Is there a way to fill in free RAM on a Linux machine with random data?
The reason I ask: I work in a group where we program numerical programming in Fortran. Sometimes people randomly work with double-precision actions, so programs that should give double-precision results give only the same accuracy.
If my understanding is correct, you can see random fluctuations in the result after a single limit of accuracy in such a program. That is, if you run the same program with the same input several times, each time you get a random result. What you see (random part) depends on random values in the free RAM of the machine. But in practice, if you run the program again on the same computer, the same parts of the memory are usually reused that have the same random data in them, each time outputting the same output.
My idea is that if you could overwrite the memory with random data, you really will see random fluctuations in the output of your program. This would make it much easier to find these errors.
Is this idea a blow, and if not, how can I fill my memory? Can I plug in / dev / random in RAM or something else?
source
share