From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 20309 invoked by alias); 3 May 2004 17:26:21 -0000 Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Archive: List-Post: List-Help: Sender: gcc-bugs-owner@gcc.gnu.org Received: (qmail 20302 invoked by uid 48); 3 May 2004 17:26:21 -0000 Date: Mon, 03 May 2004 17:26:00 -0000 Message-ID: <20040503172621.20301.qmail@sources.redhat.com> From: "aleks at physik dot tu-berlin dot de" To: gcc-bugs@gcc.gnu.org In-Reply-To: <20040401120607.14809.boeck@fhi-berlin.mpg.de> References: <20040401120607.14809.boeck@fhi-berlin.mpg.de> Reply-To: gcc-bugzilla@gcc.gnu.org Subject: [Bug c++/14809] performance depends on allocation methods X-Bugzilla-Reason: CC X-SW-Source: 2004-05/txt/msg00231.txt.bz2 List-Id: ------- Additional Comments From aleks at physik dot tu-berlin dot de 2004-05-03 17:26 ------- REOPEN the bug 14809: i can verify the problem partly. performance with global static arrays of double's is up to 10% higher as with pointers to dynamic (new or malloc) arrays of same size and type!! PSA .. pointer to static arrays time(average over 5runs): (big matrix multiplications, g++) static=21.8sec; PSA=23.4sec ;dyn.new=23.4sec; dyn.malloc=23.0sec (last week i had up to 3 sec diff.!?) result: pointers slow down my prog!? used memory: ca. 05 megabyte computer: kernel 2.4.23-1.SMP i686 Dual Processor Pentium III, 1.2 GHz, 512K L2-Cache, 2530 bogomips Red Hat Linux(2years old, but some updates (gcc,kernel,libs) GCC: -3.3.3 --enable-shared --enable-languages=c,c++,f77 Thread model: posix, gcc version 3.3.3 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14809