-
#1040
by
Confusador
on 17 Mar, 2017 13:58
-
Indirectly, Moore's law helps everything.
The fact that a noob company can work on a state of the art SC engine and get it to work so quickly is very much a consequence of ML.
The fact that you can finance a huge project using the constellation is a result of ML since the demand originates with smart phones and self driving cars.
As folks said, the computational power needed for fly-back and landing is not found in older rockets.
And of course, ISP proper has got nothing to do with it.
SpaceX is most certainly NOT a noob company at this point. And they've been working on Raptor for years, building on their knowledge of building Merlins for a decade.
It's also not machine learning (which is a specific thing by the way, and it requires a large training set to work, thus not applicable here) that they used for Raptor but an interesting algorithm for studying fractals complex fluid flows efficiently.
They didn't say anything about machine learning. From context, ML= Moore's Law
-
#1041
by
Comga
on 17 Mar, 2017 14:12
-
It is a repeat to say my guess would be a descent under parachutes to a dessert landing, softened by the SuperDracos.
I usually like to end things with a nice dessert.
What can I say?
It was a serious question now derailed by a spelling error.
At least you didn't make more jokes about soft landing in pudding.....
-
#1042
by
meekGee
on 17 Mar, 2017 14:22
-
Thx - I didn't realize there'd be an ambiguity. 2MTLAs, clearly.
I was never a fan of machine learning anyway. Or rather, I think too many things are now called that. Very OT. (!Over Temperature)
-
#1043
by
oldAtlas_Eguy
on 17 Mar, 2017 15:03
-
Ok Moore's Law is nothing more than a pointing out that IC capability incresses exponentially over a set amount of time interval (power of 2 for every 2 years).
For the LV/Space (satellites) industry the time span is different. But it does follow the same concept being a high tech advancement controlled capability industry.
For the LV industry SpaceX has set the the time span at 6 years for the doubling of payload at same cost. 2010 F9v1.0 (max ~10mt to LEO) [$6,000/kg] to the 2016 F9v1.2 (max ~20mt to LEO)[introduction to flight of the M1DFT+ [$3,000/kg].
So at this rate in 2022 the capability/cost would be 40mt for $60M cost. This would sort of coincide with higher rate operation of FH or even the regular use of New Glenn in competition with FH that lowers the price of FH flights such that FH is heavily reusing 3 cores (5 to 10 times rapid reuse with minimal refurbishment).
$1,500/kg
But do not expect a doubling of capability to cost at every 2 years.
This industry is just not that nimble.
-
#1044
by
meekGee
on 17 Mar, 2017 15:14
-
Ok Moore's Law is nothing more than a pointing out that IC capability incresses exponentially over a set amount of time interval (power of 2 for every 2 years).
For the LV/Space (satellites) industry the time span is different. But it does follow the same concept being a high tech advancement controlled capability industry.
For the LV industry SpaceX has set the the time span at 6 years for the doubling of payload at same cost. 2010 F9v1.0 (max ~10mt to LEO) [$6,000/kg] to the 2016 F9v1.2 (max ~20mt to LEO)[introduction to flight of the M1DFT+ [$3,000/kg].
So at this rate in 2022 the capability/cost would be 40mt for $60M cost. This would sort of coincide with higher rate operation of FH or even the regular use of New Glenn in competition with FH that lowers the price of FH flights such that FH is heavily reusing 3 cores (5 to 10 times rapid reuse with minimal refurbishment).
$1,500/kg
But do not expect a doubling of capability to cost at every 2 years.
This industry is just not that nimble.
You're describing a hypothetical "Musk's law"...
The ML I was referring to was the original semiconductor one. (Gates/mm2)
I don't know how $/kg will behave in the future.
$/kg for air craft certainly isn't dropping exponentially at this point.
I'd say that the cost of flying a rocket might, at best, approach that of flying an airplane of the same fuel loas. In the far future.
