1

Topic: For 3 $ a supercomputer of 2004

https://medium.com/intuitionmachine/bui … dd60d4a93c 3 $ today and at you computing power of the fastest  the beginnings of the zero. Now on such  , and what did 10 back??

2

Re: For 3 $ a supercomputer of 2004

Hello, sharpcoder, you wrote: S> Now on such  , and what did 10 back?? All modeling of physical processes can gobble up any capacity.

3

Re: For 3 $ a supercomputer of 2004

Hello, sambl4, you wrote: S> Hello, sharpcoder, you wrote: S>> Now on such  , and what did 10 back?? S> All modeling of physical processes can gobble up any capacity. It is clear, especially if is written  by the programmer.

4

Re: For 3 $ a supercomputer of 2004

Hello, sharpcoder, you wrote: S>> All modeling of physical processes can gobble up any capacity. S> it is clear, especially if is written  by the programmer. The programmer certainly here plays a role, but the main thing - models, algorithms, required accuracy.

5

Re: For 3 $ a supercomputer of 2004

Hello, sambl4, you wrote: S> All modeling of physical processes can gobble up any capacity. Aha. Here the main thing not to think of efficiency, algorithms, and to consider more signs both to scale, scale. And not to recall that a nuclear bomb calculated practically on the calculator. In summary we receive weather forecasts worse national signs.

6

Re: For 3 $ a supercomputer of 2004

Hello, sharpcoder, you wrote: S> Now on such  , and what did 10 back?? Did the same.

7

Re: For 3 $ a supercomputer of 2004

Hello, Ops, you wrote: Ops> And not to recall that a nuclear bomb calculated practically on the calculator. Here by the way just here on operation recalled and told, how at Richard Fejnmana in the book it was told, how they on adding machines did calculations for this bomb

8

Re: For 3 $ a supercomputer of 2004

Hello, Ops, you wrote: S>> All modeling of physical processes can gobble up any capacity. Ops> aha. Here the main thing not to think of efficiency, algorithms, and to consider more signs both to scale, scale. And not to recall that a nuclear bomb calculated practically on the calculator. Ops> in summary we receive weather forecasts worse national signs. When for words it is not necessary to answer that it is possible and on the calculator. You will master on the calculator http://www.bbc.com/russian/science/2014 … ter_model?

9

Re: For 3 $ a supercomputer of 2004

Hello, sambl4, you wrote: S> Here by the way just here on operation recalled and told, how at Richard Fejnmana in the book it was told, how they on adding machines did calculations for this bomb set such task now, there would be no capacity, it should clusters , and all the same, would blow up 1 time from 10.

10

Re: For 3 $ a supercomputer of 2004

Hello, Ops, you wrote: Ops> set such task now, there would be no capacity, it should clusters , and all the same, would blow up 1 time from 10. The iron Sky, a film. To the gloomy German genius got iPod Touch

11

Re: For 3 $ a supercomputer of 2004

Hello, GarryIV, you wrote: GIV> When for words it is not necessary to answer that it is possible and on the calculator. And on what your way a nuclear bomb considered? On  clusters? GIV> you will master on the calculator http://www.bbc.com/russian/science/2014 … ter_model? A question - what for? The practical purpose what? To master it is more than capacities? To draw a beautiful picture for distributing grants? To train programmers in tensioning of the known phenomena on the code? Same, as a matter of fact, the presentation displaying already known.

12

Re: For 3 $ a supercomputer of 2004

Hello, Ops, you wrote: GIV>> When for words it is not necessary to answer that it is possible and on the calculator. Ops> and on what your way a nuclear bomb considered? On  clusters? And they did not appear about calculators on what was on that considered. GIV>> you will master on the calculator http://www.bbc.com/russian/science/2014 … ter_model? Ops> the Question - what for? The practical purpose what? To master it is more than capacities? To draw a beautiful picture for distributing grants? To train programmers in tensioning of the known phenomena on the code? Ops> same, as a matter of fact, the presentation displaying already known. There quite to itself the scientific purposes. But if to you more close the application-oriented purposes you can count chemistry on the calculator - new medicines, materials and so on.

13

Re: For 3 $ a supercomputer of 2004

Hello, Ops, you wrote: Ops> Aha. Here the main thing not to think of efficiency, algorithms, and to consider more signs both to scale, scale. And not to recall that a nuclear bomb calculated practically on the calculator. That bomb weighed five tons, and it needed 4 kg of weapon uranium. Ops> in summary we receive weather forecasts worse national signs. Generally normal now forecasts (about three days, certainly). And in those days forecasts in the modern understanding at all was not. Were not able to do them on calculators.

14

Re: For 3 $ a supercomputer of 2004

Hello, sharpcoder, you wrote: S> https://medium.com/intuitionmachine/bui … dd60d4a93c S> 3 $ today and at you computing power of the fastest  the beginnings of the zero. S> now on such  , and what did 10 back?? Same videocards, and on a supercomputer general purpose processors. It is impossible so to compare. If the task on the videocard does not lay down, there is no at you a supercomputer.

15

Re: For 3 $ a supercomputer of 2004

Hello, alpha21264, you wrote: A> Generally normal now forecasts (about three days, certainly). How there, at Strugatsky? Water in  is dark.

