Goldbach's Conjecture

[science_man_88]

should have the 400 million test done in 12-13 minutes if it worked linearly

[CRGreathouse]

Originally Posted by science_man_88

should have the 400 million test done in 12-13 minutes if it worked linearly

It will take slightly longer than linear. The expected number of tests rises slowly, and the cost of each test is higher as the numbers get bigger.

[science_man_88]

Originally Posted by CRGreathouse

It will take slightly longer than linear. The expected number of tests rises slowly, and the cost of each test is higher as the numbers get bigger.

means the next test may have to be done on a faster computer as linearly it should be over 10 hours on my machine(I only have 11 until i usually try to go to bed). oh and then the next test should take over 100 hours or over 4 days.

*** last result computed in 1h, 2mn, 56,844 ms.

[science_man_88]

does your code break if it finds one before running out ? if not wouldn't that likely speed it up ? it only needs to find one not all of them.

[3.14159]

does your code break if it finds one before running out ? if not wouldn't that likely speed it up ? it only needs to find one not all of them.

If it happens to find one, just press Ctrl + C after it is found.

[science_man_88]

Originally Posted by 3.14159

If it happens to find one, just press Ctrl + C after it is found.

so i need it showing again ? I changed it to not showing also I think it only prints once it's done the gone through the loop.

[CRGreathouse]

Originally Posted by science_man_88

does your code break if it finds one before running out ?

If it finds one what, a Goldbach counterexample? No, the code doesn't break -- although you won't find one, all small numbers have already been tested.

[science_man_88]

Originally Posted by CRGreathouse

If it finds one what, a Goldbach counterexample? No, the code doesn't break -- although you won't find one, all small numbers have already been tested.

no if it finds a prime pair so it only has to find one not 1000 lol

[CRGreathouse]

Originally Posted by science_man_88

no if it finds a prime pair so it only has to find one not 1000 lol

Oh, certainly. The code would take vastly longer otherwise, and it would need a large primelimit.

[science_man_88]

is there a way we could speed it up as data is rarely random according to a ASM book I read into lol.

[CRGreathouse]

Originally Posted by science_man_88

is there a way we could speed it up as data is rarely random according to a ASM book I read into lol.

Did you have something particular in mind?

You could get a small speedup by installing (possibly as a second OS rather than replacing your primary) some form of Linux, installing gp2c, running the script through gp2c, and hand-editing the C code to type-specialize, as well as to replace general functions with custom specific functions. But for this problem, most of the time is just spent proving primality, so there's no a whole lot to be done.