Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!uwm.edu!ux1.cso.uiuc.edu!ux1.cso.uiuc.edu!ejk From: ejk@ux1.cso.uiuc.edu (Ed Kubaitis) Newsgroups: comp.windows.x Subject: xfroot timing update Message-ID: <1989Oct15.105523.18023@ux1.cso.uiuc.edu> Date: 15 Oct 89 10:55:23 GMT Sender: news@ux1.cso.uiuc.edu (News) Reply-To: ejk@ux1.cso.uiuc.edu (Ed Kubaitis) Followup-To: xfroot timing update Organization: University of Illinois at Urbana Lines: 81 Here is the 2nd updated list of xfroot fractal-points/processor_second measured on various clients. The number, a count of trips/second through the 8 line "hopalong" loop in xfroot, is a rough index of scalar double-precision floating point uniprocessor speed. New items since the last posting are marked with ">". Cray X-MP 157,000 to 194,000* (*=per processor) Cray 2 129,000 to 183,000* > Convex C2 (gcc) 117,000 to 151,000* > Convex C2 (vc3/fastmath) 108,000 to 138,000* > Convex C2 (vc3) 99,000 to 118,000* DEC DS5800 95,000 to 115,000 DEC DS5400 77,000 to 91,000 DEC DS3100 58,000 to 75,000 Gould NP1 44,000 to 60,000* DEC Vax 6400 (vcc) 50,000 to 57,000 Convex C2 (vc2) 49,000 to 55,000* Convex C2 (cc) 41,000 to 47,000* > Dec Vax 8650 28,000 to 33,000 Sun Sparcstation 1 ~25,000 to DEC MV3900 (vcc) 22,900 to 26,100 > DG AViiON (88k 16.7 MHz) 17,200 to 24,200 Sun 4/380 ~21,000 to ~23,000 > Dec Vax 8530 19,700 to 23,200 > Dec Vax 8600 19,700 to 22,400 DEC Vax 6220 16,800 to 19,200 > DEC MV3200 (vcc) 15,400 to 17,500 > IBM RT 135 (-f2 -lfm) 15,200 to 17,400 > DEC MV3600 (vcc) 14,500 to 17,400 > HP9000/360 13,700 to 15,200 DEC Vaxserver 3500 13,200 to 15,200 > Dec Vaxstation 3100 13,000 to 15,100 > IBM RT 135 10,600 to 11,500 > HP9000/350 10,500 to 11,500 Sun 3/50 (gcc 68881) ~11,200 Sequent Symmetry 9,900 to 10,500* Sun 3/60 (-f 68881) 8,000 to 8,750 > IBM RT 125 7,200 to 7,600 DEC Vaxstation 2000/vcc 5,530 to 6,330 > HP9000/330 5,730 to 6,230 > DEC Vax 780 5,410 to 6,170 > HP9000/320 5,580 to 6,150 Sun 3/50 (-f 68881) 5,480 to 6,080 DEC Vaxstation 2000 4,670 to 5,530 > DEC MVII (cc) 4,160 to 5,210 > DEC MVII (vcc) 4,080 to 5,070 Sun 3/60 1,960 to 2,060 Sun 3/50 1,270 to 1,330 Sun 3/160 (no fpa) ~950 > DEC Vax 730 340 to 360 A few notes on the results: gcc enhancements are due to inline code for sqrt and fabs. The three top Convex C2 measurements use compilers/libraries that exploit the C2 hardware sqrt. It pays to shop around for the best compiler/options/libraries available for your floating point intensive code. Thanks to: bryan%kewill@uunet.uu.net, csmith@convex.com, eric@geology.tn.cornell, evans@decvax.dec.com, glenn@mathcs.emory.edu, harrison@decwrl.dec.com, hleroy@erisa.fr, howard@aic.hrl.hac.com, idallen@watgcl.waterloo.edu, jw@pan.uucp, kline@ux1.cso.uiuc.edu, markw@airgun.wg.waii.com, rauletta@gmuvax2.gmu.edu, steved@longs.lance.colostate.edu, for sharing their results. I would appreciate hearing about measurements on other clients, or results differing significantly from those above. To perform your own 1. Get xfroot/part01 (V5-I3) and xfroot/patch1(V5-I7) from comp.sources.x. 2. Install xfroot on the client to be tested, taking care that you have verified the definition of HZ in xfroot.c. (See the README.) 3. Make the following two runs: xfroot -a 0.1 -b 0.1 -c 0.1 (lower bound) xfroot -a 3000 -b 3000 -c 3000 (upper bound) ------------------------- Ed Kubaitis (ejk@ux1.cso.uiuc.edu) Computing Services Office - University of Illinois, Urbana