Linux's getrandom() Sees A 8450% Improvement With Latest Code

[phoronix]

Phoronix: Linux's getrandom() Sees A 8450% Improvement With Latest Code

The Linux kernel's random number generator code has been seeing a number of improvements recently led by Jason Donenfeld of WireGuard fame...

https://www.phoronix.com/scan.php?page=news_item&px=Linux-getrandom-8450p

[beevvy]

Wow. Just like with WireGuard - simpler, faster, more secure than the old designs. Jason Donenfeld is such a great engineer.

[mdedetrich]

Not over 9000, not impressed.

[zx2c4]

Not over 9000, not impressed.

Complete trash, I agree.

[Random_Jerk]

Does this make my NumPy np.random.random calls on Python faster on Linux?

[sinepgib]

Does this make my NumPy np.random.random calls on Python faster on Linux?

It depends on build flags, there's a strong random mode that uses the kernel's facilities and there's a mode that only generates pseudo random numbers. Note that this is only used for the seed from what I could see in my quick check (I may or may not have missed something), so it's only for the first call. Everything else is arithmetic only.

[CochainComplex]

Not over 9000, not impressed.

Krilin should train it a bit more

[discordian]

Not over 9000, not impressed.

They had a 9000% improvement, but this did not seem random enough.

PS. I am usually against adblockers, I just dont watch sites that abuse them. Right now the adds appear over the "Post reply" button which royally pisses me off.

[Quackdoc]

PS. I am usually against adblockers, I just dont watch sites that abuse them. Right now the adds appear over the "Post reply" button which royally pisses me off.

give adnauseam a try, it hides ads doesn't block them, and apparently spoofs clicks on them (I have mine set to fairly conservative, but by default it clicks on pretty much all ads)

[rmfx]

Is it just me or there is a kernel’s bottlenecks hunt these past few months. Love it.