From owner-freebsd-current Thu May 2 10:01:03 1996 Return-Path: owner-current Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id KAA10034 for current-outgoing; Thu, 2 May 1996 10:01:03 -0700 (PDT) Received: from ormail.intel.com (ormail.intel.com [134.134.192.3]) by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id KAA10029 for ; Thu, 2 May 1996 10:00:56 -0700 (PDT) Received: from ichips.intel.com (ichips.intel.com [134.134.50.200]) by ormail.intel.com (8.7.4/8.7.3) with ESMTP id KAA25118; Thu, 2 May 1996 10:00:40 -0700 (PDT) Received: from pdx202 by ichips.intel.com (8.7.4/jIII) id JAA19776; Thu, 2 May 1996 09:58:09 -0700 (PDT) Message-Id: <199605021658.JAA19776@ichips.intel.com> To: asami@cs.berkeley.edu (Satoshi Asami) cc: current@freebsd.org, nisha@cs.berkeley.edu Subject: Re: more on fast bcopy In-reply-to: Your message of "Thu, 02 May 1996 02:35:47 PDT." <199605020935.CAA06965@silvia.HIP.Berkeley.EDU> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Date: Thu, 02 May 1996 10:00:39 -0700 From: Wayne Scott Sender: owner-current@freebsd.org X-Loop: FreeBSD.org Precedence: bulk I think you might find these results interesting. This is the script run on a P6 box. Pentium Pro 200/256 128 Meg memory 2-way interleave B-step Orion chipset The interesting results is that 'libc' is MUCH faster than any of the other results. We implimented a fast string copy mode for 'rep movs' that kicks in at about 128 elements. Wayne Scott MD6 Architecture wscott@ichips.intel.com Work #: (503) 264-4165 Disclaimer: All views expressed are my own opinions, and not necessarily those of Intel Corporation. # This is a shell archive. Save it in a file, remove anything before # this line, and then unpack it by entering "sh file". Note, it may # create directories; files and directories will be owned by you and # have default permissions. # # This archive contains: # # pdxlx008/fpunrolled.out # pdxlx008/libc.out # pdxlx008/plot.gp # pdxlx008/unrolled128.out # pdxlx008/unrolled144.out # pdxlx008/unrolled32.out # pdxlx008/unrolled48.out # pdxlx008/unrolled64.out # pdxlx008/unrolled96.out # echo x - pdxlx008/fpunrolled.out sed 's/^X//' >pdxlx008/fpunrolled.out << 'END-of-pdxlx008/fpunrolled.out' X31.927970 X28.838598 X33.324169 X34.343018 X34.618547 X34.948538 X35.172875 X35.307306 X35.343182 END-of-pdxlx008/fpunrolled.out echo x - pdxlx008/libc.out sed 's/^X//' >pdxlx008/libc.out << 'END-of-pdxlx008/libc.out' X32.017417 X32.604354 X33.238603 X39.687264 X43.481569 X45.450414 X46.639616 X47.106484 X47.434955 END-of-pdxlx008/libc.out echo x - pdxlx008/plot.gp sed 's/^X//' >pdxlx008/plot.gp << 'END-of-pdxlx008/plot.gp' Xset xtics ("32" 0, "64" 1, "128" 2, "256" 3, "512" 4, "1024" 5, "2048" 6, "4096" 7, "8192" 8) Xplot 'pdxlx008/fpunrolled.out' w linesp, 'pdxlx008/unrolled144.out' w linesp, 'pdxlx008/unrolled128.out' w linesp, 'pdxlx008/unrolled96.out' w linesp, 'pdxlx008/unrolled64.out' w linesp, 'pdxlx008/unrolled48.out' w linesp, 'pdxlx008/unrolled32.out' w linesp, 'pdxlx008/libc.out' w linesp END-of-pdxlx008/plot.gp echo x - pdxlx008/unrolled128.out sed 's/^X//' >pdxlx008/unrolled128.out << 'END-of-pdxlx008/unrolled128.out' X31.792206 X32.426736 X35.571681 X35.386643 X35.226152 X35.223980 X35.189893 X35.092952 X35.121300 END-of-pdxlx008/unrolled128.out echo x - pdxlx008/unrolled144.out sed 's/^X//' >pdxlx008/unrolled144.out << 'END-of-pdxlx008/unrolled144.out' X31.879623 X32.474122 X33.214315 X28.795623 X35.488679 X35.381009 X35.256580 X35.281147 X35.299828 END-of-pdxlx008/unrolled144.out echo x - pdxlx008/unrolled32.out sed 's/^X//' >pdxlx008/unrolled32.out << 'END-of-pdxlx008/unrolled32.out' X30.772308 X35.372562 X35.711097 X34.900055 X34.853832 X34.834406 X34.804399 X34.747559 X34.817426 END-of-pdxlx008/unrolled32.out echo x - pdxlx008/unrolled48.out sed 's/^X//' >pdxlx008/unrolled48.out << 'END-of-pdxlx008/unrolled48.out' X31.844852 X35.743966 X35.232668 X35.396664 X35.211578 X35.172565 X35.219018 X35.085257 X35.271502 END-of-pdxlx008/unrolled48.out echo x - pdxlx008/unrolled64.out sed 's/^X//' >pdxlx008/unrolled64.out << 'END-of-pdxlx008/unrolled64.out' X31.792964 X35.621098 X35.714605 X35.439315 X35.407317 X35.515463 X35.568518 X35.576427 X35.632522 END-of-pdxlx008/unrolled64.out echo x - pdxlx008/unrolled96.out sed 's/^X//' >pdxlx008/unrolled96.out << 'END-of-pdxlx008/unrolled96.out' X31.846373 X32.406507 X35.524294 X36.170290 X36.104342 X36.246987 X36.461419 X36.407657 X36.381827 END-of-pdxlx008/unrolled96.out exit