IP networking Interfaces User's Guide

TC65 JAVA User's Guide

Strictly confidential / Released

TC65 JAVA User's Guide_V05 Page 29 of 90 26.09.2005

5.9 Performance

The performance study was focused on comparable performance values under various

circumstances.

5.9.1 Java

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 Î Java instruction execution

Î reaction on GPIO).

Generated frequency A

[reference sequence]

Test MIDlet

Output pin

Change

output

pin level

Measurement loop (for frequency B)

Send URC (start performance test) ATCommandListener.ATEvent()

Generated frequency B

[measurement sequence]

ATCommand.send(...)

Measurement sequence

Reference loop (for frequency A)

ATCommand.send(...)

Figure 11: 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

measurement sequence is moved.

State jPS-Value (mean)

TC65 module in IDLE mode / Not connected ~49000

CSD connection ~46000

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.