From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 30665 invoked by alias); 2 Oct 2002 14:56:02 -0000 Mailing-List: contact gcc-prs-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Archive: List-Post: List-Help: Sender: gcc-prs-owner@gcc.gnu.org Received: (qmail 30643 invoked by uid 71); 2 Oct 2002 14:56:01 -0000 Resent-Date: 2 Oct 2002 14:56:01 -0000 Resent-Message-ID: <20021002145601.30642.qmail@sources.redhat.com> Resent-From: gcc-gnats@gcc.gnu.org (GNATS Filer) Resent-Cc: gcc-prs@gcc.gnu.org, gcc-bugs@gcc.gnu.org Resent-Reply-To: gcc-gnats@gcc.gnu.org, o.lauffenburger@topsolid.com Received: (qmail 30595 invoked by uid 61); 2 Oct 2002 14:55:52 -0000 Message-Id: <20021002145552.30594.qmail@sources.redhat.com> Date: Wed, 02 Oct 2002 07:56:00 -0000 From: o.lauffenburger@topsolid.com Reply-To: o.lauffenburger@topsolid.com To: gcc-gnats@gcc.gnu.org X-Send-Pr-Version: gnatsweb-2.9.3 (1.1.1.1.2.31) Subject: optimization/8126: Floating point computation far slower in 3.2 than in 2.95 X-SW-Source: 2002-10/txt/msg00055.txt.bz2 List-Id: >Number: 8126 >Category: optimization >Synopsis: Floating point computation far slower in 3.2 than in 2.95 >Confidential: no >Severity: serious >Priority: medium >Responsible: unassigned >State: open >Class: pessimizes-code >Submitter-Id: net >Arrival-Date: Wed Oct 02 07:56:01 PDT 2002 >Closed-Date: >Last-Modified: >Originator: Olivier Lauffenburger >Release: gcc version 3.2 >Organization: >Environment: Cygwin 1.3.12-2 / Windows 2000 SP3 / Pentium III 800MHz >Description: The enclosed file test.ii computes one hundred million vector products. Compiled with gcc 2.95, it takes 2770 ms. Compiled with gcc 3.2, it takes 4470 ms. A examination of the assembly code generated by gcc 3.2 shows that the stack is used instead of the floating point registers. The code generated by gcc 3.2 with optimization -O3 looks like the one generated by gcc 2.95 with optimization -O1. >How-To-Repeat: g++ -O3 -ffast-math -fomit-frame-pointer test.ii -o optim.exe >Fix: With -funroll-loops, the code is better (but still slower than gcc 2.95) >Release-Note: >Audit-Trail: >Unformatted: ----gnatsweb-attachment---- Content-Type: application/x-gzip-compressed; name="optim.tar.gz" Content-Transfer-Encoding: base64 Content-Disposition: attachment; filename="optim.tar.gz" H4sICOsFmz0AA29wdGltLnRhcgDtPNty27iSeZW+gpuTmZE8tsObZGU057apc7a2Kg9be3nKpFAU SSk8pkgVL7acKf/7duNCAiAoUk5m5kWoikWiL+huNBrdIJkqLqvbJHn1Wzbbse2l77+yod0t6S/0 OPTetr3F3WLxyr5b+Mul7TrOHeB7rue8suzfVCre6rIKCst69fdon2RJWRVBFddFF+/vdVhnv4dA v2/7k+VYryt0gvBweD2ltz8nWRUXWfwXcZ8muyy2otgK8/0+yKIGohG+rcvibZKFaR3Fb8sqSvLb z68B4AHYfdcH9wzEJMjKpCV2Fn1wE3H5VL4N82yb7CQOyz4OLoV7dp94CMfWjpPmYZC+3YXhjXfr vk2TDb3G32S5Wt4cwpvwafeYZG8RLPGL4q1kD/sbsfOm1dMhhhurBo1goiIL5s8qky8xqdaomm48 TbWvVywoJEv73rfh1upFyKZO0irJyENAUlij0LPL6lDcUiVXQ/OnaCq7ShHHWdXK3/FTGcWbxkdc G9br96+tX08wJSh8KfmfexKpVTbNs51F8u0WJ2/adOOUkrKd1M6KUMQUOnc8g3En//cBfqnZ7k5r C/GwDmHkXVrH01+nliV3WFckA3OsoZuKlyX5plxLSKT8539++AegMcAzKCRA/57sgEZhybp0pvfX FtkHx8eohAvUAn7gBuFcD4scPzqfGPuGP6n2lDnlATekjMO1fA/RXrn/nNeFihAFT2pHrlI8xYFK 8ahTPOkdSRmhuz6j6e9MHmEwPexHx6SyUBnWNEBrL9boYEkWtT0PeRJZsyuyzcqPnvtpPsOOuQZH MKw8itGCGhNvM+qrXPZ3J72vlZ2Um3orid74YfgZNt0rsgnKWJMcXZzPpezBzHG3h7wU/q8H9F4R qA+iDProh2Zu2nl8pP77OS/gepsGu1K+T9JYdW/Ujmy2DXW62TL5WROmtUiY5/cJ9lvNKgb6oKqK ZFNXMSGzGSFhFIcpIfM5TlYRBxFcqAyuG8PV22u+6OhEjmT6WCRVbOKaZxBVjbyROTP7AO8yju8N rBtiCGllXHHOj5/jLIzn7eI4xTlM87IrNcrWnYx6szbMdN1ONa7yaQcF6OrtR+9Tl5hkCMH4Yhgt 3bBuynmT3rPJ5/dMYZWOOigIFAVVABC6mFy3k3uYXbmA1Mtfqa7MXROMH+nic9C+TqseUBBFzFWp HL4/sPUpSrTRNS6KLF9zh++EftyHIWbTi7yu+BXQrBvDwXAoHTN3vI+LHTjH00d38anFCeuigFFJ CFFvlxc0rCpeKxBo1hG3hKSMEpAgqdbNejztbeBucZDVB7jRZk3zuGa/KuISbNwGE3pL7tcm3MOi NPVD9IVBWDLT6vxQpXHWmOYKO8Dt1sx9kzyjkVNEV8ayCaZt0LXUxJDUWV3GEfWmtYTExgBXqooq vydpUFYdMAnKsEr2McaIj+7yk4wg7bwWmwWBKWNRCXYBlBF0M4crGUoIplcgbZ4RomUtbAdAqbUN T1pdbIGQwl91xtxvwJagHPiPAVblj2EP8DGs8v2mC2QGSZd+QK2x+tQFovhByozlf+qaIYZkF3yQ LgYBeebutmbryTizpqkV04e2gUhmK8OpDpDtgxQmSEV6Fo7Bup6R06Pq8E3CwS7WXRD/tdft5je7 upqhIci2zsL5fAYSzPVgQZNJWKn4a8gf4ZfFZhqmlu5A4sQzdD3kJvtDXcTkUBUWjM8CQQHrPkj2 HMew3mE4fyA6P2Po7CQjcvVhrhC0AsHzT+F437r8dPy7b11+om+BdaOE1S6mwlNWyWyZfRB+TrJY t46jT0IHT6ugWOznZZRSRJFDEtHaSg02CrXEIs0bdUwZp6bOwlRPGWRUl21N8GfdhbJ9uib01wBH EWvYP00wqkNNUlrpmWpBzvwk74QHIjN3SMzCe1anapk6Df0cwFcVdpWHOKQ7E4Nb1YOoyxjVA8no