jiffies to seconds

[danwinslow]

Anyone have a ML routine to convert jiffies ( from the RTC at 18/19/20 ) into seconds handy?

[Chilly Willy]

Instead of dividing the jiffies by 50/60, perhaps it would be better to add your own VB handler and increment a seconds variable every 50/60 ints.

[danwinslow]

Perhaps it would, but that's not the question

[Rybags]

Yep, standard practice is to just use a VBI that does it's own counting to 50 or 60.

 

Converting RTCLOCK to seconds would require dividing a 24-bit unsigned binary by 50 or 60.

 

Maybe hit up 6502.org, they should have some code that does binary division. You'd probably need to modify a 16-bit divide to work on 24-bit numbers.

[flashjazzcat]

This is from my IDEDIAG program. It uses the FP package to work out elapsed time and put the result in LBUFF:

 

get_elapsed ; figure out elapsed time
lda rtclok+2
sec
sbc start_time+2
sta elapsed
lda rtclok+1
sbc start_time+1
sta elapsed+1
lda rtclok
sbc start_time
sta elapsed+2
;
lda secdiv ; comvert secdiv to FP
sta fr0
lda #0
sta fr0+1
jsr ifp
ldx float_p
ldy float_p+1
jsr flstor ; save seconds multiplier

lda elapsed ; convert elapsed to FP
sta fr0
lda elapsed+1
sta fr0+1
jsr ifp
jsr fmove ; move jiffies to fr1
lda test_kb
sta fr0
lda #0
sta fr0+1 ; 1024 in fr0
jsr ifp ; convert to FP
jsr fdiv ; divide 1024 by elapsed jiffies

ldx float_p
ldy float_p+1
jsr fld1r ; put multiplier in fr1
jsr fmul ; multiply, put result in fr0
jsr fasc

 

Before starting your benchmark or whatever, just copy the RTC registers into START_TIME. SECDIV is derived thus:

 

get_secdiv
lda vcount
beq get_secdiv
wait2
tay
lda vcount
bne wait2; VCount = zero, but we've saved the largest possible in Y
cpy #131 ; NTSC will never get this high
bcc ntsc
lda #50
sta secdiv
rts
ntsc
lda #60
sta secdiv
rts

 

It's developmental code and probably buggy, and I don't know if it will suit your purposes.