Stream results 386, 5x86

From: Wei Lin (wlin@NetergyNet.COM)
Date: Sun Jul 09 2000 - 22:44:22 CDT

  • Next message: Wesley Jones: "SGI Origin 3800, 400 MHz MIPS R12K: 128 CPU STREAM Result"

    Some results that may be of historical interest. A 1992 386 machine and
    a 1995 5x86 (high-end 486 compatible) machine. A special clock driver
    was used to get 10 millisecond resolution under DOS (normally it has 55
    ms resolution). Driver included for reference.

    Evergreen 586 = AMD 5x86, 16KB L1 cache, 120 MHz (3x40)
    ALI Vi15G motherboard with SiS 85C471 chipset, 256KB L2 cache
    64 MB Fast Page Mode DRAM, 60ns
    DOS 6.22, QEMM 8.01, Causeway DOS Extender
    Watcom Fortran 11.0, wfl386 -Ox -4 -fp3 -l=Causeway

    ----------------------------------------------
     Double precision appears to have 16 digits of accuracy
     Assuming 8 bytes per DOUBLE PRECISION word
    ----------------------------------------------
     Array size = 2345678
     Offset = 0
     The total memory requirement is 53 MB
     You are running each test 10 times
     The *best* time for each test is used
    ----------------------------------------------------
     Your clock granularity/precision appears to be 10000 microseconds
    The tests below will each take a time on the order
    of 1040000 microseconds
       (= 104 clock ticks)
    Increase the size of the arrays if this shows that
    you are not getting at least 20 clock ticks per test.
    ----------------------------------------------------
    WARNING -- The above is only a rough guideline.
    For best results, please be sure you know the
    precision of your system timer.
    ----------------------------------------------------
    Function Rate (MB/s) RMS time Min time Max time
    Copy: 50.7174 0.7450 0.7400 0.7500
    Scale: 45.2179 0.8443 0.8300 0.8900
    Add: 47.7087 1.1890 1.1800 1.1900
    Triad: 44.3278 1.2771 1.2700 1.3100
    Sum of a is = 2.7052722699857500D+018
    Sum of b is = 5.4105445396884370D+017
    Sum of c is = 7.2140593867314100D+017

    <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<

    AMD 386, 40 MHz + IIT 3C87 math coprocessor, 40 MHz
    Noname motherboard, Forex chipset FRX36C100/FRX36C200, 64KB cache
    8 MB DRAM, 70ns
    DOS 5.0, QEMM 7.04, Causeway DOS Extender
    Watcom Fortran 11.0, wfl386 -Ox -3 -fp3 -l=Causeway

    ----------------------------------------------
     Double precision appears to have 16 digits of accuracy
     Assuming 8 bytes per DOUBLE PRECISION word
    ----------------------------------------------
     Array size = 234567
     Offset = 0
     The total memory requirement is 5 MB
     You are running each test 10 times
     The *best* time for each test is used
    ----------------------------------------------------
     Your clock granularity/precision appears to be 10000 microseconds
    The tests below will each take a time on the order
    of 700000 microseconds
       (= 70 clock ticks)
    Increase the size of the arrays if this shows that
    you are not getting at least 20 clock ticks per test.
    ----------------------------------------------------
    WARNING -- The above is only a rough guideline.
    For best results, please be sure you know the
    precision of your system timer.
    ----------------------------------------------------
    Function Rate (MB/s) RMS time Min time Max time
    Copy: 7.3590 0.5140 0.5100 0.5200
    Scale: 5.6865 0.6610 0.6600 0.6700
    Add: 7.7118 0.7370 0.7300 0.7400
    Triad: 6.3973 0.8870 0.8800 0.8900
    Sum of a is = 2.7052630435503090D+017
    Sum of b is = 5.4105260870780820D+016
    Sum of c is = 7.2140347828127140D+016




    This archive was generated by hypermail 2b29 : Mon Jul 17 2000 - 04:46:15 CDT