vVQpJohaNIQ6s6Qi9LTqIUjXZjBAalFRKDJGkP4UsnvwEBqK/j61kyznRYjtjZtuZmrmUPDXWRKT xWWsukFTjNuw8Fwji5aYIbA25OmU49JHAjOqaSemozCiXks97HkWbBRWoIhTDYmNvDpgmJWRnq8d Hd6Y8aSZ6+5EoPFOm7buml6JF8wcJ6xWkw4W52gV8Q7Sv7jgrrU07XIG16JRl0U0RaT7+EnuawrE srW25vf6ckDO+zwSB06rk5uYIbQxO2dpkt0zFu7Y5UIV2EZkH5T3wofb6eWbNKUjgAXuJWW7nAx+ S7JJqvLjbDZb+vMf4Qc1z7fWjKPMQefVfH7jzOdv+4BzzDrUsagmnqlUOmWMNl7y3ad3zdB4Zwyq kJ1ByROVDGjOKrZxUEGaI22drmkXltG0M80/We9M06RSPFPGJ6eTn6sP7YhdNXi6oWcA+mA6mqzG 9NdTB77wD9KHIpeNNCQEPVHVXLDZZcQFgfKWXiOczpJ6fIXBsC96UGAnnDYkVudEmNfwJIofegId 5wsIzWbQ8uVkFvyqfCkVP8sw6QCgsD8ICrhBk5bQYpc9e6xQrE6ikxGckF0vRrNFRT3RnCPsBIIq Kh8a8h8lsDYjWrvuKmanU4fqM57cyVQqYF9DiaTYVYVrcVsF4rnFAOvTKBg/hjE0R2PlJ4ZYTQmL /oqDD14hP08FVp6Fcf9AxaPIHk/BhbCYbOkVUjcCPNOI03kU1A1LficqKY+L1Ic8zfTzo0xxCq3v SLxcxT/r3tijPhVddKo+rXDshs/2sSk7YTt1nFXtD3hobYkTf3FIcJIiC/bWjGECycDx7JYezloz JssI/G0KFYeEP6zEtojzQ5yBTNJ5n3pmfW01/IZO+GAHx6NbLoA4UB8WHOgeDIT0PPuaP20dof6h AAyZi6LF7e3tCB5lGGRfx0JI8QJSPvhLKI2kL2D0MGBF9enwCH5mc5zNpmz4GHU7l992F1ehtFKG Fy9SlO3w1DNHL8xDjaOdTVPqZhtLrGk3AhsHGB/JZFuMWBHna48kVCR2pDmMrZlqmKbOmJFUsYaf 9G1xAxXPy0RwYo9W27vsPJb00aFQgD+JG8N6hJPDdtouZb7HjiHFJ44tIWZ01OdHUaqDMq2koXl6 2M+hitPUmo3fdooYUpVI8vchAnwKVcRFccbWSuvX8dhFkRdnCHTgBJoLj9nAu9s3gSRDf+6M6dSI XeR0kB21gRTxPn+Iz12NRYxCn9q/6IONwc0r+ba7V/IVm/mQLOP29WTMlIxIPWORe54y8OAmnOnS sKj01UnC6bTr/GTBlAydz2VETtVhOjihI03IJhZfYOh993NMkbKNWHygu5y+nAYdOI2z/Kw04nE8 9kEvboaVOQyVKqMyhUd9zwcj+72vcY4xMmFWJkXnKSfPEzUZx3DsZfg12hO2zox8vyJIEZ5AGviO Ck5IX5qVHasYD5MjLDZOo/19iSHztFAjNNvfD7MZSg4Iyw4GdRuh1je2Es/S+xx/HIeemT9PMZZ1 fBtOmIkMciJ5Gmk9Wfw4IqqS/iX4kikoT03py7M4Ivapk4xPbV2DA4y1w8sSHsIznvNj6IjozI/+ XhzvqhOijfVTcVBzRjA/NwMiD6Pjxfmsz3fcF4zxcic+f7Cz99YX6DNyybyA88tWYzf9HfGuNKEv 58Vp0hyFt+conZO1sfzSJIuN7BpOtA0ZgpQFMBufxQLBIz25VqVmjb+leyLDrjM5pdVf78axZ1d4 3rTN5mPeVhck9DzJSNP3LnpzNjS7wpMfE630rrn6orkYlSb0Xco5TbE9+8TDFny6lYT0VRmcR0LE 68WYWM6EGQ/zqXj1loFD68/W7OZmhqCbv4Dz/mzZ1l/FJNJu6yd2hDi7EliHH3+c8+8zEpg37IZe +jmC9b1lH33btufW999bMzrAn60ffil+mM+n4v1Q6R1TErpfKQGTgTHCkbIf5s3LFkw/hAncOC1j 6XMPahyE4zo8NByf6W8RV3WRITn202d3qxNfCdJ3DqRvDi1XfzVkm+YBvkiD74LRVbEPkgw1s4Ji