16

Re: For 3 $ a supercomputer of 2004

Hello, sharpcoder, you wrote: S> https://medium.com/intuitionmachine/bui … dd60d4a93c S> 3 $ today and at you computing power of the fastest  the beginnings of the zero. S> now on such  , and what did 10 back?? Authors of article similar do not see a difference between SIMD and MIMD. Capacity of videocards is not correct for comparing to capacity CPU-klastera

17

Re: For 3 $ a supercomputer of 2004

S> 3 $ today and at you computing power of the fastest  the beginnings of the zero. latest Xeon Phi (Knights Landing) is at 3.46 teraflops double precision here   under it: https://market.yandex.ru/catalog/55323/ … ;onstock=1 In the beginning zero I considered the task  on Pentium 75, 50 nodes in  (a mouse you move nodes, the program quickly rebuilds) turned out. Now it is possible more (but ).

18

Re: For 3 $ a supercomputer of 2004

Hello, vsb, you wrote: vsb> Hello, sharpcoder, you wrote: S>> https://medium.com/intuitionmachine/bui … dd60d4a93c S>> 3 $ today and at you computing power of the fastest  the beginnings of the zero. S>> now on such  , and what did 10 back?? vsb> Same videocards, and on a supercomputer general purpose processors. It is impossible so to compare. If the task on the videocard does not lay down, there is no at you a supercomputer. Too most and in , the task too should be decomposed on parallel calculations.

19

Re: For 3 $ a supercomputer of 2004

Hello, sharpcoder, you wrote: vsb>> Same videocards, and on a supercomputer general purpose processors. It is impossible so to compare. If the task on the videocard does not lay down, there is no at you a supercomputer. S> too most and in , the task too should be decomposed on parallel calculations. Not the same. In videocards one instruction becomes over a heap of the same data. And if at you different instructions in each flow the videocard does not help you. I.e. if you have to count one million hashes for numbers from 0 to one million, you get an array on one million numbers, load the program which considers a hash and apply it to this one million numbers simultaneously. Here the videocard will work perfectly well. And if you for each number have different instructions, if and , here already the videocard loses meaning, since From 10 000 working processors will work 1-2 on slow rate, while a supercomputer in which these 10 000 processors independent and independent from each other, will be processed by all much faster.

20

Re: For 3 $ a supercomputer of 2004

Hello, Ops, you wrote: Ops> In summary we receive weather forecasts worse national signs. It  for televiewers such that is easier, accuracy sentry. The normal forecast - probability, looks it so.

21

Re: For 3 $ a supercomputer of 2004

Hello, Ops, you wrote: Ops> Aha. Here the main thing not to think of efficiency, algorithms, and to consider more signs both to scale, scale. And not to recall that a nuclear bomb calculated practically on the calculator. It did not calculate especially. Roughly estimated the main conditions for maintenance of chain reaction and all. And then blew up and looked that turns out. Really atomic explosions started to consider only in 90. Well and one more example. Earlier a path of flight -7 some days considered Keldysh's whole institute. And now it is enumerated directly in flight by the onboard computer in real time.... <<RSDN@Home 1.0.0 alpha 5 rev. 0 on Windows 8 6.2.9200.0>>

22

Re: For 3 $ a supercomputer of 2004

Hello, alpha21264, you wrote: A> That bomb weighed five tons, and it needed 4 kg of weapon uranium. Also what? Uranium there was not 4, and 64. It reacted only  them less kilogram, but not because of  calculations, and because of limitation of the gun circuit. And a plutonium bomb  on the same adding machines.... <<RSDN@Home 1.0.0 alpha 5 rev. 0>>

23

Re: For 3 $ a supercomputer of 2004

Hello, AndrewVK, you wrote: AVK> It did not calculate especially. Roughly estimated the main conditions for maintenance of chain reaction and all. And then blew up and looked that turns out. Really atomic explosions started to consider only in 90. There those conditions still, simultaneously from different directions to press out without calculations hardly it turns out. AVK> well and one more example. Earlier a path of flight -7 some days considered Keldysh's whole institute. And now it is enumerated directly in flight by the onboard computer in real time. So that at fighters. And at modern  it on GSM, from the ready Chinese unit for , in a cluster did requests.

24

Re: For 3 $ a supercomputer of 2004

Hello, Ops, you wrote: AVK>> It did not calculate especially. Roughly estimated the main conditions for maintenance of chain reaction and all. And then blew up and looked that turns out. Really atomic explosions started to consider only in 90. Ops> There those conditions still, simultaneously from different directions to press out without calculations hardly it turns out.  were approximate. And then a heap of experiments with steel balls. Ops> so that at fighters. Not only. On civil rockets too most.... <<RSDN@Home 1.0.0 alpha 5 rev. 0 on Windows 8 6.2.9200.0>>

25

Re: For 3 $ a supercomputer of 2004

Hello, AndrewVK, you wrote: AVK> Rasschety were approximate. And then a heap of experiments with steel balls. So quits, all these computing powers are not too necessary for serious achievements? AVK> not only. On civil rockets too most. You about space? So he from fighters gemmated.