From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Sun, 04 Jun 2006 18:04:54 -0400 From: Ivan Adzhubey In-reply-to: <20060604210546.GA18120@localhost.localdomain> To: Sisyphus@lists.altlinux.org Message-id: <200606041804.54270.iadzhubey@rics.bwh.harvard.edu> MIME-version: 1.0 Content-type: text/plain; charset=koi8-r Content-transfer-encoding: 8BIT Content-disposition: inline References: <20060604210546.GA18120@localhost.localdomain> User-Agent: KMail/1.9.1 Subject: Re: [sisyphus] ATLAS vs BLAS performance X-BeenThere: sisyphus@lists.altlinux.org X-Mailman-Version: 2.1.7 Precedence: list Reply-To: ALT Linux Sisyphus discussion list List-Id: ALT Linux Sisyphus discussion list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 04 Jun 2006 22:05:07 -0000 Archived-At: List-Archive: List-Post: On Sunday 04 June 2006 17:05, Alexey Tourbin wrote: > Я разобрался, как надо собирать ATLAS. > Вот предварительный benchmark. > > Fortran BLAS: > > mm <- matrix(rnorm(10^6), ncol = 10^3) > > system.time(crossprod(mm)) > > [1] 1.572 0.004 1.580 0.000 0.000 > > ATLAS w/ SSE2: > > mm <- matrix(rnorm(10^6), ncol = 10^3) > > system.time(crossprod(mm)) > > [1] 0.344 0.020 0.369 0.000 0.000 А вот с GotoBLAS 1.2 (P4 2.8GHz, R 2.3.0): > mm <- matrix(rnorm(10^6), ncol = 10^3) > system.time(crossprod(mm)) [1] 0.232 0.012 0.270 0.000 0.000 -- Иван