In article <C5sx2o.Mow@news.udel.edu> you write:
<
<Anybody have a CM-5 handy to test? (vector units preferred but not essential)
I tried the benchmark but was unable to get accurate results. Even
with the largest possible memory size I was unable to get the time
below the granularity of the system block (60 ticks/sec). Also, note
that three other users were also running jobs at the same.
Note that the CM-5 has 64 bit wide memory, like a Cray. 32 bit memory
accesses require a read-modify-write sequence which takes three times longer.
Also note that all optimizations were turned off, because otherwise the
whole benchmark would have been deleted by the optimizer.
HW: CM-5, 256 SPARCs, 1024 Vector Units, 32 Mhz Clock
SW: CMOST 7.2 Beta 1 Patch 4, CMF 2.0 Beta 1, cmf -g [no optimization]
For stream_s, n=128 million. For stream_d, n=256 million.
Script started on Tue Apr 20 17:20:46 1993
e4 2% ./stream_s
-------------------------------------
Single precision appears to have 7 digits of accuracy
Assuming 4 bytes per default REAL word
-------------------------------------
Timing calibration ; time = 21.66667 hundredths of a second
Increase the size of the arrays if this is <30 and your clock precision is =<1
/100 second
---------------------------------------------------
unction Rate (MB/s) RMS time Min time Max time
ssignment:********** 0.1000 0.1000 0.1000
caling :********** 0.0953 0.0833 0.1000
umming :-9203.5098 0.1202 0.1167 0.1333
AXPYing :********** 0.1138 0.0833 0.1167
e4 3% ./stream_d
-------------------------------------
Double precision appears to have 16 digits of accuracy
Assuming 8 bytes per DOUBLEPRECISION word
-------------------------------------
Timing calibration ; time = 3.333333134651184 hundredths of a second
Increase the size of the arrays if this is <30 and your clock precision is =<1
/100 second
---------------------------------------------------
unction Rate (MB/s) RMS time Min time Max time
ssignment:********** 0.0264 0.0167 0.0333
caling :********** 0.0279 0.0167 0.0333
umming :********** 0.0441 0.0333 0.0500
AXPYing :********** 0.0425 0.0333 0.0500
e4 4% exit
Script ended on Tue Apr 20 17:21:06 1993
Note that 0.0167s is the granularity of the system clock.
This archive was generated by hypermail 2b29 : Tue Apr 18 2000 - 05:23:02 CDT