*To*: James McDonald <mcdonald at kestrel.edu>*Subject*: Re: [isabelle] 64-bit Java is 6x faster than 32-bit for a recursive fibonacci*From*: Gottfried Barrow <igbi at gmx.com>*Date*: Mon, 19 May 2014 04:52:35 -0500*Cc*: Makarius <makarius at sketis.net>, Richard Waldinger <waldinger at AI.SRI.COM>, cl-isabelle-users at lists.cam.ac.uk*In-reply-to*: <52196068-A081-4408-808D-664DBD11D97C@kestrel.edu>*References*: <5374E9FA.1010201@gmx.com> <5374F12E.2030901@gmx.com> <alpine.LNX.2.00.1405161628320.3424@lxbroy10.informatik.tu-muenchen.de> <5376399E.5060306@gmx.com> <8B7099FA-D82C-4E2A-B510-4F006D368326@ai.sri.com> <5376A3BA.3040007@gmx.com> <E5A9AA96-C476-445D-97ED-F5C10BC1624F@kestrel.edu> <52196068-A081-4408-808D-664DBD11D97C@kestrel.edu>*User-agent*: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:11.0) Gecko/20120312 Thunderbird/11.0

On 14-05-17 18:25, James McDonald wrote:

More generally, if you are really trying to benchmark variouslanguages and implementations,you might want to consider a broader suite of tests. E.g., withinthe lisp world, you couldlook at Dick Gabriel’s “Performance and Evaluation of Lisp Systems”from around 1985.(http://www.dreamsongs.com/NewFiles/Timrep.pdf)

James (and Makarius below),

http://github.com/gc44/prelim/tree/master/1400/1405A_bigint_prog_lang_perf_tests On 14-05-16 15:44, Makarius wrote:

On Fri, 16 May 2014, Gottfried Barrow wrote:A starting point is to try to and find as many languages that performgood with recursive functions that use pattern matching.You should look in the vicinity of ML and Haskell. The latter isparticularly well-known to beat most other languages in the usualbenchmarks.

If a recursive fibonacci function isn't relevant, then why isIsabelle/HOL filled with impractical recursive functions?There is nothing impractical about recursion. The slowness of thenaive fibonacci implementation is caused by the non-linearity of theinvocation, which causes an exponential blowup.

But surely blowups are good for stress tests.

I know that recursive functions are bad performers, and recursivefunctions with pattern matching are even worse.This was disproven in the 1980s or so, when the first efficientimplementations of ML came about. Recursion without pattern matchingwas already known to be efficient much earlier, in the olden days ofLISP.

I want as much speed as possible, because I think I'll need it.

Regards, GB

**References**:**[isabelle] 64-bit Java is 6x faster than 32-bit for a recursive fibonacci***From:*Gottfried Barrow

**Re: [isabelle] 64-bit Java is 6x faster than 32-bit for a recursive fibonacci***From:*Gottfried Barrow

**Re: [isabelle] 64-bit Java is 6x faster than 32-bit for a recursive fibonacci***From:*Makarius

**Re: [isabelle] 64-bit Java is 6x faster than 32-bit for a recursive fibonacci***From:*Gottfried Barrow

**Re: [isabelle] 64-bit Java is 6x faster than 32-bit for a recursive fibonacci***From:*Gottfried Barrow

**Re: [isabelle] 64-bit Java is 6x faster than 32-bit for a recursive fibonacci***From:*James McDonald

- Previous by Date: Re: [isabelle] building the afp sample submission
- Next by Date: [isabelle] Isar parser for constant names
- Previous by Thread: Re: [isabelle] 64-bit Java is 6x faster than 32-bit for a recursive fibonacci
- Next by Thread: [isabelle] building the afp sample submission
- Cl-isabelle-users May 2014 archives indexes sorted by: [ thread ] [ subject ] [ author ] [ date ]
- Cl-isabelle-users list archive Table of Contents
- More information about the Cl-isabelle-users mailing list