Is a MSX2 machine slower than a MSX1 one?

Page 3/3
1 | 2 |

By jltursan

Prophet (2180)

jltursan's picture

18-10-2019, 23:35

Just as a curiosity, this is the new performance of the NMS-8245 after aplying the following patch (left, patched hook):

The code (very limited now):

htimi:	ld hl,REPCNT
	ld a,(hl)
	dec a
	jr z,.incr
	cp 1
	jr z,.end
	ret
.incr:  ld a,2
.end:	ld (hl),a
	ret

By Grauw

Ascended (8455)

Grauw's picture

18-10-2019, 23:48

Even just tweaking the key repeat counter makes such a difference? I thought it would only make a 5 scanline or so difference… (SCNCNT being the bigger factor.)

By akumajo

Resident (34)

akumajo's picture

19-10-2019, 10:47

jltursan wrote:
htimi:	ld hl,REPCNT
	ld a,(hl)
	dec a
	jr z,.incr
	cp 1
	jr z,.end
	ret
.incr:  ld a,2
.end:	ld (hl),a
	ret

dec a vs. cp 1
dec a is a smaller and faster way to check if a equal 1 Wink

By Grauw

Ascended (8455)

Grauw's picture

19-10-2019, 12:26

It writes a to (hl) after the compare so using dec a is not functionally identical.

By akumajo

Resident (34)

akumajo's picture

19-10-2019, 13:06

You're right, and the gain is too insignificant if we do "inc a" after, forget what I said oO

By jltursan

Prophet (2180)

jltursan's picture

19-10-2019, 13:20

Quite shocking indeed, between the above ISR installed or not there're slightly more than a dozen scanlines. I've double-checked this and that's what i'm getting.

EDIT: In fact, disabling all OpenMSX effects, I can count more than twice the above...

By Manuel

Ascended (15754)

Manuel's picture

19-10-2019, 22:57

jltursan, what do you mean exactly? openMSX effects have no influence on the timing of the running MSX programs of course...

By jltursan

Prophet (2180)

jltursan's picture

19-10-2019, 23:55

I mean that in my previous screenshots, the scanlines and blur make difficult to manually count the scanlines difference. With these effects disabled, is easier to spot a bigger difference in speed. Of course it has nothing to do in how the program runs Smile

Page 3/3
1 | 2 |