CCBASIC Geschwindigkeit

Kennt jemand die Geschwindigkeit von CCBASIC? Gibt's da eine Tabelle?

Ich suche nicht die Geschwindigkeit für jeden möglichen CCBAISC token unter allen denkbaren Umständen und ganz genau. Mir geht's mehr um einen Überblick.

Ich hab mal versucht, auf die Schnelle das auszutesten; aber jetzt ist's schon so spät und ich hab immer noch nicht alles drin, und ich hab's auch noch nicht ausgewert.

Ich füge hier mal mein Testprogramm ein (upload geht momentan bei mir nicht). Das Programm sollte auf jeder CC1 laufen; es braucht keinen S19-Treiber, sondern nur ein ASCII Terminal an der RS232.

Bin in einer Woche zurück -- vielleicht kann jemand was dazu sagen!

Code:

---

'-----------------------------------
' CCBASIC Performance Messungen
' V1.0, 2004-02-23 / JF
' Es werden keine S19 Treiber benötigt,
' die Ausgabe erfolgt über RS232 an ein ASCII Terminal
'-----------------------------------

define lbyte    byte[1]
define delta    word[4] ' delta timer
define base    word[5] ' timer for empty loop
define i        word[6] ' counter
define temp    word[7] ' GP
define j        word[8] ' GP
define k        word[9] ' GP

'########## start
    beep 368,10,0 : pause 50
    print "Start CCBASIC performance tests, 2004-02-23 / JF"
    print "All lines are <text>TAB<value>"
    print "All values are [ms] per 1000 iterations"

'########## get base: empty loop
    delta = timer
    for i = 1 to 1000
    next i
    delta = timer - delta
    base = delta
    print "empty loop:", delta * 20

'########## measurements

    gosub test_for
    gosub test_assignment
    gosub test_gosub
    gosub test_sys
    gosub test_if

'########## show end

    print "Done CCBASIC performance tests."
    beep 368,10,0 : pause 50
    beep 368,10,0 : pause 50
 end

'########## test cases
'########## test cases
'########## test cases

#test_if
    delta = timer : for i = 1 to 1000
          if i = 0 then temp = i
    next i
    delta = timer - delta - base : print "if i = 0 then temp = i", delta * 20

    delta = timer : for i = 1 to 1000
          if i <> 0 then temp = i
    next i
    delta = timer - delta - base : print "if i <> 0 then temp = i", delta * 20

    delta = timer : for i = 1 to 1000
          if i = 0 then temp = i else temp = i
    next i
    delta = timer - delta - base : print "if i = 0 then temp = i else temp = i", delta * 20

    delta = timer : for i = 1 to 1000
          temp = i : if i = 0 then temp = i
    next i
    delta = timer - delta - base : print "temp = i : if i = 0 then temp = i", delta * 20

    delta = timer : for i = 1 to 1000
          temp = i : if i <> 0 then temp = i
    next i
    delta = timer - delta - base : print "temp = i : if i <> 0 then temp = i", delta * 20

    delta = timer : for i = 1 to 1000
          if i <> 0 then temp = i else temp = i
    next i
    delta = timer - delta - base : print "if i <> 0 then temp = i else temp = i", delta * 20

    delta = timer : for i = 1 to 1000
          if i = 0 then goto goto_never
    next i
    delta = timer - delta - base : print "if i = 0 then goto goto_never", delta * 20

    delta = timer : for i = 1 to 1000
          if i <> 0 then goto goto_1
          #goto_1
    next i
    delta = timer - delta - base : print "if i <> 0 then goto goto_1", delta * 20

    delta = timer : for i = 1 to 1000
          if i = 0 then goto goto_never else goto goto_2
          #goto_2
    next i
    delta = timer - delta - base : print "if i = 0 then goto goto_never else goto goto_2", delta * 20

    delta = timer : for i = 1 to 1000
          if i <> 0 then goto goto_3 else goto goto_never
          #goto_3
    next i
    delta = timer - delta - base : print "if i <> 0 then goto goto_3 else goto goto_never", delta * 20
return
#goto_never
    print "reached goto_never!!"
    beep 368,10,0 : pause 50
    beep 368,10,0 : pause 50
    beep 368,10,0 : pause 50
    beep 368,10,0 : pause 50
end


