name: Ron Sercely establishment: Convex Computer Corporation email: sercely@convex.com telephone: 214.497.4667 date sent: 29/Nov/95 date benchmarked (DD/MMM/YY): 29/Nov/95 benchmark suite and revision: Parkbench message passing library and revision: PVM 3.3.10.2 (to be released Dec '95) benchmark: poly1 problem size (if applicable): number of nodes: 4 file number if one of a group: manufacturer: Convex model number: Exemplar 1200 cpu: HP-PA/RISC 7200 cpu speed (MHz): 120Mhz fpu: integrated primary cache: 250kbytes secondary cache: none other cache: none memory: 512M disk and connection: SCSI interconnect type: shared memory switching: none other hardware: none operating system and version: SPP-UX 3.2 (to be released Dec ' 95) compilers, libraries and versions: Convex FORTRAN, version 9.3 Convex CC, version 6.3 compiler switches: -O2 (full optimization) additional tuning parameters: other software: additional comments: PARKBENCH Version 1.0 *** UNOFFICIAL release test package *** This run was with modified code. Modifications were: Within the timing loops, the code segments like: CALL PVMFINITSEND(PVMDEFAULT,INFO) CALL PVMFPACK(BYTE1,A,ILEN,1,INFO) CALL PVMFSEND(TIDS(NSLAVE),10,INFO) CALL PVMFRECV(TIDS(NSLAVE),20,INFO) CALL PVMFUNPACK(BYTE1,A,ILEN,1,INFO) was changed to: CALL PVMFPSEND(TIDS(NSLAVE),10,a,ilen,byte1,INFO) CALL PVMFPRECV(TIDS(NSLAVE),20,a,ilen,byte1,iatid, iatag,ialen,INFO) The PSEND/PRECV functions are Convex versions of SEND/RECV optimized shared memory PVM implementations. It is a documented, standard Convex feature. Also, in synch1, one line of code was inserted just before the timing loop:: CALL PVMFFREEZEGROUP(GROUP,NNODE,INFO) This function is a documented, standard Convex feature, which allows optimized shared memory group operations, such as PVMFBARRIER Environment Variables: PVM_ARCH = CSPP PVMBUFSIZE=1000000 PVMPSEND=0 (enables the PSEND functions) ================================================= === === === GENESIS / PARKBENCH Parallel Benchmarks === === === === POLY1 === === === === Program: In-cache (r-hat f-half) === === Version: Standard Fortran 77 === === Author: Roger Hockney === === === ================================================= POLYNOMIAL BENCHMARK FOR FHALF ------------------------------ (DOUBLE PRECISION) This benchmark tests severity of memory bottlenecks by varying the amount of arithmetic per memory reference which is called the computational intensity of the loop. The performance for long loop (vector) lengths, RINF, is represented as : RINF = RHAT/(1 + FHALF/F) where RHAT = peak Mflop/s rate of arithmetic pipeline approached as F goes to infinity and F = computational intensity = ratio floating operations/memory references FHALF = F required to obtain RINF=RHAT/2 The loop executed is polynomial evaluation by Horners rule, F varies as the order increases from 1 to 10 Roger