RFR: 8284493: Improve computeNextExponential tail performance and accuracy [v16]

[Xin Liu]

On Thu, 1 Dec 2022 21:31:28 GMT, Chris Hennick <duke at openjdk.org> wrote:

>> Chris Hennick has updated the pull request incrementally with one additional commit since the last revision:
>> 
>>   Add parameter to enable/disable fixed PRNG seed
>
> @simonis @navyxliu @turbanoff @rgiulietti This is ready for review with the benchmark results.

hi, @Pr0methean 

I notice that  there is a big regression in L64X1024MixRandom.nextExponential p100 on [c7g.16xlarge](https://aws.amazon.com/ec2/instance-types/c7g/)

L64X1024MixRandom.nextExponential() p100 ns/op: 560 before, 3760 after


but the average ns/op improves. Can you explain the discrepancy? 

btw: I updated the JBS issue to ensure the subject is aligned with your PR. 

Thanks,
--lx

-------------

PR: https://git.openjdk.org/jdk/pull/8131