Path: csiph.com!usenet.pasdenom.info!news.albasani.net!newsfeed.freenet.ag!news2.euro.net!postnews2.euro.net!news.wanadoo.nl!not-for-mail From: mhx@iae.nl (Marcel Hendrix) Subject: Re: Forth as a polarising language Newsgroups: comp.lang.forth Message-ID: <16910094948435@frunobulax.edu> Date: Sun, 9 Sep 2012 03:12:19 +0200 References: <2097404.2psJB3vpnK@sunwukong.fritz.box> X-Newsreader: iForth 2.0 console (October 21, 2006) Lines: 36 Organization: Wanadoo NNTP-Posting-Date: 09 Sep 2012 01:12:19 GMT NNTP-Posting-Host: s529d937f.adsl.wanadoo.nl X-Trace: 1347153139 dr7.euro.net 225 82.157.147.127:58703 X-Complaints-To: abuse@wanadoo.nl Xref: csiph.com comp.lang.forth:15547 Bernd Paysan writes Re: Forth as a polarising language > Marcel Hendrix wrote: > > At least your heap/heap1 code is easy to understand and interesting, > > but Mini-OOF results in a ridiculous 7..50 times slowdown. > Ok, I did the same changes to heap.fs which make VFX happy, and here are > the benchmark results with VFX: [.. see below ..] > The worst slowdown is for the 4k ins. > If you see 7..50 times slowdown, your compiler probably "sucks" or > whatever the derogative word of the month is ;-). No, I don't see that. In the earliest postings I showed the correct numbers, but the "7..50 times" come from (1) erroneously comparing the linear heap with the binary one, not from mini-OOF binary to normal binary, (2) from erratic timing / slow cache warming on Windows. Below I use the best of 3 runs results. heap heap1 heap heap1 -----------------------------------------+------------------- VFX VFX | iForth iForth -----------------------------------------+------------------- 10 inserts 11,617ns 7,323ns | 13 us 11 us 10 deletes 7,421ns 5,064ns | 9 us 7 us 256 inserts 65,836ns 45,393ns | 53 us 26 us 256 deletes 173,988ns 116,564ns | 100 us 67 us 4096 inserts 1,052,728ns 462,301ns | 527 us 282 us 4096 deletes 3,781,767ns 2,640,110ns | 1,894 us 1,317 us There is not much of a difference. iForth indeed inlines DOES> code. -marcel