Base Flags and Notes:
ONESTEP=yes for all benchmarks
C: -fast -xopenmp -xalias_level=std -xipo=2 -Xc
-xprefetch_level=2 -xcode=abs44 -xarch=v9a -lmtmalloc
-xprofile
f90: -fast -openmp -xcode=abs44 -xarch=v9a -xipo=2 -autopar
-xprofile
submit = ppgsz -o heap=4m,stack=4m $command
Extra art Base Flags:
331.art_l: -DINTS_PER_CACHELINE=16 -DDBLS_PER_CACHELINE=8
Peak Flags and Notes:
ONESTEP=yes for all benchmarks
311.wupwise_l: -fast -openmp -autopar -xunroll=2
-Qoption iropt -Athr,-Ainline:inc=800:cp=1
-xipo=2 -xprefetch_level=3 -xarch=v9a
-xprofile
submit = ppgsz -o heap=512k,stack=512k $command
313.swim_l: -fast -openmp -xarch=v9a -xipo=2 -autopar
-xprofile
submit = ppgsz -o heap=512k,stack=512k $command
315.mgrid_l: -fast -openmp -xipo=2 -xprefetch_level=3
-xprefetch=latx:2 -xprofile
-Qoption iropt -Apf:largedim -xarch=v9a
submit = ppgsz -o heap=512k,stack=512k $command
OMP_NUM_THREADS=128
317.applu_l: -fast -openmp -xipo=2
-Qoption iropt -Aujam:inner=g -xunroll=1
-xprefetch=latx:1.8 -xarch=v9a -xprofile
submit = ppgsz -o heap=4m,stack=4m $command
OMP_NUM_THREADS=133
321.equake_l: -fast -xopenmp -xipo=2 -xcode=abs44
-xrestrict -xalias_level=strong
-xprefetch_level=3 -xprefetch=latx:2.5
-W2,-Apf:pdl=1 -Wc,-Qlp-ip=1-ol=1
-xarch=v9a -lmtmalloc -xprofile
submit = ppgsz -o heap=64k,stack=64k $command
OMP_NUM_THREADS = 117
325.apsi_l: -fast -openmp -xcode=abs44
-xipo -autopar -xarch=v9a -xprofile
submit = ppgsz -o heap=4m,stack=4m $command
OMP_NUM_THREADS = 140
327.gafort_l: -fast -openmp -autopar -xarch=v9a -xprofile
submit = ppgsz -o heap=4m,stack=4m $command
329.fma3d_l: -fast -openmp -xipo=2 -xprefetch_level=3
-Qoption iropt -Athr,-Apf:pdl=1
-Qoption cg -Qlp-ip=1
-xarch=v9a -lmtmalloc -xprofile
submit = ppgsz -o heap=512k,stack=512k $command
331.art_l: -fast -xopenmp -xipo=2
-xalias_level=strong -xprefetch_level=2
-W2,-Apf:outer=0:pdl=1 -Wc,-Qlp-ip=1
-xarch=v9a -lmopt -lm -lmtmalloc -xprofile
submit= ppgsz -o heap=512k,stack=512k $command
Alternate Source for Peak:
Change initial data distribution for WORK array
Submitted Source: ompl2001-dd-20040128.tar.gz
325.apsi_l
Feedback optimization (-xprofile) is done as follows,
unless otherwise noted:
fdo_pre0: rm -rf `pwd`/../feedback.profile
PASS1: -xprofile=collect:./feedback
PASS2: -xprofile=use:./feedback
Base and Peak User Environment:
unlimit stacksize (in /bin/csh)
setenv STACKSIZE 16384
setenv OMP_NUM_THREADS 143
setenv OMP_DYNAMIC FALSE
setenv MT_BIND_PROCESSOR "0 4 1 5 2 6 3 7
8 12 9 13 10 14 11 15 16 20 17 21 18 22 19 23
24 28 25 29 26 30 27 31 32 36 33 37 34 38 35 39
40 44 41 45 42 46 43 47 48 52 49 53 50 54 51 55
56 60 57 61 58 62 59 63 64 68 65 69 66 70 67 71
72 76 73 77 74 78 75 79 80 84 81 85 82 86 83 87
88 92 89 93 90 94 91 95 96 100 97 101 98 102 99 103
104 108 105 109 106 110 107 111
112 116 113 117 114 118 115 119
120 124 121 125 122 126 123 127
128 132 129 133 130 134 131 135
136 140 137 141 138 142 139 143"
Kernel Paramters (/etc/system):
set autoup=900
set tune_t_fsflushr=1
Sun Studio 9 will be generally available July 2004
System configuration details located at
http://www.sun.com/servers/
|