ground attenuates a lot.

At 10GHz and 3m it can't be near field. This implies that some adverse effects will be felt at a longer distance. Many possibilities exist, depending on whether you seek interference, destruction of weak equipment, destruction of purposely shielded equipment. 30MHz-10GHz in one equipment is a bit ambitious, I'd split it over 2 or 3 generators+antennas. This is standard warfare technology for decades, well-known but not very public. Armed forces are equipped and use it, especially to destroy radars at the beginning of a war to get the control of the sky. Range exceeds 10m by much. Consider 100m for a luggage-sized weapon, much more for a weapon installed in an aeroplane or in a bomb.

Fuel Propellant depots are a well-studied option, with some (one?) strong advocates at Nasa. They permit to split the launch mass, which can be useful by itself: If the existing launchers are too small, it may sometimes avoid to develop a new one; But if a new launcher is necessary anyway, a bigger one isn't much more difficult, and developing costs more than recurrent manufacturing. In my scenario for a manned Mars mission, I split a trip over three SLS-class launches, so a Falcon wouldn't have coped with the needs. http://www.scienceforums.net/topic/83289-manned-mars-mission/ They offer advantages beyond splitting the mass: If a manned mission must travel faster than the economical (say Hohmann) path, then presetting heavy hardware (like fuel) by a slow and economical means saves much launch mass. Newer propulsion, like ionic or my sunheat engine http://www.scienceforums.net/topic/76627-solar-thermal-rocket/ save much launch mass over chemical propulsion but their faint thrust takes more time to reach a location, say a high orbit. They can serve to preset fuel for a crew, say to land on the Moon and come back. If you produce propellant in-situ, say oxygen by electrolysis of Mars' atmosphere or Moon's regolith or an asteroid's ice, you have to store it. Storing liquid oxygen in Lunar orbit or at a Lagrange point would reduce the launch mass (but cost??) because it's easier to put there starting from the Moon, and it can serve to send a vessel or craft that dives to and accelerates near Earth to take advantage of the Oberth effect https://en.wikipedia.org/wiki/Oberth_effect you might consider doing the same at Mars' moons. Whether you store only the fuel or as well the tank, the engines, the vessel or craft - that depends on mission needs and choices.Moving propellants between tanks hasn't been done in space up to now, but I suppose we need only to decide it. I've no firm opinion about swapping the tank instead of moving the liquid. ---------- Nitrogen tetroxide and hydrazine and its derivates are stored in space since ever, despite hydrazine freezes so easily. Solar-powered resistors keep it warm. Though, we want to get rid of these toxic and not so efficient propellants. We can insulate a tank well http://www.scienceforums.net/topic/60359-extruded-rocket-structure/page-2#entry761740(and nearby messages) and this suffices to store hydrogen for a week or if needed a month, enough to reach the geosynchronous orbit directly or for a short manned Moon mission. Easy, light, cheap. We can design cold tanks by making them white or putting sunshades. This suffices at Sun-Earth distance to store passively oxygen, methane, cyclopropane and spiropentane, farnesane and most propellants. It doesn't suffice for hydrogen, but hydrogen is very much necessary, say to put craft in orbit around Uranus and Neptune, and to fuel my sunheat engine. To my eye, active cooling is a better choice. It works for hydrogen too, near Venus and Mercury as well, and more easily on low orbit around the lukewarm Earth. A big tank needs only 100W cold. The cooler can consume very little of the propellants or run with sunlight. Again, it needs only to be decided http://saposjoint.net/Forum/viewtopic.php?f=66&t=2051 (mess there, some relates to a generator, some (May 17, 2010) with a cryocooler using the same hardware) with redundancy I'd rely on active cooling. Oxygen is the first and necessary step to avoid toxic propellants and improve efficiency, so storage must focus on cold propellants. Hypergolic propellants aren't necessary at attitude thrusters, say for a lander, as I describe an igniter there http://forum.nasaspaceflight.com/index.php?topic=27308.0on 01/13/2012 I consider long-term storage as a fundamental enabling technology for space exploration, both of oxygen and hydrogen (hence active), and it's easy.