Hockney, August 1989 ------------------------------------------------------------ POLYNOMIALS : A(I)=S0+B(I)*(S1+B(I)*(S2+B(I)*S3)) ICASE= 1 NCASE= 1 NMAX= 10000 NTIMES= 1000000 NPROCS= 1 COMPUTATIONAL INTENSITY = flop per mem ref 1.0 ------------------------------------------ Floating operations per iteration = 2 Memory references per iteration = 2 LOOP LOOP RINF N1/2 S1/2 ERROR R-AVERAGE T0 NTIM LENGTH SECS MFLOP/S VEC.LEN FLOP % MFLOP/S SECS repeat 1 2.676E-07 .000000 .000 .000E+00 .00 7.472728 2.676E-07 0 5 4.691E-07 39.718224 4.315 8.630E+00 .00 21.319290 2.173E-07 0 10 7.278E-07 39.097504 4.210 8.419E+00 2.00 27.480118 2.153E-07 0 20 9.437E-07 56.974247 8.151 1.630E+01 60.70 42.387257 2.861E-07 0 50 1.740E-06 69.645126 11.691 2.338E+01 38.64 57.467632 3.357E-07 0 100 3.124E-06 71.622116 12.405 2.481E+01 20.18 64.020027 3.464E-07 0 200 5.713E-06 74.268669 13.744 2.749E+01 12.05 70.013008 3.701E-07 0 300 8.273E-06 75.522675 14.595 2.919E+01 8.90 72.523613 3.865E-07 0 400 1.106E-05 75.136253 14.253 2.851E+01 6.49 72.311852 3.794E-07 0 600 1.630E-05 75.301613 14.462 2.892E+01 4.23 73.623955 3.841E-07 0 800 2.169E-05 75.181992 14.266 2.853E+01 3.07 73.778435 3.795E-07 0 1000 2.700E-05 75.156494 14.214 2.843E+01 2.37 74.079437 3.783E-07 0 COMPUTATIONAL INTENSITY = flop per mem ref 2.0 ------------------------------------------ Floating operations per iteration = 4 Memory references per iteration = 2 LOOP LOOP RINF N1/2 S1/2 ERROR R-AVERAGE T0 NTIM LENGTH SECS MFLOP/S VEC.LEN FLOP % MFLOP/S SECS repeat 1 3.153E-07 .000000 .000 .000E+00 .00 12.688269 3.153E-07 0 5 6.440E-07 48.663189 2.835 1.134E+01 .00 31.053852 2.331E-07 0 10 9.797E-07 54.380821 3.454 1.382E+01 16.07 40.828030 2.540E-07 0 20 1.296E-06 79.662125 7.109 2.844E+01 65.09 61.705444 3.570E-07 0 50 2.506E-06 93.999451 9.787 3.915E+01 36.62 79.795753 4.165E-07 0 100 4.601E-06 95.512878 10.161 4.064E+01 18.40 86.940292 4.255E-07 0 200 8.612E-06 97.525848 10.878 4.351E+01 9.82 92.894005 4.461E-07 0 300 1.248E-05 99.212051 11.699 4.680E+01 7.75 96.179901 4.717E-07 0 400 1.675E-05 98.603096 11.306 4.523E+01 5.80 95.510689 4.587E-07 0 600 2.463E-05 99.086159 11.758 4.703E+01 4.01 97.461815 4.746E-07 0 800 3.284E-05 98.959610 11.604 4.642E+01 2.90 97.452736 4.690E-07 0 1000 4.133E-05 98.372330 10.710 4.284E+01 3.13 96.785683 4.355E-07 0 COMPUTATIONAL INTENSITY = flop per mem ref 3.0 ------------------------------------------ Floating operations per iteration = 6 Memory references per iteration = 2 LOOP LOOP RINF N1/2 S1/2 ERROR R-AVERAGE T0 NTIM LENGTH SECS MFLOP/S VEC.LEN FLOP % MFLOP/S SECS repeat 1 3.264E-07 .000000 .000 .000E+00 .00 18.383259 3.264E-07 0 5 7.932E-07 51.414307 1.797 1.078E+01 .00 37.822433 2.097E-07 0 10 1.297E-06 55.812801 2.158 1.295E+01 12.92 46.277317 2.320E-07 0 20 1.748E-06 82.192299 5.262 3.157E+01 70.89 68.631439 3.841E-07 0 50 3.461E-06 