attached are Stream results for a Superdome with:
16 cells (in a single partition)
64 cpus (fully populated)
256GB of memory (fully populated)
552MHz PA8600 cpus, 500MHz system frequency
running HP-UX 11i
compiled with fortran90 in 64 bit mode as follows:
f90 -o stream_d.mp +extend_source +autodbl4 +DA2.0W +noppu +DS2.0 +O3 -Wl,+pd,L -Wl,-aarchive +Oparallel strea
m_d.f second_wall.o
and source modified to adjust array size and put arrays in COMMON:
for 64 cpu runs:
63c63
< PARAMETER (n=2000000,offset=0,ndim=n+offset,ntimes=10)
--- > PARAMETER (n=53477816,offset=0,ndim=n+offset,ntimes=10) 88c88 < * COMMON a,b,c --- > COMMON a,b,c---------------------------------------------- Double precision appears to have 16 digits of accuracy Assuming 8 bytes per DOUBLE PRECISION word ---------------------------------------------- Array size = 53477816 Offset = 0 The total memory requirement is 1224 MB You are running each test 10 times -- The *best* time for each test is used *EXCLUDING* the first and last iterations ---------------------------------------------------- Your clock granularity/precision appears to be 2 microseconds ---------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 25230.6682 0.0345 0.0339 0.0359 Scale: 21163.1446 0.0409 0.0404 0.0414 Add: 26102.0771 0.0494 0.0492 0.0497 Triad: 26717.1565 0.0486 0.0480 0.0500 ---------------------------------------------------- Solution Validates! ----------------------------------------------------
This archive was generated by hypermail 2b29 : Mon May 21 2001 - 08:38:48 CDT