The Ehang 184, one more electric helicopter design, with secondary batteries instead of fuel cells: http://www.ehang.com/ehang184("More about product" there) http://www.wired.co.uk/news/archive/2016-01/07/ehang-184-personal-drone-car("View gallery") Opinions should expectedly vary over its general design choices. My personal dislikes are: - It knows only a fully automatic mode. As an old engineer, I claim that reliability results from humans in the loop. - It has four rotor sites instead of six and apparently no parachute. But its numbers for mass, power, energy, flight duration add up. Maybe this one breaks through, and if not, an other will

The varied flavours of Gamestation can also benefit from spreading the Cpu among the Dram, stacking of Cpu and Dram chips, and networks I've described. I take as an example the Playstation 4 https://en.wikipedia.org/wiki/PlayStation_4 https://en.wikipedia.org/wiki/PlayStation_4_technical_specifications with the same 8GB of Dram and 1152 single float Cpu, on 28nm Amd process like the parent Radeon HD 7870 Pitcairn XT. Each of the 16 compute packages contains now a compute chip with 72 Cpu, is 18mm2 big because of individual sequencers, and dissipates 7W at 800MHz. The Jaguar Cpu drops away. The package contains also, stacked, a 72*8MB Dram chip. One Cpu accesses 3 Dram banks in a cycle to exchange three 32b words. With addresses, 360 contacts in 0.25mm2 by my adaptive connection leave 26µm*26µm pitch. The compute chip has an internal switch matrix among the 72 Cpu and to the 32 in and 32 out external shared links accessed by through-silicon vias. This accepts 170-ball compute packages and switch packages. 16 packages of 32x32 switch matrices operate in parallel, with 2*2 links at 2GB/s to each compute package. The printed circuit can have 7+6 layers and be 120mm*120mm small, plus some power converters. Each compute package can have 8GB of Mlc Flash at its side to load the games 16x faster than an Ssd would. The no-latency Dram provides 3 words per cycle to each Cpu and L1, or 11TB/s. The network transfers 500GB/s between any machine halves, three times as much as the PS4's Gddr5; the nodes can receive simultaneously a word each in 4 cycles. The design is easy to cool and scale up. This design can spread among gamestations earlier than among graphic cards because fewer companies must adapt. Marc Schaefer, aka Enthalpy

If you didn't understand the example it's because I botched the parallel version. Better: BeginMacro 0 R0 = R0 + R1 R1 = R0 Blt R4++, Maxloop, Macro 0 Endmacro Macro 0 It shows as well that the macro-op can't be executed sequentially during its definition. Making a block of the definition is then better. Compilers are already smart enough to rearrange the registers as needed. ----- A processor with explicit macro-op and several execution units can do as much as a vector processor. It doesn't need extra vector instructions, and combines different instructions and registers more flexibly in one cycle. The sequencer is much simpler because the compiler does the job.

Microprocessors, and previously mainframes, execute several instructions per cycle. The sequencer may decide it during the first pass in a loop or on the fly, the instruction cache may store the original instructions or the resulting "macro-op": many variants exist. It gains speed but complicates the sequencer. I suggest instead that the compiler does this job, using special explicit macro-op instructions that are defined dynamically at execution and always executed in one cycle. Example: ClrMacro 0 DefMacro 0 R0 = R1 + R2 DefMacro 0 R2 = R1 DefMacro 0 R1 = R0 DefMacro 0 Blt R4++, Maxloop, Macro 0 Macro 0 would define Macro 0 then execute it as: #pragma HeyCompilerMacroThat do { R0 = R1 + R2; R2 = R1; R1 = R0; } while (r4++ < Maxloop); The sequencer can be dumb. It can just stall the execution if data isn't available yet. The machine still needs conditional instructions, maybe register renaming. The defined macros must be saved at context switching, so there are very few macros. A loop can contain several macros, and a macro can contain several passes of the source code loop. Each execution and branch unit can memorize locally the part of a macro definition it contributes and recognize its call. Branching to a macro, not only to addresses, helps the hardware. A complicated sequencer must run faster when some instruction unexpectedly lags, say a memory access, but this is unfrequent at the suggested database - web server - AI engine where the tiny unit Dram responds in very few cycles. Advantageously, several loop passes are easier to run in a cycle when the compiler decides it. I suppose it was done long ago, at a mainframe, a Dsp... Here at a database - web server - AI engine, whose execution units are tiny (multicycle floats and mult), it reduces the consumption and size of the sequencer and keeps it efficient. Marc Schaefer, aka Enthalpy

