From: Radical DOT NetSurfer AT delorie DOT com Newsgroups: comp.os.msdos.djgpp Subject: Benchmark Testing Date: Wed, 26 Sep 2001 01:09:24 -0400 Organization: Posted via Supernews, http://www.supernews.com Message-ID: X-Newsreader: Forte Agent 1.8/32.548 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Complaints-To: newsabuse AT supernews DOT com Lines: 62 To: djgpp AT delorie DOT com DJ-Gateway: from newsgroup comp.os.msdos.djgpp Reply-To: djgpp AT delorie DOT com Listen up...some of you might find this quite interesting. For some time I've had a nice little project in Borland that attempts to do simple spell-checking and correcting, quite simply, all it does is look for specific sequence of characters, either delimited by whitespaces or punctuation, and then if it matches what is hoped to be a mis-spelling, replaces that sequence of characters with a string from its vocabulary. #define Vocab_a 512 char *voc_a[Vocab_a] = { "aword1", "aword1_substitute", "aword2", "aword2_substitute", ... "awordn", "awordn_substitute" }; and there are dozens of such arrays hard-coded into an include file. The point to all this is: Under pure MSDOS 6.22 it takes 10 minutes 10 seconds to process a 2MB file, using a 16-bit non-dpmi version of the program, Under Borland Console32 (which must run under Win95/Dos7Shell), the best time I get for the same identical 2MB file is 8 mins 15 secs, NOW! Introduce the DJGPP 2.03/2.95.3 port or the very same program, and behold: Pure MSDOS 6.22, above mentioned identical 2MB file takes 6 mins 35 secs, Under Win95/Dos7: 6 mins 45 secs ** TEN WHOLE SECONDS FASTER in "pure" MSDOS 6.22 ** The very worst times I've logged are under Windows 95's DosShell on a 486DX2/66, using "pure" MSDOS 6.22 on a 486 VASTLY improves performance. (lots to consider why there) I do not know if I'll want to contaminate my project by porting it to (ick) msvc 6.0, but I suppose to be thorough I will eventually have to do just that....(God help me for what my hands will be touching). DJGPP's SIX MINUTE PROCESSING time really surprised me! and the output file, from all tests, are thoroughly compared... I know the same exact task is being performed.... Some of DJGPP's (seemingly) silly compiler statements notwithstanding, I need to give 2.03 more consideration. email: radsmail AT juno DOT com http://members.tripod.com/~RadSurfer/ Join us on Yahoo at: http://groups.yahoo.com/group/BorlandCPPBuilder for informal discussions about all versions of C++Builder... [Computer programming for Windows 95, and Console32] (you must be a Yahoo member to join)