#test_sys
    delta = timer : for i = 1 to 1000
          sys &h91 &h819d
    next i
    delta = timer - delta - base : print "sys rts_push1", delta * 20

    delta = timer : for i = 1 to 1000
          sys &h91 &h0, &h819d
    next i
    delta = timer - delta - base : print "sys rts_push2", delta * 20

    delta = timer : for i = 1 to 1000
          sys &h91 &h0, &h0, &h819d
    next i
    delta = timer - delta - base : print "sys rts_push3", delta * 20

    delta = timer : for i = 1 to 1000
          sys &h91 &h0, &h0, &h0, &h819d
    next i
    delta = timer - delta - base : print "sys rts_push4", delta * 20

    delta = timer : for i = 1 to 1000
          sys &h91 &h0, &h0, &h0, &h0, &h819d
    next i
    delta = timer - delta - base : print "sys rts_push5", delta * 20

    delta = timer : for i = 1 to 1000
          sys &h91 &h0, &h0, &h0, &h0, &h0, &h819d
    next i
    delta = timer - delta - base : print "sys rts_push6", delta * 20

    delta = timer : for i = 1 to 1000
          sys &h91 &h0, &h0, &h0, &h0, &h0, &h0, &h819d
    next i
    delta = timer - delta - base : print "sys rts_push7", delta * 20

    delta = timer : for i = 1 to 1000
          temp = sys_rts_push7
    next i
    delta = timer - delta - base : print "sys_rts_push7", delta * 20
return
#sys_rts_push7
    sys &h91 &h0, &h0, &h0, &h0, &h0, &h0, &h819d
return

#test_gosub
    delta = timer : for i = 1 to 1000
          gosub gosub_empty
    next i
    delta = timer - delta - base : print "gosub gosub_empty", delta * 20

    delta = timer : for i = 1 to 1000
          temp = gosub_empty_return
    next i
    delta = timer - delta - base : print "temp = gosub_empty_return", delta * 20

    delta = timer : for i = 1 to 1000
          gosub gosub_word_word
    next i
    delta = timer - delta - base : print "gosub gosub_word_word", delta * 20

    delta = timer : for i = 1 to 1000
          temp = gosub_word_word_return
    next i
    delta = timer - delta - base : print "temp = gosub_word_word_return", delta * 20
return
#gosub_empty
return
#gosub_empty_return
return 123
#gosub_word_word
    temp = temp
return
#gosub_word_word_return
    temp = temp
return temp

#test_assignment
    delta = timer : for i = 1 to 1000
          temp = temp
    next i
    delta = timer - delta - base : print "word = word", delta * 20

    delta = timer : for i = 1 to 1000
          lbyte = lbyte
    next i
    delta = timer - delta - base : print "byte = byte", delta * 20

    delta = timer : for i = 1 to 1000
          temp = lbyte
    next i
    delta = timer - delta - base : print "word = byte", delta * 20

    delta = timer : for i = 1 to 1000
          lbyte = temp
    next i
    delta = timer - delta - base : print "byte = word", delta * 20
return

#test_for
    ' test empty loops running from 1..10
    for k = 1 to 10
        delta = timer
        for i = 1 to 1000
            for j = 1 to k
            next j
        next i
        delta = timer - delta - base : print "for j = 1 to "; k; " : next j", delta * 20
    next k
return

---

Anscheinend geht die "Länge" der Basic-Befehle ziemlich stark in die Ausführungszeit ein. Das verwundert nicht: jedes Byte eines jeden Basic tokens muss ja einzeln aus dem EEPROM gelesen werden, bevor das token ausgeführt werden kann.

Ich habe deshalb das Programm so erweitert, dass auch die Länge der getesteten Basic Anweisung(en) in einem Unterprogramm gemessen und mit ausgegeben wird:

Code:

---

#getCodeAddress
    ' reversed: lda _stack2 / sta $91 / lda _stack2 + 1 / sta $92 / rts, nop
    sys &h91 &h819d, &hb792, &hb6c3, &hb791, &hb6c2
    return

---

Die Messungen habe ich entsprechend erweitert, und auch die GOTOs explizit gemessen:

Code:

---

#test_goto
    delta = timer : codes = getCodeAddress
    for i = 1 to 1000
          goto label_11
          #label_11
    next i
    delta = timer - delta - base : codes = getCodeAddress - codes - baseCodes
    print "goto label_11", delta * 20, codes

---

und so weiter.

Ergebnis bisher (4 MHz):

1. Nächstes Byte aus EEPROM lesen und ausführen: ca. 160 [us]
2. Sonderfall GOTO: ca. 600 + 3 * 160 = 1050 [us]
(Token ist zwar nur 3 Bytes lang, ändert aber EEPROM Adresse)
3. Sonderfall leeres GOSUB / RETURN: overhead ca. 1900 [us]
(2 Tokens = 4 Bytes, aber 2 * EEPROM Adresse ändern)
4. Sonderfall leeres FOR / NEXT: overhead ca. 2500 [us]
(dynamisch 14 bytes pro Durchlauf, mit 1 * EEPROM Adresse ändern)

Das heisst:

Eine 4 MHz CC1 macht im Schnitt so etwa 1000 bis maximal 2000 Basic-Befehle pro Sekunde.

Kann das jemand bestätigen?

So Tief bin ich in dem Thema nicht drin, aber die Schaltung eines Ausgangs mit Toggelfunktion am Oszilloskop ergibt Pulse der Dauer um 1ms.
Manfred