From: groepaz (groepaz_at_gmx.net)
Date: 2001-10-03 15:14:58
Hello Ullrich,
mmmmh, i was just playing around with that plasma thingy again....
a) i unrolled the whole thing the same way i would do in a "real"
demo.... the compiler (2.6.2) generates code like
lda _xbuf+17
jsr pusha0
lda _ybuf+8
jsr tosaddax
sta $E151
for each pixel, which is surprisingly damn close to the "perfect" code
which would look like
lda _xbuf+17
clc
adc _ybuf+8
sta $E151
.... by that optimization, the framerate was more than doubled,
resulting in 6.7 fps (with no vertical-blank waiting)
depending on what kind of optimization the current
development-compiler adds... i'd say it's entire possible it could
spit out the "perfekt" code for this ,=) (resulting in >50 fps
probably)
b) i noticed that the compiler i am using (win32 2.6.2 package) has a
bug with (static) initialized variables (the frame-counter in your revised
plasma code wont work here).... so first thing i did was checking the
compiler-version and SURPRISE... the compiler from the 2.6.2 package
prints 2.6.1 as version number ?!? is that a mistake in the package or
just a typo ? :o)
....whatever.... updated snapshot of my stuff coming this weekend
probably (added floating-point routines, some integer math, more
sample code)
oh another thing..... is the linker smart enough to just link the
routines from a .lib file that are needed? or does it work simelar to
the gnu-linker? (links all routines at once that have been in one .o
-or .c/.s file respectivly- before)
i'd just like to write my libraries in a way they can be used most
efficient ,=)
--
Best regards,
groepaz mailto:groepaz_at_gmx.net
----------------------------------------------------------------------
To unsubscribe from the list send mail to majordomo_at_musoftware.de with
the string "unsubscribe cc65" in the body(!) of the mail.
This archive was generated by hypermail 2.1.3 : 2001-12-14 22:05:42 CET