-
#1045
by
IainMcClatchie
on 17 Mar, 2017 15:57
-
Semiconductor memories, primarily flash, have been making good progress in the last decade as they are starting to exploit the third dimension on chip. But Moore's Law, as applied to logic, ended 5 years ago, because the 20nm and then 14nm process nodes are not cheaper per transistor than the 28nm node.
Here is a nice article on Nvidia's view of the problem. Note that costs are coming down over time, but that's just the capital cost of the fabs getting amortized. That'll bottom out soon.
A separate but also serious issue is that wires haven't gotten faster for a long time. Transistors continue to get faster (e.g. the 600 GHz Ft I noted earlier), but at this point wires dominate. That's why CPU frequencies haven't budged in 8 years. There has been progress in better system integration (on CPU hardware video decoding, or flash memory as the backing store rather than rotating disk drives), and a little progress in work-per-cycle which comes from better use of more on-chip memory (better branch prediction).
-
#1046
by
ChrisWilson68
on 17 Mar, 2017 16:33
-
If someone were interested in discussion of the SpaceX Crewed Circumlunar Mission in 2018, where should they go? Clearly not this thread.
-
#1047
by
kevinof
on 17 Mar, 2017 17:09
-
If someone were interested in discussion of the SpaceX Crewed Circumlunar Mission in 2018, where should they go? Clearly not this thread.
My thoughts exactly.
Sent from my SM-J510FN using Tapatalk
-
#1048
by
dglow
on 17 Mar, 2017 21:45
-
If someone were interested in discussion of the SpaceX Crewed Circumlunar Mission in 2018, where should they go? Clearly not this thread.
I believe it was Mr. Bradbury who, lampooning The Bard, told us "digression is the soul of wit."
-
#1049
by
MATTBLAK
on 17 Mar, 2017 22:01
-
I've been a bit unwell the last couple of days - which is why I haven't finished my long-short/short-long story "Flight Of The Dragon". In it, I speculate what the mission is going to be like in the major - and some minor - details. It'll be finished in a day or two; then I'll spend half a day editing and combing it for typos and polishing it. Hopefully, this will spark some discussion about the 'nitty-gritty' of the interesting details.
EDIT: I'm up to about 10 thousand words and climbing. I'm going to try and cut to the chase of the story and not make it a novella...
-
#1050
by
TomH
on 18 Mar, 2017 00:03
-
I am the one who brought up Moore's Law. The context was in relation to the Human Genome Project. My point was that the HGP was such a success solely due to Moore's Law, and I meant that in a completely literal, denotational manner. The HGP was expected to take 10-20 years to decode a single genome. The project was completed far ahead of schedule and far under budget simply because genome decoding is done BY COMPUTER and the immense processing speed increases after the project began led to its success. Today, you can pay a modest fee and have your own personal genome decoded in a relatively short time.
These other figurative allusions to Moore's law were not what I was implying nor what I intended anyone to infer. I was trying to point out that the HGP is not highly instructive in relation to advancement in space technology. Faster processing contributes to computer modeling and to the ability of control in landing a booster. Yet in the total scheme of things, I was implying that the HGP mention did not belong here. I think we need to drop this and get back on topic before Chris deletes the last several pages.
-
#1051
by
su27k
on 18 Mar, 2017 01:58
-
Ok, back on topic: Would a flight on New Shepard be useful in terms of training? Seems to me this would be the closest analog to the ascend and landing phase of a Dragon mission, plus you got some weightlessness in a capsule as bonus.
-
#1052
by
IntoTheVoid
on 18 Mar, 2017 15:10
-
Ok, back on topic: Would a flight on New Shepard be useful in terms of training? Seems to me this would be the closest analog to the ascend and landing phase of a Dragon mission, plus you got some weightlessness in a capsule as bonus.
What would the benefit be in training for an automated flight, on another automated flight, on a completely different rocket with 1/10th the flight history?
-
#1053
by
RonM
on 18 Mar, 2017 17:40
-
Ok, back on topic: Would a flight on New Shepard be useful in terms of training? Seems to me this would be the closest analog to the ascend and landing phase of a Dragon mission, plus you got some weightlessness in a capsule as bonus.
