Re: [Tails-dev] performance test: randomsound vs haveged

Delete this message

Reply to this message
Autor: Jacob Appelbaum
Data:  
Dla: tails-dev
Temat: Re: [Tails-dev] performance test: randomsound vs haveged
adrelanos:
> Hi,
>
> I've done a performance test to answer the following questions:
> - Is randomsound faster than haveged or vice versa?
> - Do they block each other or result in even more entropy available?
>


I've given up on randomsound for actual long term use - it makes sound
entirely unusable in non-obvious ways.

I settled on rng-tools, ekeyd, haveged and ~60 seconds of randomsound; I
kill randomsound after a minute so I may use audio again.

All the best,
Jacob

> All tests have been done in Virtual Box.
>
> Every 1.0s: cat /proc/sys/kernel/random/entropy_avail
> watch -n 1 cat /proc/sys/kernel/random/entropy_avail
>
> no package installed                ~130 and ~180
> rng-tools in VM without rng hardware         no effect
> ekeyd in VM without entropykey             no effect
> randomsound only                 ~130 and ~3800
> haveged only                     ~1100 and ~4000
> randomsound and haveged             ~1100 and ~4000

>
> (~ is defined as, for exmaple ~1100 could be up to 1199.)
>
> Off topic:
>
> http://en.gentoo-wiki.com/wiki/Generating_better_random_numbers#Dieharder
> is an excellent source for information about entropy in general.
>
> I am conducting some entropy quality tests and will share the results soon.
>
> Cheers,
> adrelanos
> _______________________________________________
> tails-dev mailing list
> tails-dev@???
> https://mailman.boum.org/listinfo/tails-dev
>