Topic: [ARCHIVE] Bitcoin challenge discusion - page 10. (Read 29562 times)

Table update with bitcrack key/rate per sec:


Added RTX 2080 and Tesla K80

https://mrxtraf.github.io/brainwalletXX/#game

how can i play the same? )

Thanks ,that was usefull in my ubuntu machine.
so now we have just to specify a range and search into it to optimize our search.
Hope can do it as soon as we can, even gpu membre are more powerfull XD



Thanks

on ubuntu first run these command with sudo
apt-get install libgmp-dev
apt-get install libmpfr-dev
apt-get install libmpc-dev

then run direct install pip install gmpy2
or
downloaded gmpy2 wheel files
you will success

thanks but its look like no hope ,because I use python 3.7 and x64 also, I changed to try python27 but still same error. thanks for your support.
and about puzzle, What if we minimize the range of search ?!! it's possible to find it fast than full range.


Thanks

You should use the same bit version (32 or 64) and the same Python version (cp37 in the file name means Python version 3.7). The error looks like you use the incorrect wheel version.

Yeah I tried that too but still nothing,


if i use ubuntu i get that need gmp .if i tried windows i get this problem . i readed also that gmpy add 8% fast calcul.

Thanks

It is not very difficult. Go here and download the correct version for your Python: https://www.lfd.uci.edu/~gohlke/pythonlibs/#gmpy
For eample, for python 3.7 (64bit) use this file: gmpy2-2.0.8-cp37-cp37m-win_amd64.whl
Copy that file to the root folder of your phyton, and run with administrator rights the following code: python -m pip install gmpy2-2.0.8-cp37-cp37m-win_amd64.whl

Thank you so much for explaining ,and sure he or she will not give the gpu code until he finish and find the key.
and about gmpy2 ,i tried to use it but can't, its problem with installing gmp and the wheel is damaged.
So now we can just go to #64 and let gpu users doing it 

I think this question should be adressed to that one who made the code kanagaroo  
But if you read some mesages before, he (or she, i can not be sure) found the last key #105, and he also wrote that he used GPU to find that one. He wrote that: "Solving #105 takes 1 month of working of GTX 1080ti at 270-275 Mh/s, and last 2 week of work of all my available GPUs (total h/s was approximately 1800Mh/s). So, it was not so simple as the Python code presented. I haven't reached published by j2002ba2 1600Mh/s. I has obtained only 470Mh/s with a cloud GPU server and one Tesla V100" (the whole post is here: https://bitcointalksearch.org/topic/m.52535752)


The 10 times difference could be because you use the phyton code to calcualte the scalar inversion. It is slow. Try to import gmpy2 library and use function gmpy2.invert - you will reach the same 100Kh/sec (depending on your CPU power). But ususaly the difference is exactly 10 times.

I do not argue that the keys from the puzzle were obtained randomly. What was this generator, bitwise, ekdska, super random, etc. irrelevant.
But here we go into the theory of probability. For example, if you get a random number from 1 to 100, then theoretically one kind of number, for example 3, can fall 10 times in a row. But it is theoretically. If in practice this happens, then it will be more like a generator failure. Another example. We create a million random numbers from 1 to 100. In theory, some numbers will turn out much more, and some much less. But in practice, the quantitative discrepancy will be no more than 10 percent on all numbers. That is, not entirely by accident.
From this we see that the numbers found have some rules or conditions. The number of set bits can be from 30 to 70 percent, a sequence of up to 8 identical bits in a row, the same rotation no more than 10 times, etc. The same rules can be chosen for the 8-decimal version and the 16-decimal representation. Repeating the same numbers, ladder, etc.
And applying these rules, we can discard most of the keys so as not to check them for correctness. Yes, there may be exceptions to the rules, but most should be subject to these rules.
This increases the speed of search. And you can start not with boundary values, but let's say averaged ones.

Sort, less than 50 and more than 50.

Count 35 Average 60,77%

Count 30 Average 41.43%

Count 5 Average 50% not used in calculation.

That is, the search can begin with a range from 41 to 61 percent.
And if we search for those already found with this range, then we will find 58 keys out of 71. That is more than 81 percent of the locations.
In this case, the search is performed in only 20 percent of the possible options, that is, 80 percent of the options recline !!!
If you have not found, increase the range, say up to 31 - 71, while also excluding options from 41 to 61 percent. Etc. Or we proceed to search for the next key.

Please I Have Question to you MrFreeDragon ,Why the one who code kangaroo use CPU ,why didn't use GPU to accelerate the code and get the solution faster , I use the same code of Kangaroo and I have only 10k H/s but other membres result is 100k H/s !! thats mean my cpu is low, but why not use cuda gpu ? and thanks

So you just confirm that the key is random! Your example like "that the same bit could not repeat more than 8 times" just prove that the key is random.

How do you think the real random works? The random number is not selected within the eide range, no. The random is applied to every bit. So, the random entropy was used to create each bit in random way, which is absolutely 50%/50% for every bit.
That's why there is no pattern. That's why it is so hard to guess and find any connections. When you apply 50% probability for every single bit, so you do not guess and imagine how is the whole key will look, it is really not important.

I beleive that all the keys for this puzzle transactions were generated randomly on the lowest level - on bit level.

Yes, the visualizer is designed for chance. The speed is small enough, it’s just sometimes not interesting to watch. But there are not just random numbers, there are also some checks that were discussed in this or a neighboring branch. For example, a repeating sequence of identical bits cannot exceed 8, this is a type, for example. Another algorithm changes only one random bit from the last option.

How many private key can be generated per second using this tool ? I am not sure but when I watch th video its look like take more time + don't forget that using bitcrack and gpu can create more than 100Mkey if you have good resources, so this game should take lot and lot of time than bitcrack.
and at all its just chance, you can get it fast if you have chance

Funny game ) I found there a square 16x16 visualizing the key in BIN format. Last month I also created a tool to generate a bitcoin address in coin mode and visualize the existing keys (https://btckeygen.com/). I made the visualizations of all known puzzle keys to find the pstterns, but could not. I made a conclusion that the keys were random   With no patterns  

It is not clear what you are doing there and what result you get?
Write me in PM.

Thanks for the info regarding the "challenge". I was not sure about the right form :-) As to the percentage ranges - I took them personally into account, but I do not plan to base them in my actions, because they would be shots on the basis of statistics, and again complicating my life. It is not known whether now there will be three subsequent addresses had their key in the range of 90-100% .... and maybe even the rest of the keys below 50% :-)

That is a very complex problem and addressed to experts in programming, I will be watching to see who sends the solution.