SPEC CFP2000 Summary Hewlett-Packard Company AlphaServer GS1280 Model 64 Tue Jul 15 17:20:51 2003 SPEC License #2 Test date: Jun-2003 Hardware availability: Oct-2003 Tester: HP Software availability: Oct-2003 Base Base Base Peak Peak Peak Benchmarks # Copies Run Time Rate # Copies Run Time Rate ------------ -------- -------- -------- -------- -------- -------- 168.wupwise 64 184 644 64 77.3 1536 168.wupwise 64 184 647* 64 77.3 1537 168.wupwise 64 183 648 64 77.3 1537* 171.swim 64 89.2 2580 64 89.2 2580 171.swim 64 88.8 2591 64 88.8 2591 171.swim 64 88.9 2588* 64 88.9 2588* 172.mgrid 64 257 519 64 171 781* 172.mgrid 64 257 519* 64 171 782 172.mgrid 64 257 519 64 171 781 173.applu 64 141 1106* 64 139 1119 173.applu 64 141 1107 64 139 1123* 173.applu 64 141 1105 64 139 1123 177.mesa 64 155 669* 64 131 795 177.mesa 64 155 669 64 130 797 177.mesa 64 155 669 64 131 796* 178.galgel 64 140 1538* 64 139 1544 178.galgel 64 140 1540 64 139 1546* 178.galgel 64 140 1538 64 139 1549 179.art 64 141 1372 64 84.8 2275 179.art 64 140 1374 64 84.4 2287 179.art 64 141 1372* 64 84.5 2284* 183.equake 64 256 377* 64 84.2 1147 183.equake 64 256 377 64 84.0 1149* 183.equake 64 256 377 64 83.9 1150 187.facerec 64 178 791 64 160 879 187.facerec 64 174 809* 64 157 897 187.facerec 64 174 809 64 157 897* 188.ammp 64 307 532 64 267 612 188.ammp 64 307 532 64 267 611* 188.ammp 64 307 532* 64 267 611 189.lucas 64 133 1118 64 124 1194 189.lucas 64 133 1114 64 122 1214 189.lucas 64 133 1115* 64 123 1209* 191.fma3d 64 209 745 64 159 981 191.fma3d 64 209 745* 64 158 984* 191.fma3d 64 209 746 64 158 986 200.sixtrack 64 242 338* 64 221 370 200.sixtrack 64 242 337 64 223 367* 200.sixtrack 64 242 338 64 223 365 301.apsi 64 218 884* 64 207 934 301.apsi 64 218 884 64 207 931 301.apsi 64 219 883 64 207 932* ======================================================================== 168.wupwise 64 184 647* 64 77.3 1537* 171.swim 64 88.9 2588* 64 88.9 2588* 172.mgrid 64 257 519* 64 171 781* 173.applu 64 141 1106* 64 139 1123* 177.mesa 64 155 669* 64 131 796* 178.galgel 64 140 1538* 64 139 1546* 179.art 64 141 1372* 64 84.5 2284* 183.equake 64 256 377* 64 84.0 1149* 187.facerec 64 174 809* 64 157 897* 188.ammp 64 307 532* 64 267 611* 189.lucas 64 133 1115* 64 123 1209* 191.fma3d 64 209 745* 64 158 984* 200.sixtrack 64 242 338* 64 223 367* 301.apsi 64 218 884* 64 207 932* SPECfp_rate_base2000 813 SPECfp_rate2000 1068 HARDWARE -------- Hardware Vendor: Hewlett-Packard Company Model Name: AlphaServer GS1280 Model 64 CPU: Alpha 21364 CPU MHz: 1150 FPU: Integrated CPU(s) enabled: 64 cores, 64 chips, 1 core/chip CPU(s) orderable: 2 to 64 Parallel: No Primary Cache: 64KB(I)+64KB(D) on chip Secondary Cache: 1.75MB on chip per CPU L3 Cache: None Other Cache: None Memory: 256GB (64 * 10 * 512MB RIMMs, both controllers populated) Disk Subsystem: MFS 16GB (Memory File System) Other Hardware: None SOFTWARE -------- Operating System: Tru64 UNIX V5.1B (Rev. 2650) +PK3 Compiler: Compaq C V6.5-011-48C5K Program Analysis Tools V2.0 Spike V5.2 (509 DTK) Compaq Fortran X5.5-2602-48C8L Compaq Fortran 77 X5.5-2602-48C8L KAP Fortran V4.3 000607 KAP Fortran 77 V4.1 980926 KAP C V4.1 000607 File System: MFS 16GB System State: Multi-user NOTES ----- Baseline C: cc -arch ev7 -fast -O4 ONESTEP Fortran: f90 -arch ev7 -fast -O5 ONESTEP Peak: All use -v -g3 -arch ev7 -non_shared ONESTEP except these (which use only the tunings shown below): 173.applu 188.ammp 191.fma3d 200.sixtrack Individual benchmark tuning: 168.wupwise: kf77 -g3 -arch ev7 -non_shared ONESTEP -call_shared -inline all -tune ev67 -unroll 12 -automatic -align commons -arch ev67 -fkapargs=' -aggressive=c -fuse -fuselevel=1 -so=2 -r=1 -o=1 -interleave -ur=6 -ur2=060 ' +PFB 171.swim: same as base (f90...) 