xwallman@yale.UUCP (11/21/87)
How they do it ? I don't know.
What's my guess, given the constraints of the hardware (and there isn't
much in there except some RAM, a 68000 and the SHIFTER) ?
Allright usually on my old and beloved 800 everyone used DLIs (now in ST-Talk
that's called HBLs) for multiple colors. An interrupt occured after (*) a
scanline. And one had the time to change color registers, in perfect synch
with the display.
But this really isn't neccessary.
You don't have to switch colors on scanlines. You can switch them anytime
you wan't. (True for ST and 800)
Problem is that you need to get the CPU somehow in sync, with the display.
This was impossible (?(*)) if you have a co-processor stealing cycles (as
ANTIC does) but the SHIFTER doesn't and this makes things easier for the ST.
From previous postings (I have never seen the actual product, much less a
picture) I read that only 199 lines can be shown with SPECTRUM 512.
My guess is that the first DLI (Ooops HBL) is used for starting
synchronization. Then the CPU will spring into action, kerchunking away,
changing registers on the fly. Here is where above mentiones RAM comes into
the picture. Displays with 300/200/9? Bits are BIG.
So I think this is a theory that would work in praxis, I would be very
surprised if SPECTRUM 512 wouldn't work that way. (Surprise me !)
(*) Not really after, somewhere in the middle of the scanline.
(**) Footnote : I managed once to get a somewhat appealing display
giving the ANTIC some cycles, but usually it's just
a mess.
Out-of-YALE-callers please use only these addresses
<<<<wallman-george@yale.uucp>>>>>
<.......................IIII, ..<
<......III, ..........IXI, .....<
<.....IXIIXI, ......IXI,, ......<
<....IXI,,IXI, ...IXI,, ........<
<...IXI,, .IXI, IXXI,, .........<
<..IXXI,, ..IXXXXXI,, ..........<
<.IXXI,, ....IXXXI,, ...........<
<IXXXI,, .....III,, ............<
<<<<<<<<<wallman@yalecs>>>>>>>>>
And grand new opening at WALLMANN@CTSTATEU.BITNET
(yale addresses preferred)c9c-eh@dorothy.Berkeley.EDU (Warner Young (WHY)) (11/21/87)
In article <19278@yale-celray.yale.UUCP> xwallman@gator.UUCP (Natuerlich!) writes: > >Problem is that you need to get the CPU somehow in sync, with the display. >This was impossible (?(*)) if you have a co-processor stealing cycles (as >ANTIC does) but the SHIFTER doesn't and this makes things easier for the ST. > >So I think this is a theory that would work in praxis, I would be very >surprised if SPECTRUM 512 wouldn't work that way. (Surprise me !) > I believe someone mentioned recently that they purchased the Spectrum 512 program? Could this same person please comment on whether there is any appreciable slowing of the machine? I've played with DEGAS Elite somewhat, and even that's a little slow at times. If Spectrum indeed has most of DEGAS's features, as the ad claims, then is it also as slow, or slower? \ / Disclaimer: I'm not associated \ /\ /arner with the latest revision \/ \__/ of SANITY. |oung \___| Last known address: c9c-eh@dorothy.Berkeley.EDU or ucbvax!dorothy!c9c-eh
wheels@mks.UUCP (11/24/87)
In article <5021@zen.berkeley.edu>, c9c-eh@dorothy.Berkeley.EDU (Warner Young (WHY)) writes: > I believe someone mentioned recently that they purchased the > Spectrum 512 program? Could this same person please comment > on whether there is any appreciable slowing of the machine? > I've played with DEGAS Elite somewhat, and even that's a little > slow at times. If Spectrum indeed has most of DEGAS's features, > as the ad claims, then is it also as slow, or slower? I wasn't the person who commented, but I just got Spectrum 512 and love it. All those colours! Some operations are slower than the other programs, but they are justified. For example, copying a block from the buffer to the screen is slower, but they have a lot of work to do to merge the pallettes of the two pictures. The manual says that about 80% of the cpu time is taken by the pallette switches. During some intensive operations, such as anti-aliasing, the program gives up the pallette switching to provide faster computation. During this time, the screen looks like garbage. Actual drawing speed is good. Rubber-band lines keep up with the mouse, no problem. I have no complaints. -- Gerry Wheeler Phone: (519)884-2251 Mortice Kern Systems Inc. UUCP: uunet!watmath!mks!wheels 43 Bridgeport Rd. E. BIX: mks Waterloo, Ontario N2J 2J4 CompuServe: 73260,1043