30 years ago some research papers claimed to have made optical "gates" but these were only light subtracters, a dark fringe if you wish. They were linear, which doesn't make a logic gate. I don't know whether things have progressed meanwhile, but one doesn't need a hot topic to do research. Is optical computing still interesting at all, since electric components have improved so much? We can have a metal line every 30nm, but optical lines that close would interfere horribly, or rather, they wouldn't even carry light for being too narrow. Making an optical circuit 302 times less dense would need serious reasons. The power consumption looks bad too. Presently one electron has 0.6eV energy, one photon rather 1eV - but once you have charged 500 electrons at a node the bit remains without consuming power except for leaks. In contrast, light may need 100 photons right from the beginning of the transition, and more and more photons all the time. That's why I suggested instead - as everyone does - to concentrate on optical transmissions rather than optical computing. Beware, though, that electric transmissions are good: for instance 7Gb/s over 500 closely spaced pins between a graphics processor and its Ram. That is the kind of figures that need a strong improvement, so showing a fibre with a 10mm connector that transmits 40Gb/s won't bring anything. It's the reason why I propose printed circuits for the described supercomputers. Including all air, the crossboards transport a 20Gb/s signal in 0.75mm2. To be serious competitors, fibres must be very closely packed including all connectors, transmitters and receivers, or carry each >100 quick links but have tiny filters. To my knowledge, this isn't solved, and is a current research topic - for telecomms too. Computers need <100m range hence would accept more diverse wavelengths, presently they even use multimode fibres outside the transparency window; maybe it helps.

Happy new year! ¡Feliz año nuevo! Feliz ano novo! Bonne année ! Frohes neues Jahr! Felice anno novo!

Compute chips are even smaller than my last figure. The 14nm Knights Landing has no on-chip L3 and carries 1152 mul-acc on 35mm*25mm estimated from pictures, or 0.76mm2 per scalar mul-acc. If the common complicated vector sequencer and the L2 pay for individual simple sequencers, then an 8 scalar nodes chip takes 6mm2.