172.mgrid: kf90 -g3 -arch ev7 -non_shared ONESTEP -call_shared -arch generic -O5 -inline manual -nopipeline -transform_loops -unroll 9 -automatic -fkapargs='-aggressive=a -fuse -interleave -ur=2 -ur3=5 -cachesize=128,16000 ' +PFB 173.applu: kf90 ONESTEP -O5 -transform_loops -fkapargs=' -o=0 -nointerleave -ur=14 -ur2=260 -ur3=18' +PFB 177.mesa: kcc -g3 -arch ev7 -non_shared ONESTEP -fast -O4 +CFB +IFB 178.galgel: f90 -g3 -fixed -arch ev7 -non_shared ONESTEP -O5 -fast -unroll 5 -automatic 179.art: kcc -g3 -arch ev7 -non_shared ONESTEP -assume whole_program -ldensemalloc -call_shared -assume restricted_pointers -unroll 16 -inline none -ckapargs=' -fuse -fuselevel=1 -ur=3' +PFB 183.equake: cc -g3 -arch ev7 -non_shared ONESTEP -call_shared -arch generic -fast -O4 -ldensemalloc -assume restricted_pointers -inline speed -unroll 13 -xtaso_short +PFB 187.facerec: f90 -g3 -arch ev7 -non_shared ONESTEP -O4 -nopipeline -inline all -non_shared -speculate all -unroll 7 -automatic -assume accuracy_sensitive -math_library fast +IFB 188.ammp: cc ONESTEP -arch host -O4 -ifo -assume nomath_errno -assume trusted_short_alignment -fp_reorder -readonly_strings -ldensemalloc -xtaso_short -assume restricted_pointers -unroll 9 -inline speed +CFB +IFB +PFB 189.lucas: kf90 -g3 -arch ev7 -non_shared ONESTEP -O5 -fkapargs='-ur=1' +PFB 191.fma3d: kf90 -arch ev6 -non_shared ONESTEP -O4 -transform_loops -fkapargs='-cachesize=128,16000 ' +PFB 200.sixtrack: f90 -arch ev7 -non_shared ONESTEP -fast -O5 -assume accuracy_sensitive -notransform_loops +PFB 301.apsi: kf90 -g3 -arch ev7 -non_shared ONESTEP -O5 -inline none -call_shared -speculate all -align commons -fkapargs=' -aggressive=ab -tune=ev5 -fuse -ur=1 -ur2=60 -ur3=20 -cachesize=128,16000' Most benchmarks are built using one or more types of profile-driven feedback. The types used are designated by abbreviations in the notes: +CFB: Code generation is optimized by the compiler, using feedback from a training run. These commands are done before the first compile (in phase "fdo_pre0"): mkdir /tmp/pp rm -f /tmp/pp/${baseexe}* and these flags are added to the first and second compiles: PASS1_CFLAGS = -prof_gen_noopt -prof_dir /tmp/pp PASS2_CFLAGS = -prof_use -prof_dir /tmp/pp (Peak builds use /tmp/pp above; base builds use /tmp/pb.) +IFB: Icache usage is improved by the post-link-time optimizer Spike, using feedback from a training run. These commands are used (in phase "fdo_postN"): mv ${baseexe} oldexe spike oldexe -feedback oldexe -o ${baseexe} +PFB: Prefetches are improved by the post-link-time optimizer Spike, using feedback from a training run. These commands are used (in phase "fdo_post_makeN"): rm -f *Counts* mv ${baseexe} oldexe pixie -stats dstride oldexe 1>pixie.out 2>pixie.err mv oldexe.pixie ${baseexe} A training run is carried out (in phase "fdo_runN"), and then this command (in phase "fdo_postN"): spike oldexe -fb oldexe -stride_prefetch -o ${baseexe} When Spike is used for both Icache and Prefetch improvements, only one spike command is actually issued, with the Icache options followed by the Prefetch options. vm: vm_bigpg_enabled = 1 vm_bigpg_thresh = 6 vm_swap_eager = 0 ubc_maxpercent = 50 proc: max_per_proc_address_space = 34359738368 max_per_proc_data_size = 34359738368 max_per_proc_stack_size = 34359738368 max_proc_per_user = 2048 max_threads_per_user = 4096 maxusers = 2048 per_proc_address_space = 34359738368 per_proc_data_size = 34359738368 per_proc_stack_size = 34359738368 Portability: galgel: -fixed Information on UNIX V5.1B Patches can be found at http://ftp1.service.digital.com/public/unix/v5.1b/ Processes were bound to CPUs using "runon". ----------------------------------------------------------------------------- For questions about this result, please contact the tester. For other inquiries, please contact webmaster@spec.org. Copyright 1999-2004 Standard Performance Evaluation Corporation Generated on Wed Apr 13 13:14:59 2005 by SPEC CPU2000 ASCII formatter v2.1