mflop
play

Mflop 300 200 100 0 6 8 10 12 14 16 18 20 22 Vector - PDF document

Mar 05, 2024 •273 likes •658 views

Performance of scalar product benchmark 600 R10000/195MHz R8000/75MHz Alpha/500Mhz IBM Power 2/66Mhz 500 IBM PPC604e/166Mhz IBM PPC604e/233Mhz PPro 200/Mhz, NAG F90 PPro 200/Mhz, PG F77 400 Mflop 300 200 100 0 6 8 10 12 14 16

  15. Performance of scalar product benchmark 600 R10000/195MHz R8000/75MHz Alpha/500Mhz IBM Power 2/66Mhz 500 IBM PPC604e/166Mhz IBM PPC604e/233Mhz PPro 200/Mhz, NAG F90 PPro 200/Mhz, PG F77 400 Mflop 300 200 100 0 6 8 10 12 14 16 18 20 22 Vector length 2^n

  16. 600 "sk2.a500" "sk1.a500" "sk2.R10000" 500 "sk1.R10000" 400 300 200 100 0 6 8 10 12 14 16 18 20 22

  17. 120 "sk3.a500" "sk3.r10" "sk3.rs6" 100 "sk3.r8" 80 60 40 20 0 6 8 10 12 14 16 18 20 22

  21. CPU 32 Registers 1000 W. Lev. 1 Cache 12000 W. Level 2 Cache 0.5 MW ext. Level 3 Cache 64 MW Main Memory 1 GW Disk Space

  23. 180 red black(1) red black(2) red black(4) 160 140 120 100 Mflop 80 60 40 20 0 4 16 64 256 1024 grid size

  24. 180 red black(4) fused (1, 1) fused (2, 2) 160 fused (4, 0) 140 120 100 Mflop 80 60 40 20 0 4 16 64 256 1024 grid size

  25. 180 optimised RB melt(2, 2) melt(3, 3) 160 melt(4, 4) 140 120 100 Mflop 80 60 40 20 0 4 16 64 256 1024 grid size

More recommend