< previous page	page_116	next page >
Page 116 Table 2.7 Dynamic instruction count relative to HLL[136]: (a) simple compilation; (b) optimized compilation. Arch. Instruction Set Result (relative to an HLL operation) HLL HLL 1.00 Stack P-code Pascal 5.46 R/M S/370 Fortran Hopt1 3.10 R+M VAX UNIX Fortran 1.99 (a) Arch Instruction Set Result (relative to an HLL operation) L/S PA RISC Fortran 1.98 R/M S/370 Hopt3 Fortran H opt 3 1.78 R+M VAX/VMS VMS Fortran 1.14 (b) Table 2.8 Instruction traffic (per HLL operation) [136]: (a) Instruction activities (per HLL operation)optimized compilation; (b) Dynamic instruction bandwidthoptimized (average) I-bandwidth relative to instruction count. Architecture L/S PA RISC R/M> S/370 R+M VAX Instructions Fetched 1.98 1.78 1.14 Instruction Bytes Fetched 7.92 6.3 - (a) Bytes/I inst/HLL op I-bw rel. to HLL op count R/M S/370 Hopt3 3.54 1.78 6.30 L/S PS RISC 4.0 1.98 7.92 (b) Table 2.9 Memory object reads/writes per HLL read/write [136]. (a) Simple Reads Writes R/M S/370 Fortran Hopt1 0.53 0.89 R+M VAX UNIX Fortran 1.11 1.66 (b) Optimized Reads Writes L/S PA RISC Fortran 0.29 0.83 R/M S/370 Fortran Hopt3 0.38 0.84 R+M VAX/VMS Fortran 0.32 0.74
< previous page	page_116	next page >