Siemens AC65 Benutzerhandbuch

5.12 Performance

s

wm_java_usersguide_v12

Page 43 of 123

2008-02-25

Confidential / Released

The performance study was focused on comparable performance values under various circum-
stances.

This section gives information about the Java command execution throughput (“jPS”= Java
statements per second). The scope of this measurement is only the statement execution time,
not the execution delay (Java command on AT interface

 

→

 Java instruction execution 

→

 reac-

tion on GPIO).

Figure 13:  Test case for measuring Java command execution throughput

The following Java instruction was used for calculation of the typical jPS:
value = ( 2 x number of calculation statements ) / 

( (1 / frequencyB ) - ( 1 / frequencyA ) );

Measurement and calculation were done using:
•

duration of each loop = 600 s

•

number of calculation statements = 50 “result=(CONSTANT_VALUE/variable_value);”-
Instructions (executed twice per pin cycle)

•

frequencyA as measured with a universal counter

•

frequencyB as measured with a universal counter

The reference loop has the same structure as the measurement loop except that the measure-
ment sequence is moved.

Since only a small amount of Java code is executed in this test, it is easily optimized by the
CLDC’s HI just-in-time compiler. More complex applications might not reach that execution
speed.

ME in IDLE mode / Not connected

~49000

CSD connection

~46000