Importance of JVM parameters

[Pascal66]

With "-Xms4g", "-Xmx4g" Combined
->SIQS 1 thread: 2732664,400 µs
->PSIQS 7 threads: 2510790,210 µs
Without
->SIQS 1 thread: 3252015,594 µs
->PSIQS 7 threads: 3249436,100 µs

Using 2x 128bits 189527046769670452665699925639138782629 x 145320076287112015128258773072589293279
"27542084895039556995681746820634403757320668248657547717737190830239511650491"

[Pascal66]

Before you ask why:
SISQ GC :
SIQS.testCombined:�gc.count 1 avgt 2,000 counts
SIQS.testCombined:�gc.time 1 avgt 38,000 ms
PSIQS GC :
SIQS.testCombined:�gc.count 7 avgt 55,000 counts
SIQS.testCombined:�gc.time 7 avgt 1312,000 ms

PSIQS use too much Arraylist or any Object creation in the while(true) loop
Too much GC impose 1s of work loosed

[TilmanNeumann]

When this project was hosted on my homepage, I had a proposition for good JVM parameters. It certainly contained the proposal that Xmin and Xmax should be the same, and more stuff. I did not add that advice here because GC behavior has been changing a lot in the last years, and I'ld have to re-test all options.

[Pascal66]

Hello. I’ve made a java version of a twin prime sieve. Maybe usefull in your math library ? PAscal

[TilmanNeumann]

Hi Pascal, I would be interested.

[Pascal66]

Sorry, missed to give you the link :
TwinsPrimesSieve

[TilmanNeumann]

This is a pretty fast program.
Two things that didn't work for me:

1. comma-separated inputs like "0, 1000" give a InputMismatchException
2. How do I print all twin primes? Commenting in printprms() does not seem to help...

[Pascal66]

Hello,
first it's one comma or one space (you entered both)
Second, i've not really tested with low values
it's more usable for a big range of integers (ie 0,200000000000) or more to count numbers of twins pairs
in this case (0,2e11) this is not acceptable to print 424.084.653 of all :)

[TilmanNeumann]

Hi Pascal,
your program looks stable and very fast. I'ld like to put it into my canonical package structure, i.e. somewhere near de.tilman_neumann.jml.primes. Is that ok for you? (I'ld play with it later on, cause right now I have another project)

Btw. I am always a fan of seeing results for small inputs, just for debugging/QA ;-)

[Pascal66]

Hello, I’ve asked Jabari (the mathematician who’s create this sieve) and he’s happy that you put my java code in your package. I just have to modify some http links of his theory. I’ve a little modified the debugging part, but due to the structure of the code it’s not very easy and little buggy. HAvae a nice day PAscal From: TilmanNeumann Sent: Wednesday, December 25, 2019 8:25 PM To: TilmanNeumann/java-math-library Cc: Priveyes ; State change Subject: Re: [TilmanNeumann/java-math-library] Importance of JVM parameters (#18) Hi Pascal, your program looks stable and very fast. I'ld like to put it into my canonical package structure, i.e. somewhere near de.tilman_neumann.jml.primes. Is that ok for you? (I'ld play with it later on, cause right now I have another project) Btw. I am always a fan of seeing results for small inputs, just for debugging/QA ;-) — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub, or unsubscribe.

[TilmanNeumann]

Hi Pascal,
thanks for asking the author.
The program can now be found in package de.tilman_neumann.jml.primes.exact.
Just give me a pull request if you have improvements.
Best wishes
Till