PICAXE instruction times (partial list)

[hungarian]

PICAXE "time" variable vs clock frequency

Symbol:	k31	k250	k500	m1	m2	m4	m8	m16	m32
Actual frequency:	31kHz	250kHz	500kHz	1MHz	2MHz	4MHz	8MHz	16HMz	32MHz
Time increments every:	untested	untested	33.3s	16.7s	2s	1s	2s	1s	0.5s
Assumed frequency:	untested	untested	15kHz	60kHz	4MHz	4MHz	16MHz	16MHz	16MHz

Under the instructions "enabletime" and "disabletime" the PICAXE PDF states that:

Effect of increased clock speed:
The time function will work correctly at 4MHz or 16 MHz.
At 2MHz or 8MHz the interval will be 2s.
At 16MHz the interval will be 0.5s
​

It is inconsistent about 16MHz. In fact the last statement is false.


Instuction tests


Instructions were timed using variations of the following code, using 10000, 20000, 40000 or 50000 for n:
w3=time
for w6=1 to n
next
w4=time
for w6=1 to n
instruction under test
next
w5=time


us=microseconds. Times are +/- 100us

Instruction Under Test (IUT)	IUT Time	Clock
b4=85	550us	4MHz
b4=b0+85	1050us	4MHz
b4=b0+b2	1050us	4MHz
b4=b0*13	1200us	4MHz
b4=b0/14	1300us	4MHz
w2=w0*251	1250us	4MHz
w2=w0+21845	1200us	4MHz
lookup b0,( ten byte constants ),b2	2350us	4MHz
lookup b0,( ten word constants ),w1	3600us	4MHz
goto	850us	4MHz
gosub {& return}	2350us	4MHz
if b0>85 then goto	FALSE	850us	4MHz
if b0>85 then goto	TRUE	1300us	4MHz
if b0>85 then gosub {& return}	FALSE	1300us	4MHz
if b0>85 then gosub {& return}	TRUE	4050us	4MHz
if b0>85 then: endif	FALSE	1300us	4MHz
if b0>85 then: endif	TRUE	850us	4MHz
if b0>85 then: else: endif	FALSE	1300us	4MHz
if b0>85 then: else: endif	TRUE	1700us	4MHz
readADC C.1,b4	800us	4MHz
pauseus 1	650us	4MHz
pauseus 1: pauseus 1	1300us	4MHz
pauseus 2	700us	4MHz
pauseus 4	750us	4MHz
pauseus 8	750us	4MHz
pauseus 10	800us	4MHz
pauseus 20	950us	4MHz
pauseus 40	1100us	4MHz
pauseus 80	1500us	4MHz
pauseus 100	13300us	500kHz
pauseus 100	6700us	1MHz
pauseus 100	3300us	2MHz
pauseus 100	1675us	4MHz
pauseus 100	850us	8MHz
pauseus 100	425us	16MHz
pauseus 100	220us	32MHz

Conclusions

For timing purposes (Note ¶ represents new line):

Source: if x then gosub label0
Equivalent: if not(x) then goto label1 ¶ gosub label0 ¶ label1:

Source: if x then ¶ <x-is-true> ¶ endif
Equivalent: if not(x) then goto label0 ¶ <x-is-true> ¶ label0:

Source: if x then ¶ <x-is-true> ¶ else ¶ <x-is-false> ¶ endif
Equivalent: if not(x) then goto label0 ¶ <x-is-true> ¶ goto label1 ¶ label0: ¶ <x-is-false> ¶ label1:

"pauseus n" is supposed to be a pause of about n*10 microseconds but the actual length is about (640+10*n)*4/f microseconds, where f is the clock frequency in MHz

Hope this is useful.

 

[hippy]

Before putting too much effort into determining instruction timings it may be worth reading other posts on the forum on the subject.

Because the PICAXE uses an interpreter and interprets variable-sized tokens, instruction timing depends on a number of factors including where the it is in memory. At best one can determine what the minimum and maximum instruction times may be and even that will be inaccurate because changing the code under test alters the timing of the code doing the timing. Using the 'time' variable will also bring its own inaccuracies.

Westaust55 has previously undertaken analysis of instruction timings and has documented his findings in the PDF files linked in this thread ...

http://www.picaxeforum.co.uk/showthread.php?17782-PICAXE-program-Size-Optimisations-and-Speed