98.603859 7.864 4.718E+01 39.98 86.676483 4.785E-07 0 100 6.437E-06 100.592018 8.294 4.977E+01 19.91 93.211945 4.947E-07 0 200 1.213E-05 102.911049 9.035 5.421E+01 10.59 98.931694 5.268E-07 0 300 1.757E-05 104.955650 9.942 5.965E+01 8.57 102.440163 5.684E-07 0 400 2.359E-05 104.507057 9.676 5.806E+01 6.18 101.754219 5.555E-07 0 600 3.472E-05 105.092468 10.183 6.110E+01 4.31 103.680130 5.814E-07 0 800 4.634E-05 104.976509 10.052 6.031E+01 3.10 103.589195 5.745E-07 0 1000 5.789E-05 104.814148 9.822 5.893E+01 2.44 103.642715 5.622E-07 0 COMPUTATIONAL INTENSITY = flop per mem ref 4.0 ------------------------------------------ Floating operations per iteration = 8 Memory references per iteration = 2 LOOP LOOP RINF N1/2 S1/2 ERROR R-AVERAGE T0 NTIM LENGTH SECS MFLOP/S VEC.LEN FLOP % MFLOP/S SECS repeat 1 3.593E-07 .000000 .000 .000E+00 .00 22.267252 3.593E-07 0 5 9.619E-07 53.102455 1.385 1.108E+01 .00 41.585201 2.086E-07 0 10 1.678E-06 54.668987 1.497 1.198E+01 4.69 47.689373 2.191E-07 0 20 2.368E-06 76.951744 3.905 3.124E+01 63.92 67.577034 4.059E-07 0 50 4.832E-06 91.342430 6.088 4.870E+01 36.32 82.788315 5.332E-07 0 100 9.030E-06 94.005737 6.658 5.326E+01 18.42 88.590225 5.666E-07 0 200 1.716E-05 96.181473 7.364 5.891E+01 9.87 93.212967 6.125E-07 0 300 2.521E-05 97.240410 7.848 6.278E+01 6.91 95.202858 6.456E-07 0 400 3.373E-05 96.895042 7.634 6.108E+01 5.01 94.858887 6.303E-07 0 600 4.985E-05 97.321663 8.024 6.419E+01 3.46 96.287537 6.596E-07 0 800 6.685E-05 96.927322 7.551 6.041E+01 2.80 95.732826 6.232E-07 0 1000 8.315E-05 96.932777 7.559 6.048E+01 2.15 96.210663 6.239E-07 0 COMPUTATIONAL INTENSITY = flop per mem ref 5.0 ------------------------------------------ Floating operations per iteration = 10 Memory references per iteration = 2 LOOP LOOP RINF N1/2 S1/2 ERROR R-AVERAGE T0 NTIM LENGTH SECS MFLOP/S VEC.LEN FLOP % MFLOP/S SECS repeat 1 3.934E-07 .000000 .000 .000E+00 .00 25.416868 3.934E-07 0 5 1.133E-06 54.107925 1.129 1.129E+01 .00 44.142208 2.086E-07 0 10 2.152E-06 51.090328 .930 9.304E+00 9.24 46.470421 1.821E-07 0 20 3.358E-06 64.302513 2.311 2.311E+01 43.03 59.555340 3.594E-07 0 50 7.318E-06 72.360413 3.574 3.574E+01 23.92 68.320045 4.940E-07 0 100 1.398E-05 74.075279 3.978 3.978E+01 12.16 71.545845 5.370E-07 0 200 2.696E-05 75.547546 4.531 4.531E+01 6.85 74.183601 5.998E-07 0 300 3.998E-05 76.022461 4.790 4.790E+01 4.62 75.028374 6.300E-07 0 400 5.372E-05 75.583710 4.461 4.461E+01 3.77 74.465454 5.901E-07 0 600 7.956E-05 75.852959 4.764 4.764E+01 2.63 75.414787 6.281E-07 0 800 1.059E-04 75.930801 4.881 4.881E+01 1.92 75.528137 6.428E-07 0 1000 1.328E-04 75.801476 4.636 4.636E+01 1.60 75.329414 6.115E-07 0 COMPUTATIONAL INTENSITY = flop per mem ref 6.0 ------------------------------------------ Floating operations per iteration = 12 Memory references per iteration = 2 LOOP LOOP RINF N1/2 S1/2 ERROR R-AVERAGE