Re: Abnormous memory use for gaussian gcd()?

[hermann@stamm-wilbrandt.de]

> ..., but it correctly computes the sum
> of squares for p given sqrtm1 in single digit milliseconds!!
> p and sqrtm1 are 36401 decimal digit numbers ...
That was aready really cool.

But 10x more digits biggest twin prime (388342-digits) is absolutely 
mind-blowing with halfgcd() !!
I committed new GP script for that:
https://github.com/Hermann-SW/RSA_numbers_factored/blob/main/pari/388342.halfgcd.gp

First on 2.5GHz i7-11850H as in previous postings.
Not single digit microseconds anymore, but sub 100ms ...

$ gp < 388342.halfgcd.gp
...
2095902026788285990253069715467593810181489383352762[+++]
foobar
  ***   last result computed in 87 ms.
done
Goodbye!
$


I use Raspberry 1.8GHz Pi400 computer often (much slower than 
i7-11850H).
Even on that computer largest twin prime halfgcd() gets computed in sub 
half second !

pi@pi400-64:~/RSA_numbers_factored/pari $ gp < 388342.halfgcd.gp
...
8285990253069715467593810181489383352762[+++]
foobar
  ***   last result: cpu time 463 ms, real time 465 ms.
done
Goodbye!
pi@pi400-64:~/RSA_numbers_factored/pari $


Regards,

Hermann Stamm-Wilbrandt.