---------- Bytes-to-Flops update Up to now I estimated it from PC figures, but supercomputers have a smaller Dram for a given computing capability. Ratio Bytes Flops Machine ----------------------------------------- 0.026 1.4P 55P Tianhe-2 0.026 0.7P 27P Titan 0.094 1.5P 16P Sequoia 0.077 0.8P 10P Mitra ----------------------------------------- 0.032 512M 16G 8 nodes 1GHz ----------------------------------------- 0.14 16G 112G 4-core PC ----------------------------------------- Ratio Bytes Flops Machine The top-4 machines could have addressed more Ram, so their capacity is a designer's choice. >512MB Dram chips hence permit 8 nodes each, needing 4x fewer chips than previously estimated. >64MB per node made Windows Nt4 comfortable. A (hyper-) crossbar eases the packages over a cube or torus. ---------- Dram update Ddr4 chips offer 1GB=8Gb capacity in 2015, which didn't stall as I alleged on 13 September 2015. Organizing the Dram for speed reduces the density a bit, so I take 512MB+ per chip in this thread since 13 September 2015. A picture of a 1GB chip is there; in the 10mm*6mm, I believe to see 32 subgroups of 32MB, possibly 8192b*32768b. http://www2.techinsights.com/l/8892/2015-03-27/kbhjk Take this with mistrust, since capacities were 1000x smaller when I was in the job: Bit and word lines, of 25nm*75nm tungsten, explain a 40ns access time very easily by their resistance and capacitance over 2mm. This access time scales as a length squared or a subgroup capacity. 17 banks (easy Euclidean division) at each of the 8 nodes are 8.5x smaller than the 32 subgroups of the pictured 1GB Dram hence respond in 5 cycles at 1GHz. A L1 suffices. Accessing 16 of them delivers 3+ words per cycle, enough for a number cruncher. A database-Lisp-Prolog-Inference engine would have 6 to 12x more nodes (slower mult and float save power as on 25 October 2015) per Dram. Then, 17 banks respond in 1 cycle and deliver 16 words per cycle, enough for three Alu. ---------- Cpu update 14nm Finfet consume far less than my estimate on 31 January 2015. The Knights Landing https://software.intel.com/en-us/articles/what-disclosures-has-intel-made-about-knights-landing has 144 vector processors at 1.3GHz that mul-acc 512b = 8 doublefloats per cycle, or 1152 mul-acc on 64b per cycle. It's said to consume 160-215W, or 0.19W per 64b mul-acc at 1.3GHz https://en.wikipedia.org/wiki/Xeon_Phi which extrapolates as F2 to 0.11W per 64b mul-acc at 1GHz. Since power saving is vital to supercomputers, but Finfet and Dram processes are highly specialized, stacked chips made by different processes are better, as I describe on Nov 12, 2015. A number cruncher chip whose 8 nodes exchange each three 64b words with the Dram at 1GHz pace needs 1536 signals, and a database engine more: the small adaptive connection I describe on Nov 12, 2015 achieves it, be it by contact (sketch, electrochemical deposition can make small bumps) or capacitive. By the way, capacitive coupling is feasible (with effort) between separately packaged chips pressed against an other, permitting the Dram to evolve. The compute chip is small, estimated under the 10mm*6mm of a 1GB Dram. Comparing with 32mm*21mm for Intel's Xeon E5-2699v3: http://ark.intel.com/products/81061/Intel-Xeon-Processor-E5-2699-v3-45M-Cache-2_30-GHz?q=E5-2699%20v3 https://en.wikipedia.org/wiki/List_of_Intel_Xeon_microprocessors#Xeon_E5-16xx_v4_.28uniprocessor.29 http://www.chiploco.com/intel-haswell-ep-xeon-e5-2600-36020/ (die picture) The L3 vanishes. 2.2G transistors over 5.6G. The L2 vanishes too. Let's say, area *0.60. Fewer registers are necessary because the Dram provides the throughput to feed the execution unit. Faster context switching. The E5-2699v3 has 18 Avx2 cores that mul-acc 4 double floats per cycle. That's 9x more than 8 scalar cores. A 14nm process could slash the area by 2 over the 22nm Haswell - maybe. The scalar sequencer is simpler but each mul-acc has one. Neutral? Is hyperthreading needed? It obtains 30% more from the same execution units but doubles the registers, caches, and more than doubles the sequencer. 30% more cores is simpler, maybe it saves power and money. The 30x smaller compute chip takes then 22mm2. A database-Lisp-Prolog-Inference chip has more sequencers and caches hence is bigger. The tiny dissipation would enable more nodes per package, which makes sense only if abandoning some network throughput. Two Dram and compute chips fit without through-silicon vias (Tsv), or a bigger database chip with Tsv can carry several Dram. More companies offer 14nm or 16nm finfets besides Intel: http://www.globalfoundries.com/technology-solutions/leading-edge-technology/14-lpe-lpp http://www.tsmc.com/english/dedicatedFoundry/technology/16nm.htm http://www.samsung.com/global/business/semiconductor/file/media/Samsung_Foundry_14nm_FinFET-0.pdf Samsung also manufactures Dram, Flash and offers the Arm Cortex A7. ---------- Supercomputer update The exaflops machine draws 55MW "only", not 200MW - excellent news. If one accepts 1/4 the previous network throughput, then the whole machine is 4x smaller. That's 16 cycles per double float. Depending on how much remote data an algorithm needs: If it makes 2*N2*Log2(N) mul-acc (2*32b complex) from N2 data points, 2562=N2=0.5MB chunks keep the execution unit busy. But N3 mul-acc from N2 data points need just 162=N2=2kB chunks. Marc Schaefer, aka Enthalpy

