X-Recipient: archive-cygwin AT delorie DOT com X-SWARE-Spam-Status: No, hits=-2.2 required=5.0 tests=AWL,BAYES_00 X-Spam-Check-By: sourceware.org MIME-Version: 1.0 Date: Tue, 16 Jun 2009 12:06:49 +0200 From: "Vincent R." To: Subject: Re: Optimize cygwin on recent windows version (Vista and Seven) In-Reply-To: <50535.99.237.216.211.1245125771.squirrel@www.sidefx.com> References: <7c6dcbb9c797277cc8ffb1fc985844af AT mail DOT smartmobili DOT com> <3353982C81F6441590DD8E4B4C2D0841 AT desktop2> <50535 DOT 99 DOT 237 DOT 216 DOT 211 DOT 1245125771 DOT squirrel AT www DOT sidefx DOT com> Message-ID: X-Sender: forumer AT smartmobili DOT com User-Agent: RoundCube Webmail/0.2 Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset="UTF-8" X-IsSubscribed: yes Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm Precedence: bulk List-Id: List-Unsubscribe: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-owner AT cygwin DOT com Mail-Followup-To: cygwin AT cygwin DOT com Delivered-To: mailing list cygwin AT cygwin DOT com On Tue, 16 Jun 2009 00:16:11 -0400 (EDT), "Edward Lam" wrote: > On Mon, June 15, 2009 19:53, Sisyphus wrote: >> Here are some timings I did recently for building the mpc-0.6 library. >> On Vista and XP, (in the same version of the MSYS shell, and using the >> same >> version of MinGW's gcc) I ran: >> >> ./configure --disable-shared --enable-static >> CPPFLAGS=-I/usr/local/include >> LDFLAGS=-L/usr/local/lib && make && make check >> >> On linux (mdk-9.1) and cygwin, it was the same command, but without the >> CPPFLAGS and >> LDFLAGS arguments (as they're not necessary on linux and cygwin). >> >> Times taken were: >> Linux : 1.5 mimutes >> XP (mingw): 6.5 minutes >> Vista (mingw): 16.5 minutes >> Vista (cygwin): 23.25 minutes Yes these figures are a good example of what I am talking about and here is an additional benchmark done when compiling binutils (time ../build-mingw32ce.sh -j2 --comp=binutils) Core 2 Duo 2.4 GHz - Windows Seven 64 bits - no antivirus - Windows Search(indexing) disabled real 16m54.470s user 3m20.164s sys 6m25.794s Core 2 Duo 2.6 GHz - Windows XP 32 bits - antivirus disabled real 5m44.375s user 15m22.277s ???? (don't really understand this figure) sys 5m44.702s Even if processor is slightly more powerful on XP, time is smaller by a factor 3 !!! I also tried to launch mintty with administrator rights just to be sure the problem doesn't come from security limitations and I havent' noticed any improvements. Now the question is how do you profile time consumed by APIs ? I tried to used StraceNT but it just killed my console when I attached it and since during a compilation process are created then destroyed it's hard to monitor. I read also on some blogs dedicated to low level stuff that it's better to use malloc before 16 kB and VirtualAlloc for bigger size, I am not saying that this is the solution but things could be improved by that kind of tricks. Maybe it's better to use some API in some context. Ideally I would like to stay on WIndows XP but on the other hand I cannot stop time forever ... -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Problem reports: http://cygwin.com/problems.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/