What would the benefit be in training for an automated flight, on another automated flight, on a completely different rocket with 1/10th the flight history?
Maybe more psychological than anything else. After the suborbital flight they can be asked "Okay, was that great or are you having second thoughts about this?"
-
#1054
by
Coastal Ron
on 18 Mar, 2017 17:54
-
Ok, back on topic: Would a flight on New Shepard be useful in terms of training? Seems to me this would be the closest analog to the ascend and landing phase of a Dragon mission, plus you got some weightlessness in a capsule as bonus.
What would the benefit be in training for an automated flight, on another automated flight, on a completely different rocket with 1/10th the flight history?
Considering the relatively low cost compared to their lunar trip, flying on Virgin Galactic to get the weightless jitters out of the way would seem like a good idea. And who knows, maybe these two individuals were one of the original ticket holders for Virgin Galactic sub-orbital flights?
-
#1055
by
AncientU
on 18 Mar, 2017 18:10
-
Ok, back on topic: Would a flight on New Shepard be useful in terms of training? Seems to me this would be the closest analog to the ascend and landing phase of a Dragon mission, plus you got some weightlessness in a capsule as bonus.
Astros prepare for flight, in part, by flying on the 'Vomit Comet' to get an introduction to weightlessness. New Shepard could indeed be a reasonable training step. Kinda doubt that SpaceX would spring for the flight(s) and the timing of commercial flights/existing queue of paying customers on NS may be problematic.
-
#1056
by
ChrisWilson68
on 18 Mar, 2017 18:16
-
What would the benefit be in training for an automated flight, on another automated flight, on a completely different rocket with 1/10th the flight history?
How does New Shepard have "1/10th the flight history" when Falcon Heavy has a flight history of exactly zero?
Because Falcon Heavy has a huge amount of commonality with Falcon 9. Obviously, there could be failure modes that are specific to Falcon Heavy, but the vast majority of potential failure modes for Falcon Heavy would also be failure modes for Falcon 9, and much of the risk of those has been retired by Falcon 9 flights.
-
#1057
by
Space Ghost 1962
on 19 Mar, 2017 20:39
-
There has been no crewed, American HSF vehicles flown since Shuttle. No relevant flight history for any of these vehicles. They all start off equal.
There have been qualifications passed to fly for Orion, Dragon 2, Starliner, and Dream Chaser. Neither VG nor BO have similar qualifications passed.
NS has not been flying at the rate BO said it would this year. In fact, am beginning to doubt if either VG or BO represents "a going concern" in the business of space tourism.
It is more likely that BO will make revenue (of a sort) from engine sales, long before it makes revenue off of HSF. Perhaps both VG/BO will launch unmanned spacecraft before the 50 or so flights that NS was supposed to do before lofting people. Or perhaps they will do something like SS1, lofting a few to show that it can be done and then not doing it again.
Much more likely that if SX succeeds with a lunar free return mission, that similar (or variations) will follow, possibly as many as a hundred. As these are the only "commercial HSF" options at the moment, can't talk about much more than that.
Jon Goff's excellent suggestion of EML 1/2 cruise flyby, with extended consumables, certainly sounds like a "stretch" mission concept worth considering, and within the scope of projected vehicle operations, modulo radiation exposure accumulation issues. Extended stays might even be possible with a small hab, possibly using CRS resupply for perhaps monthly stays - all within the realm of existing deployable systems.
None of this complicated by other political, scientific, or engineering issues. Given Musk's penchant for popular culture, perhaps title it the "Restaurant at the End of the Universe" or some such. Like dining at the top of a tall building, only more so ... perfect for these decadent times.
-
#1058
by
AncientU
on 19 Mar, 2017 20:44
-
This would be the opportunity Bigelow has been (supposedly) awaiting.
-
#1059
by
oiorionsbelt
on 20 Mar, 2017 01:06
-
SpaceX says
This presents an opportunity for humans to return to deep space for the first time in 45 years and they will travel faster and further into the Solar System than any before them.
How far?
How fast?