T0 NTIM LENGTH SECS MFLOP/S VEC.LEN FLOP % MFLOP/S SECS repeat 1 4.676E-07 .000000 .000 .000E+00 .00 25.662003 4.676E-07 0 5 1.267E-06 60.009483 1.338 1.606E+01 .00 47.337612 2.676E-07 0 10 1.915E-06 75.208374 2.293 2.751E+01 38.45 62.657494 3.658E-07 0 20 3.215E-06 85.134468 3.177 3.812E+01 28.63 74.643044 4.478E-07 0 50 7.179E-06 89.392159 3.724 4.469E+01 13.12 83.578392 5.000E-07 0 100 1.365E-05 91.148346 4.062 4.875E+01 7.17 87.900345 5.348E-07 0 200 2.669E-05 91.698952 4.231 5.077E+01 3.57 89.913651 5.536E-07 0 300 3.975E-05 91.802795 4.277 5.132E+01 2.26 90.556099 5.591E-07 0 400 5.279E-05 91.861443 4.313 5.176E+01 1.62 90.919991 5.634E-07 0 600 7.928E-05 91.590233 4.062 4.874E+01 1.34 90.815117 5.322E-07 0 800 1.050E-04 91.793716 4.312 5.174E+01 1.21 91.453133 5.637E-07 0 1000 1.311E-04 91.855522 4.409 5.290E+01 .96 91.511086 5.759E-07 0 COMPUTATIONAL INTENSITY = flop per mem ref 7.0 ------------------------------------------ Floating operations per iteration = 14 Memory references per iteration = 2 LOOP LOOP RINF N1/2 S1/2 ERROR R-AVERAGE T0 NTIM LENGTH SECS MFLOP/S VEC.LEN FLOP % MFLOP/S SECS repeat 1 4.688E-07 .000000 .000 .000E+00 .00 29.862980 4.688E-07 0 5 1.413E-06 59.304508 .986 1.380E+01 .00 49.536942 2.327E-07 0 10 2.207E-06 73.005730 1.775 2.485E+01 36.57 63.424236 3.403E-07 0 20 3.773E-06 82.426369 2.573 3.602E+01 27.80 74.208168 4.370E-07 0 50 8.484E-06 87.280891 3.182 4.454E+01 13.41 82.509270 5.104E-07 0 100 1.633E-05 88.440849 3.403 4.764E+01 6.78 85.736153 5.387E-07 0 200 3.220E-05 88.439301 3.402 4.763E+01 3.18 86.959595 5.386E-07 0 300 4.787E-05 88.671303 3.507 4.910E+01 2.12 87.744888 5.538E-07 0 400 6.350E-05 88.866341 3.630 5.082E+01 1.68 88.195892 5.719E-07 0 600 9.490E-05 88.992615 3.750 5.250E+01 1.14 88.514595 5.900E-07 0 800 1.263E-04 89.034096 3.802 5.323E+01 .83 88.643745 5.979E-07 0 1000 1.585E-04 88.822235 3.463 4.848E+01 1.10 88.315170 5.458E-07 0 COMPUTATIONAL INTENSITY = flop per mem ref 8.0 ------------------------------------------ Floating operations per iteration = 16 Memory references per iteration = 2 LOOP LOOP RINF N1/2 S1/2 ERROR R-AVERAGE T0 NTIM LENGTH SECS MFLOP/S VEC.LEN FLOP % MFLOP/S SECS repeat 1 5.191E-07 .000000 .000 .000E+00 .00 30.822340 5.191E-07 0 5 1.546E-06 62.303402 1.021 1.634E+01 .00 51.735229 2.623E-07 0 10 2.390E-06 77.557076 1.866 2.986E+01 38.50 66.939102 3.850E-07 0 20 3.461E-06 107.312714 4.274 6.838E+01 63.31 92.472122 6.372E-07 0 50 8.130E-06 106.742241 4.210 6.735E+01 24.11 98.405861 6.310E-07 0 100 1.484E-05 112.287766 5.128 8.204E+01 15.06 107.806099 7.306E-07 0 200 2.905E-05 112.943031 5.297 8.475E+01 7.20 110.163574 7.504E-07 0 300 4.326E-05 112.922478 5.289 8.463E+01 4.52 110.957535 7.494E-07 0 400 5.748E-05 112.850594 5.252 8.404E+01 3.22 111.341072 7.447E-07 0 600 8.600E-05 112.681992 5.124 8.198E+01 2.09 111.628967 7.276E-07 0 800 