FzJ3vMLrh4+f5tP2nRGksB6c4zX8ecI/X9YazEWYizC3A/MQ5iHM+9K8rSJQWHCfvf6fKiiqv/6S vZa+u6HGmUkvpoLh6evyc/bdBWsgF9jYuZW+13GeoMdWer7wnrbLPepI7pPOyG3IVKm3OTvWthKE r+HnZ8uxeYNb8A7ji+acrYdDo5RXdIgbKh9eP60NyE8UmSEcKfKREZqQv1BkhvBEkdkox/W0i00t R2FdEB3VKBC1JR2+C6NGBf0MZNS6oI0JRDl6kkLPvW6SH/6NeUkH9t/0uwLg9B0EN/YPMK9VB2wC u6Dmi8zGB/OG739xNZFdGLq35W/2jfHp77/thWPfie+/Hdtx8fvvO39x+f7792iTW6wSJtKX3OgM sC3uD9Af3f40JTSBanpIeEjrEv/9NJ3cQtSdAALG2fXktgzTiQu/GI8nHl7FGWBMb6v4WE0/vLeR JCjDJLFEQHTcX+zXAHIUEC4CAXEliGkJMLxGlDpKHqLEOyHNh/eexPJ/k32MDBNrXza8Avzi3fKn t7s036QW1U8MMaArhQP/sgbCN67tX38Xl4fp5FCXn1Pru3gjXUdJe11K15vjdBIGKQxM+NBBFAG3 G8dVuL0BkwpMFiWmk33+gHTtpgIUwbHLgeL5MwTOOYoYPmiGxx1qOtmm0RdO8Maxlwtv4bnO8tpZ rWbIa04xSku9t74rq5kjXbt4XVaI5ws8JpQnZALwobKWDXh7DD/LtAD02zHoveeq985S3DMN/iOu 3rMvw/6ryMO4LP9mc1VQz2un5acbKMXN3XnH+V2rM7dpEdwGQZ5OvGYIwirX6hyXLYLfDnFUnYAK 6ty1MpinCZTk2qE3l39zhY6y+IHg9+Ydb9AZik7Xtls9jj0m4p0RdC5crXNzVBxN9TNUlzMGd5Bn id5Kk4i3/kq5bf3hX/uD9cFZTj84q5/YhCve0XMjO1GzsoHNT61v2o0Pt3qpzrevU3qzAAOVVR/U F9BRvHydl98skSHSpZm03lC9r6W1h8xaawN9i6FYxxvNwCz7gttdRZfXvoopj+TOO1PRkGEhajFH /VdWwvSvpPm15303XudGtqM5nPYH50hbdp11KSNQL2fByO+GMqcNVa4WyxzH7QSv4XW9MS3hNtLA HuHoewRXX2xNbEm2i5BtXSsFaq96NUmtU4Re15MHCRVYo4Zr3OqkyKOHGspVDq+IVG42alBup+6N 3VzRYqd//sXmzJONJugtuQbs1l8ptzcrzdvaYA7aeeZJalge84IHZE6YH8QuwS/pnLPLqL3cNOO3 uQjUBCKbwR2eZzOeIZvhWEymE3h/dBp7aS9sov7z/rj6z79z7Kb+c5wFrf8Wl///63dp3fpvbFF3 RsU1OVEA0hrP61R/tNsfVfp9eL8YV8/BBnJ7cHkGeA057NfUdzRwT1h8xY1ngqH1mgdcDNYTGqzl MD9py4LJjY+b9QZ2IDnVmLR1weSG7pwco5QwNnxrm0CS7VMghvgsgo4b2C54B24mMDv0S/+gueWa UoGxery25IxjQp91c7BSQ0p5z0QUjXIf01Xl1RSPDs8MJzcLrpKoKTHZndzQBH/D8yC8t6V7SraU yFQ2DGG5UhEUjohw11Aw2XrLQk0TqYsZnk2N60uT19oACiOryfj4NK9aTMlakKNYK22QCPoc1zCy WtNOetNCGE2SS5R2XRmY87i2cBU+VAhD3bxrDCUJcPNuqfS2FSTnqC2rD7y2UqYOE/kJQ8cqQpkU MUvaPEuOwCikeQYlCpVkLA/dN1Z2r3d15WSjrnQPHeQhS84ofI1iOSSnvgokDiCWimAeo8vS1RB0 KfUxO1O21BTvzMdSk6FjOn3trvrXLhSAkw/LxjU30kLtrrbQuFo3R8Xf26Ul8fS1vuHFes66xGBv XIGK9BjwRZT2zFFaN7hA91t0WklN2iLUMAPpxHE1uK/CXVsbnRcoVIkmNKhC0/pJt9obW7YZFmKT Jt4Iy0gmN1jcVyWRCjApWqnTBwZZdOzHNWDu4cj7rHHzhkpq0pZak7bUmrSlFk8FpNIKZ2m4tHpB AWZKS/7oPPLSLu3SLu3SLu3SLu3SLu3SLu3SLu3SLu3SLu3SLu3SLu3SLu3SLu3SLu3SLu3S/rj2 /76a0IsAeAAA