This set includes only 32-bit results --- i.e. results based on either 32-bit data elements or 32-bit arithmetic. Many of the 64-bit results actually move data in 32-bit chunks, but as long as the operands and operations are 64-bit, those cases remain in the other tables.
Revised to Sun May 12 12:45:56 EDT 1996 -------------------------------------------------- -------------------------------------------------- -------------------------------------------------- Alliant_VFX80 3 2.1 23.7 Balance is Peak MFLOPS divided by Triad Bandwidth in MW/s Convex_C1XP 1 5.4 2.5 Convex_C3210 1 9.1 1.8 Convex_C3220 2 17.4 1.9 Convex_C3410 1 12.9 5.2 Convex_C3420 2 21.7 6.2 DEC_4000-710 1 6.6 19.3 DEC_5000-200 1 1.5 4.4 FPS_510_EA_scalar 1 1.0 22.4 FPS_511_EA_vector 1 7.8 2.8 FPS_511_SPARC_scalar 1 2.2 20.2 FPS_511_SPARC_vector 1 7.9 5.6 FPS_MCP104 1 7.8 3.4 FPS_MCP728 7 32.9 5.7 HP_9000-720 1 4.3 7.7 HP_9000-730 1 3.3 13.2 IBM_3090J-VF 1 14.5 6.1 IBM_3090J_scalar 1 5.7 15.4 IBM_RS-6000-250 1 4.4 9.9 IBM_RS6000-320 1 3.3 8.0 IBM_RS6000-320H 1 4.5 7.4 IBM_RS6000-355 1 7.1 7.8 IBM_RS6000-530 1 5.0 6.7 IBM_RS6000-580 1 13.9 6.0 IBM_RS6000-950 1 8.3 6.7 Machine ID ncpus MFLOPS Balance NEC_SX3-12 1 416.2 2.2 NeXTStation_68040 1 0.7 9.5 Omron_Luna88k 1 1.1 15.0 SGI_4D-240 1 1.3 5.1 SGI_4D-25 1 0.8 6.7 SGI_4D-35 1 2.1 4.6 SGI_Challenge_100 1 4.0 16.7 SGI_Challenge_150 1 5.3 19.0 SGI_Crimson 1 4.4 7.7 SGI_Indigo 1 2.2 4.0 STREAM Machine Balance --- John D. McCalpin, mccalpin@udel.edu Sun_4-490 1 1.3 16.3 Sun_670 1 2.0 13.3 Sun_SS1- 1 0.7 19.3 Sun_SS10-30 1 2.8 8.4 Sun_SS10-41 1 3.2 10.6 Sun_SS2_(4-75) 1 1.7 15.5 TMC_CM-5E 32 669.7 5.1 VAX_9000-420 1 6.5 6.4