1.151E-04 112.203987 4.641 7.426E+01 2.04 111.205055 6.619E-07 0 1000 1.428E-04 112.421738 4.920 7.872E+01 1.74 112.078590 7.002E-07 0 COMPUTATIONAL INTENSITY = flop per mem ref 9.0 ------------------------------------------ Floating operations per iteration = 18 Memory references per iteration = 2 LOOP LOOP RINF N1/2 S1/2 ERROR R-AVERAGE T0 NTIM LENGTH SECS MFLOP/S VEC.LEN FLOP % MFLOP/S SECS repeat 1 5.940E-07 .000000 .000 .000E+00 .00 30.305471 5.940E-07 0 5 1.748E-06 62.394188 1.059 1.906E+01 .00 51.490192 3.055E-07 0 10 2.752E-06 75.580246 1.796 3.233E+01 33.33 65.412926 4.277E-07 0 20 4.097E-06 100.782387 3.865 6.957E+01 55.36 87.864830 6.903E-07 0 50 9.605E-06 100.968582 3.886 6.995E+01 21.12 93.702385 6.928E-07 0 100 1.779E-05 105.032494 4.585 8.252E+01 12.55 101.155411 7.857E-07 0 200 3.491E-05 105.446831 4.697 8.454E+01 5.97 103.112679 8.018E-07 0 300 5.204E-05 105.398903 4.678 8.421E+01 3.75 103.760536 7.989E-07 0 400 6.919E-05 105.320694 4.636 8.344E+01 2.67 104.062927 7.923E-07 0 600 1.041E-04 104.758873 4.180 7.524E+01 2.30 103.705574 7.182E-07 0 800 1.379E-04 104.903976 4.336 7.805E+01 1.72 104.446320 7.440E-07 0 1000 1.721E-04 104.997520 4.464 8.035E+01 1.36 104.619835 7.653E-07 0 COMPUTATIONAL INTENSITY = flop per mem ref 10.0 ------------------------------------------ Floating operations per iteration = 20 Memory references per iteration = 2 LOOP LOOP RINF N1/2 S1/2 ERROR R-AVERAGE T0 NTIM LENGTH SECS MFLOP/S VEC.LEN FLOP % MFLOP/S SECS repeat 1 5.695E-07 .000000 .000 .000E+00 .00 35.117290 5.695E-07 0 5 1.514E-06 84.721436 1.413 2.825E+01 .00 66.059311 3.335E-07 0 10 2.140E-06 115.958122 2.829 5.657E+01 54.14 93.459099 4.879E-07 0 20 4.020E-06 113.012115 2.645 5.290E+01 25.28 99.503090 4.681E-07 0 50 9.094E-06 116.110062 2.930 5.861E+01 10.60 109.962616 5.047E-07 0 100 1.719E-05 119.674477 3.434 6.867E+01 7.44 116.346848 5.738E-07 0 200 3.413E-05 119.237556 3.334 6.668E+01 3.53 117.192062 5.592E-07 0 300 5.039E-05 120.125305 3.632 7.263E+01 2.98 119.067047 6.046E-07 0 400 6.740E-05 119.893448 3.524 7.047E+01 2.20 118.694389 5.878E-07 0 600 1.025E-04 118.434235 2.497 4.995E+01 3.81 117.094643 4.217E-07 0 800 1.365E-04 117.894661 1.991 3.982E+01 3.12 117.202568 3.378E-07 0 1000 1.687E-04 118.401649 2.597 5.194E+01 2.92 118.581985 4.387E-07 0 CALCULATION OF FHALF -------------------- F RINF RHAT FHALF ERROR f/m Mflop/s Mflop/s flop/mref % 1.0 75.156494 .000000 .0000 .00 2.0 98.372330 142.341644 .8939 .00 3.0 104.814148 130.577957 .7099 1.04 4.0 96.932777 108.493294 .3079 3.50 5.0 75.801476 79.209671 -.3150 7.56 6.0 91.855522 85.447510 -.1561 7.51 7.0 88.822235 86.465553 -.1262 5.78 8.0 112.421738 100.941414 .3550 9.64 9.0 104.997520 105.902229 .5362 7.84 10.0 118.401649 115.972336 .9359 8.82 LAST VALUES : RHAT = 115.972336 Mflop/s FHALF = .9359 flop/mref TOTAL EXECUTION TIME IS 3.0630049706E+00 SECONDS.