Re: Historical latency values

From: Ron Nicholson (rhn@waltz.engr.sgi.com)
Date: Tue Aug 01 1995 - 17:33:09 CDT


Here are the cstream results for some Mac systems that I have at home.
The oldest machine is a 1989 vintage Mac.
The compilers used were from Metrowerks Codewarrior DR6, all optimisations on.
I wanted to try cstream on my 1984 vintage Mac 512KE, but I have to dig out
an older compiler and libraries that will run under System 6.0.x before
I can do this.

---

PowerMac 7100/80 (PPC601/80 256k 2nd level cache) System 7.5.0

sizeof(real) = 8 N = 523776 NTIMES = 10 Timing calibration ; time = 327.41 usec. Increase the size of the arrays if this is < 300 and your clock precision is =< 1/100 second. --------------------------------------------------- Function Rate (MB/s) RMS time Min time Max time Assignment: 48.128 174.516 173.786 175.383 Scaling : 48.116 174.272 173.831 174.604 Summing : 52.308 240.609 239.852 241.329 SAXPYing : 52.323 240.470 239.783 241.042

--

PowerMac 7100/80 emulating a 68020 Mac System 7.5.0

sizeof(real) = 10 NTIMES = 10 Timing calibration ; time = 1050.00 usec. Increase the size of the arrays if this is < 300 and your clock precision is =< 1/100 second. --------------------------------------------------- Function Rate (MB/s) RMS time Min time Max time Assignment: 8.944 591.725 583.333 600.000 Scaling : 3.598 1465.008 1450.000 1466.667 Summing : 4.891 1610.021 1600.000 1616.667 SAXPYing : 3.194 2456.680 2450.000 2466.667

-

Mac SE/30 68030/68882 16MHz System 7.0

sizeof(real) = 12 N = 64449 NTIMES = 10 Timing calibration ; time = 966.67 msec. Increase the size of the arrays if this is < 300 and your clock precision is =< 1/100 second. --------------------------------------------------- Function Rate (MB/s) RMS time Min time Max time Assignment: 2.442 638.379 633.333 650.000 Scaling : 1.450 1068.345 1066.667 1083.333 Summing : 1.698 1378.354 1366.667 1383.333 SAXPYing : 1.547 1508.356 1500.000 1516.667

-

Mac SE/30 68030/68882 16MHz System 7.0

sizeof(real) = 8 N = 64449 NTIMES = 10 Timing calibration ; time = 366.67 msec. Increase the size of the arrays if this is < 300 and your clock precision is =< 1/100 second. --------------------------------------------------- Function Rate (MB/s) RMS time Min time Max time Assignment: 4.125 250.000 250.000 250.000 Scaling : 1.263 823.475 816.667 866.667 Summing : 1.856 838.368 833.333 850.000 SAXPYing : 1.547 1001.679 1000.000 1016.667

-

Mac SE/30 68030/68882 16MHz System 7.0

sizeof(real) = 4 N = 130944 NTIMES = 10 Timing calibration ; time = 483.33 msec. Increase the size of the arrays if this is < 300 and your clock precision is =< 1/100 second. --------------------------------------------------- Function Rate (MB/s) RMS time Min time Max time Assignment: 3.308 316.667 316.667 316.667 Scaling : 0.706 1498.675 1483.333 1583.333 Summing : 1.122 1408.417 1400.000 1450.000 SAXPYing : 0.842 1868.340 1866.667 1883.333

-

PowerBook Duo 210 68030/no-fpu 16MHz System 7.1

sizeof(real) = 4 N = 130944 NTIMES = 2 Timing calibration ; time = 550.00 msec. Increase the size of the arrays if this is < 300 and your clock precision is =< 1/100 second. --------------------------------------------------- Function Rate (MB/s) RMS time Min time Max time Assignment: 2.857 366.667 366.667 366.667 Scaling : 0.050 20758.336 20750.000 20766.666 Summing : 0.055 28758.346 28733.334 28783.334 SAXPYing : 0.040 39408.336 39400.000 39416.668

-

PowerBook Duo 210 68030/no-fpu 16MHz System 7.1

sizeof(real) = 10 N = 64449 NTIMES = 2 Timing calibration ; time = 1016.67 msec. Increase the size of the arrays if this is < 300 and your clock precision is =< 1/100 second. --------------------------------------------------- Function Rate (MB/s) RMS time Min time Max time Assignment: 1.933 666.667 666.667 666.667 Scaling : 0.213 6050.000 6050.000 6050.000 Summing : 0.318 6091.672 6083.333 6100.000 SAXPYing : 0.167 11558.337 11550.000 11566.667

--- Ronald H. Nicholson, Jr. rhn@engr.sgi.com, rhn@netcom.com #include <canonical.disclaimer> // I speak only for myself, etc.



This archive was generated by hypermail 2b29 : Tue Apr 18 2000 - 05:23:05 CDT