Consider c-random?

[mrjbq7]

If we use the rand() from libc:

FUNCTION: int rand ( ) ;
SINGLETON: c-random
M: c-random random-32* drop rand ;

It's faster than the mersenne-twister that we use by default:

IN: scratchpad 10,000,000 iota >array

IN: scratchpad c-random [ gc [ randomize ] time ] with-random
Running time: 2.38198561 seconds

IN: scratchpad gc [ randomize ] time
Running time: 2.666762887 seconds

[mrjbq7]

See d66fdbb.

[mrjbq7]

The rand() function has RAND_MAX of 2^31-1, which is not quite the same as our random-32* which has max value of 2^32-2. Calling it twice to get one more bit of randomness reduces the performance advantage considerably.

[bjourne]

The difference is even less pronounced on my (linux) machine:

IN: scratchpad gc [ randomize ] time
Running time: 1.222153295 seconds
IN: scratchpad c-random [ gc [ randomize ] time ] with-random
Running time: 1.160819849 seconds

I wouldn't be suprised if factor outperforms windows libc's rand() function.

[bjourne]

Nope, the difference is bigger on windows:

IN: scratchpad gc [ randomize ] time
Running time: 1.425305261 seconds
IN: scratchpad c-random [ gc [ randomize ] time ] with-random
Running time: 1.14878115 seconds

The rand() function works exactly the same on windows and linux so the random.c vocab doesn't need to be exclusive to unix.

[erg]

The good RNG is PCG:

http://www.pcg-random.org/

[erg]

https://en.wikipedia.org/wiki/List_of_random_number_generators

https://github.com/tkaitchuck/Mwc256XXA64/blob/main/impl/src/gen64.rs

[alex-ilin]

I looked through the help-wanted issues, and this one caught my eye, since I like crypto and stuff.
I have a question, though. If I implement it, how will it be used? I mean, should I implement it in Factor, or in C++ of the VM?

[mrjbq7]

I’m not sure this would be faster due to the signed/unsigned issue, but I suggest if you want to play with the code you can try from Factor first: https://github.com/factor/factor/blob/master/extra/random/c/c.factor

…

On Jan 2, 2022, at 6:20 PM, Alexander Ilin ***@***.***> wrote:  I looked through the help-wanted issues, and this one caught my eye, since I like crypto and stuff. I have a question, though. If I implement it, how will it be used? I mean, should I implement it in Factor, or in C++ of the VM? — Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you authored the thread.

[alex-ilin]

Should I change the issue title to "Implement PCG RNG"?

[mrjbq7]

Or open a new issue. That would be cool to have!!

…

On Jan 2, 2022, at 6:46 PM, Alexander Ilin ***@***.***> wrote:  Should I change the issue title to "Implement PCG RNG"? — Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you authored the thread.

[alex-ilin]

If I open a new issue, did I resolve a help-wanted issue or created a new one?

[mrjbq7]

Either!

…

On Sun, Jan 2, 2022 at 6:48 PM Alexander Ilin ***@***.***> wrote: If I open a new issue, did I resolve a help-wanted issue or created a new one? — Reply to this email directly, view it on GitHub <#783 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAF5A2B32BL6WN65PRDCX3UUEFBPANCNFSM4ADV7D7A> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you authored the thread.Message ID: ***@***.***>

[alex-ilin]

I mean, if my goal was to reduce the open issue count, what should I have done to close this one?
Should it maybe be closed already with no action?

[mrjbq7]

The current random.c vocab has an issue returning only 31 bits but it’s also not currently used anywhere. I’d be fine with closing this issue as prototyped in extra/random.c. There are potentially a zillion “new feature” issues so you don’t necessarily need to make one for PCG unless you’d like to have a place to discuss.

…

On Jan 2, 2022, at 6:50 PM, Alexander Ilin ***@***.***> wrote:  I mean, if my goal was to reduce the issue count, what should I have done to close this one? Should it maybe be closed already with no action? — Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you authored the thread.

[alex-ilin]

Some critical comments about the PCG MWC: rust-random/rngs#18

[alex-ilin]

Added PR #2542 with the PCG MWC implementation.

[erg]

Related to #1518