frcX vs. mvmeX on performance

[Johnny Tang <tang@cebaf.gov>]

This was the results of Motorola VME card with EPICS benchmark I posted in june:

        mv167(33MHZ)              mv177(50MHZ)              mv177(50MHZ)
                                  with 68040 epics          with 68060 epics*
       ---------------------      ---------------------     ---------------------
       scanPeriod    CPU IDL      scanPeriod    CPU IDL     scanPeriod    CPU IDL
       CPU usage                  CPU usage                 CPU usage

1  sec     2%          90%            1%           90%          1%          90%
.5 sec     6%          84%            2%           89%          2%          90%
.2 sec    14%          56%            7%           79%          7%          80%
.1 sec    29%          35%           13%           66%         13%          66%
        (event lost)
average on 1 - IDL     33.75%                      19%

In comparison with Noboru's measurements for Force CPU:

mv167(33mHz):mv177(50mHz) = 1:2.56  (average on scanPeriod CPU % usage measure)
                            1:1.78  (average on 1-IDL CPU % usage measure)

Noboru's scanPeriod measure:
>  frc40:frc64:frc604RT=1:1.3:4.1
>
Marty Kraimer's (1-IDL) measure:
>  frc40:frc64:frc604RT=1:1.2:5.25


I will report the measurements on mvme1604(100mHz) shortly.

Johnny Tang

--
_____________________________________________

Johnny Tang     

Tel: (757)269-7239  
Fax: (757)269-7049
E-Mail: tang@cebaf.gov

Thomas Jefferson National Accelerator Facility
(Formerly CEBAF)
Accelerator Software Control
12000 Jefferson Avenue, MS 85A
Newport News, Virginia 23606
_____________________________________________