Dr. McCalpin,
NASA Ames Research Center submits the following stream results for your
survey, the first TRIAD result to exceed 1 Terabyte/second. This run was
made on a 512 processor SGI Altix 3000 at NASA Ames, with 1.5GHz Itanium2
cpus. Note that the system is a single, coherent shared memory machine, and
the results are "standard" (OpenMP parallel). The run was done by Bron
Nelson (bron@sgi.com).
There were 2 changes made to the source code:
1) Due to an INTEL compiler problem, the statement:
COMMON a,b,c
was replaced by the three statements:
COMMON /common_a/ a
COMMON /common_b/ b
COMMON /common_c/ c
2) Because the generated numbers were so large (i.e. >1TB/s),
the output FORMAT statements were changed to accommodate the
larger values.
The code was compiled with the Intel 7.1.027 Fortran compiler, using
the options:
-i8 -extend_source -O3 -mP2OPT_hlo_pref_hint=3 -openmp
Regards,
Bob Ciotti
================================================================
Robert B. Ciotti TeraScale Applications Lead
NASA Advanced Supercomputing (NAS) Division TEL (650) 604-4408
NASA Ames Research Center FAX (650) 604-4377
Moffett Field, CA 94035-1000 ciotti@nas.nasa.gov
================================================================
----------------------------------------------
Double precision appears to have 16 digits of accuracy
Assuming 8 bytes per DOUBLE PRECISION word
----------------------------------------------
Array size = 5083693056
Offset = 5000
The total memory requirement is 116356 MB
You are running each test 100 times
--
The *best* time for each test is used
*EXCLUDING* the first and last iterations
----------------------------------------------------
Your clock granularity/precision appears to be 1 microseconds
----------------------------------------------------
Function Rate (MB/s) Avg time Min time Max time
Copy: 854061.5533 0.0954 0.0952 0.0958
Scale: 854438.0171 0.0953 0.0952 0.0957
Add: 1008593.7370 0.1218 0.1210 0.1876
Triad: 1007828.0053 0.1212 0.1211 0.1217
----------------------------------------------------
Solution Validates!
----------------------------------------------------
Received on Thu Nov 13 18:20:58 2003
This archive was generated by hypermail 2.1.8 : Sat Nov 15 2003 - 10:11:45 CST