These elements of the descent-ascent module shield the astronauts: In Martian orbit and during the descent, the ascent propellants bring 190kg/m2 of O and N and the tanks 30kg/m2 of Al-Zn. Just after landing, the atmosphere offers 162kg/m2 of O and C at the Zenith. The astronauts sleep and shelter in the ascent module until they've built the sand walls. The room below the descent module shields decently too and is quickly accessible but unpressurized.Propellants don't need to shield the descent module, only the ascent one. Once built, the sand wall brings 1100kg/m2 of O, Si, metals. The astronauts move in the descent module. Soon after take-off, no propellant shields the astronauts until they join the return vessel.They accumulate dose and need luck for few hours. Apollo crews had luck for a week. Extra shielding mass is best put at the bucket-seats, followed by the control panel above the astronauts. ---------- These skins and tanks are imperfect but let estimate the masses and engine pressures. The frustrum's (and ceiling's) skin contains the ascent's Mon-33 in AA7022 tubes, spin-formed then milled into cones with integral stiffeners, for 1mm wall thickness and 2mm equivalent mass. Two layers of tubes are brazed together, transverse stiffeners welded on them, and sheets welded on the stiffeners, milled for 1mm wall thickness and 2mm equivalent mass. The tubes break at 68bar, the 36m2 frustrum and ceiling weigh 28kg/m2 or 1,0t. Extrusions welded together compose the floor that holds the ascent's farnesane. 1mm thin AA6005A weighs 24kg/m2 or 0.6t and breaks at 67bar, allowing 29bar in the chamber. Extrusions welded together compose the floor that holds the heavy descent's Mon-33. 1.5mm thin AA6082 weighs 2.6t and breaks at 139bar, allowing 61bar in the chamber. A separate torus, Di=3.0m De=4.1m, holds the descent's farnesane - lighter combination than previously sketched. 2mm maraging weigh 0.3t and break at 146bar. Extrusions welded together make the descent module's frustrum. They widen in steps to fit the cone, while milled parts make the tapers. 2mm thick AA7022A weighs 23kg/m2 or 2.0t and resists the sand's 100t easily. Channels at <1bar separate the tanks from the air. Pumps reinject small leaks to the proper location, bigger leaks are vented. Maybe leaked Mon can be decomposed catalytically, but this is hot. Some graphite tank holds the helium. ---------- There are many variants for the tanks and skins. Higher chamber pressures don't pay for the heavier tanks. Steel and titanium would need sheets too thin. I don't trust graphite fibres to retain the habitat's air. But aluminium sheets can be milled and diffusion-bonded to make a pseudo honeycomb sandwich that lets the propellants flow (sketch). Easier than extrusions at a frustrum, and not limited to Mars. Mind the materials compatibility. Sheets can be milled in a quilted shape and brazed or welded on a space truss assembled from tubes to make a sandwich (sketch). Easier than extrusions at a frustrum, and not limited to Mars. The ascent module can be smaller, with lighter walls and thicker propellant. It can be cylindrical (consider a fairing) for ease. Petals (or a ballute?) can stabilize the descent module and brake it instead of parachutes (sketch): http://www.scienceforums.net/topic/77901-controlled-aerobraking/ rockets then brake from 400m/s without parachutes instead of 100m/s. The descent module can then be a cylinder with several floors. ---------- The descent engine pushes 1050kN in a D=1.25m nozzle. Expanding from 61bar to 73kPa, it achieves Isp=2760m/s=282s. Braking 60t by 216m/s from the 400km Martian orbit leaves 55.5t, the 3.8t shield+parachutes 51.7t, 486m/s to brake and hop two times 43.3t landed. This consumes 12.9t propellants. The ascent engine pushes 180kN in a D=1.1m nozzle. Expanding from 28bar to 12kPa, it achieves Isp=2962m/s=302s. To bring 3.6t in the 400km Martian orbit, 4100m/s for ascent and manoeuvre need 14.4t at lift-off. This consumes 10.8t propellants. Some 10t structure leave nearly 20t for the crew, their life support, tools, experiments. I describe igniters for attitude control thrusters there: http://forum.nasaspaceflight.com/index.php?topic=27308.msg849905#msg849905 but here's the sketch ---------- Other propulsion options include: Aza variants of the alkane. Easier to produce, 3s more efficient BUT hypergolic with Mon-33. Well, I'm not even sure if Mon-33 lights an alkane too. Oxygen would fit in a wide torus that can be vacuum-insulated. It has other drawbacks. Then, fuels cryogenic too would gain only 1t freight over farnesane. Batteries and electric pumps gain too little pressure to pay for their mass. But my Mon-33 recomposition cycle would achieve Isp=318s and 341s, gaining 3.5t freight http://www.scienceforums.net/topic/83156-exotic-pumping-cycles-for-rocket-engines/?p=805383 is it responsive enough to land and hop? Marc Schaefer, aka Enthalpy

Thanks for your interest! The transfers to and from Mars are to take 80 days in my scenario, which relies on the sunheat engine and on aerobraking, because of radiation concern. That's a short period for zero gravity: "long term" means rather a year now, with the record around 600 days.

