[
https://issues.apache.org/jira/browse/RNG-57?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16627396#comment-16627396
]
Alex D Herbert commented on RNG-57:
-----------------------------------
{quote}I don't understand the purpose of a CachedUniformRandomProvider interface
{quote}
It was a class private marker interface. ;)
I showed the method in case you could spot something silly. I am going to think
about it and try some variants.
Anyway the private classes implemented the marker interface to avoid double
wrapping. Here is the public API:
{code:java}
/**
* Factory class for wrapping instances of {@link UniformRandomProvider} to
cache
* values that can be reused to provision the interface methods.
*
* <p>The int values generated by an {@link IntProvider} can be cached to enable
* fast provision of {@link UniformRandomProvider#nextBoolean()}.
* <p>The long values generated by a {@link LongProvider} can be cached to
enable
* fast provision of {@link UniformRandomProvider#nextBoolean()} and
* {@link UniformRandomProvider#nextInt()}.
*/
public final class CachedUniformRandomProviderFactory {
// PRIVATE STUFF
/**
* Wrap the source of randomness.
*
* <p>The returned provider will cache values from an {@link IntProvider} or
* {@link LongProvider} to enable fast provision of
* {@link UniformRandomProvider#nextBoolean()},
* and in the case of a {@link LongProvider} also
* {@link UniformRandomProvider#nextInt()}.
*
* <p>If the source of randomness cannot be wrapped then it is returned
unmodified.
*
* @param rng the source of randomness
* @return the wrapped uniform random provider
*/
public static UniformRandomProvider wrap(UniformRandomProvider rng) {
// Avoid double wrapping
if (rng instanceof CachedUniformRandomProvider) {
return rng;
}
if (rng instanceof LongProvider) {
return new CachedLongProvider((LongProvider)rng);
}
if (rng instanceof IntProvider) {
return new CachedIntProvider((IntProvider)rng);
}
// Unknown implementation
return rng;
}
}
{code}
Anyway the API details can be sorted. My aim was to see if I could make it work.
Currently the {{LongProvider}} uses bit masking on longs. However the results
appear to show the {{nextInt}} is faster or equal for all {{LongProviders}}. So
I'm going to test a version that uses {{int}} bit masking for the
{{nextBoolean}}, having generated an {{int}} using a cached {{long}}.
> CachedUniformRandomProvider for nextBoolean() and nextInt()
> -----------------------------------------------------------
>
> Key: RNG-57
> URL: https://issues.apache.org/jira/browse/RNG-57
> Project: Commons RNG
> Issue Type: Improvement
> Components: sampling
> Affects Versions: 1.2
> Reporter: Alex D Herbert
> Priority: Minor
> Labels: performance
>
> Implement a wrapper around a {{UniformRandomProvider}} that can cache the
> underlying source of random bytes for use in the methods {{nextBoolean()}}
> and {{nextInt()}} (in the case of {{LongProvider}}). E.g.
> {code:java}
> LongProvider provider = RandomSource.create(RandomSource.SPLIT_MIX_64);
> CachedLongProvider rng = new CachedLongProvider(provider);
> // Uses cached nextLong() 64 times
> rng.nextBoolean();
> // Uses cached nextLong() twice
> rng.nextInt();
> IntProvider provider = RandomSource.create(RandomSource.KISS);
> CachedIntProvider rng2 = new CachedIntProvider(provider);
> // Uses cached nextInt() 32 times
> rng2.nextBoolean();
> // This could be wrapped by a factory method:
> UniformRandomProvider rng = CachedUniformRandomProviderFactory.wrap(
> // Any supported source: IntProvider or LongProvider
> RandomSource.create(RandomSource...));
> {code}
> The implementation should be speed tested to determine the benefit for
> {{nextBoolean()}} and if {{nextInt()}} can be improved for {{LongProviders}}.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