Here's how the descent-ascent module can look like. Its oxidizer is Mon-33, or 33% NO in N2O4, because oxygen tanks fit badly in the module. 33% NO prevent freezing at -100°C but the toxic oxidizer has 8 bar vapour pressure at 298K. The fuel is farnesane or phytane, freezing at -100°C and safe; suggestion for synthesis there http://www.chemicalforums.com/index.php?topic=56069.msg297847#msg297847 The module's skin needs some strength and thickness anyway, so it holds both propellants too, which help (141kg/m2) shield against radiations, especially after landing. The skins can consist of aluminium extrusions like described there http://www.scienceforums.net/topic/60359-extruded-rocket-structure/ or of diffusion-bonded parts, or similar. As the vessel's skin holds pressure easily, the engines are pressure-fed. Electric pumps and batteries would be possible http://www.scienceforums.net/topic/73571-rocket-engine-with-electric-pumps/ or the original cycle that decomposes the Mon-33 http://www.scienceforums.net/topic/83156-exotic-pumping-cycles-for-rocket-engines/?p=805383 The astronauts bring bags, things like a dust sucker or motor brushes to fill the bags with Martian sand, wheelbarrows or some conveyor for local transport, and a crane to lay the bags on the module. 0.8m thickness adds 1100kg/m2 shielding to Mars' atmosphere (162kg/m2 at the Zenith). Compare with Earth's atmosphere: 2700kg/m2 at airliners' 10km altitude. I prefer this over a home printed on Mars in advance because the astronauts need no accurate landing. Marc Schaefer, aka Enthalpy

Switch cabinets for the exaflops computer contain 60 stacked switch boards like the compute cabinets do. From 32 compute boards in the same line or row of compute cabinets, sets of 500 nodes converge their links over 32 cables to one switch board with 32 connectors. A Google server or Tianhe-2 equivalent would instead have one big dimension of 48+ compute cabinets, and the switch board 48+ connectors. Numbers here refer to the exaflops computer. The switch boards shuffle subsets of 32 link pairs from nodes at identical positions in the 32 compute cabinets to send them to one of the five-hundred 32x32 switch matrices - or rather, it sends 96 link pairs to one of hundred and sixty-eight 96x96 switch matrices as these exist aleady. The main board carries the 16,000 link pairs to switch vessels. At 100mm pitch, two connectors or 2000 links take >10 signal and 10 supply layers. In the other direction, 84 switch vessels carry each 192 link pairs to two switch components; the vessels are 13mm tall and comprise >10+10 layers. The line amplifiers on the main board and switch vessels are not sketched. Switch boards for the Tianhe-2 equivalent are 1.2m*1.2m instead of 0.8m. Two stacked or opposed switch vessels could lay flat on the main board. Boards with 30+ layers would avoid the switch vessels. More cables with fewer links each would ease the switch boards. Maybe the links of a few compute boards can be partly shuffled elsewhere before landing on smaller switch boards. Marc Schaefer, aka Enthalpy

Yes, only a part of the kinetic energy is radial versus Earth or an observer, so a v2/c2 shift will be observed. Thanks! And if the observer were below or above a clock falling vertically, would he see a v2/c2 shift?

Here's a cut through a computing or switch board, cables and connectors. Not to scale, nor actual numbers of components and flex circuits. Coupling is capacitive as suggested on 22 February 2015. Several smaller chips ease the sockets for 500 links in and 500 out per 40mm*40mm connector, but redundancy eases with chips not too small. A connector can consist of several sockets too; if flatness were a worry, a flat tool can hold the sockets during reflow. At the connectors, the cables can have a piece of stiff printed circuit for flatness. The flexible printed circuits are loose in a protective sleeve, except at the connectors where the soldered contacts hold the flex together, helped by glue put by vacuum impregnation or in advance on the flex and reflown. Some shape at the board and connectors hold these at accurate position. A small clamping force keeps the connectors in place. The connectors can be completely smooth and have a protective insulating layer. Marc Schaefer, aka Enthalpy