about ULA emulation
ZXNet echo conference «zxnet.pc»
Hello, All
Many emulators use this approach to emulating a ULA video controller:
after the processor executes the next command, the number of clock cycles is taken,
spent on this command, and the corresponding amount of ULA drawing is performed.
for example: a certain PUSH took 11 clock cycles, therefore (in the case of a direct beam path)
ULA draws 22 points. and so on.
BUT - after all, if we have a machine with separate memory fields, ULA and Z80 access
to the data simultaneously, that is, a situation may arise when PUSH is put on
stack located in screen memory, high byte of word, meanwhile ULA
counted the byte at the previous address, and then PUSH put the low byte, but ULA
has already jumped forward, and thus only the most significant byte of the word appears on the screen,
which I put on the stack PUSH.
in the emulation algorithm mentioned above this will not be taken into account; both will appear on the screen
bytes, that is, strictly speaking, such emulation is incorrect.
Am I right?
Hello Titus
It’s high time to fully emulate not only the percentage “from the inside” and the spinning top, but also
bus (clock state). And all sorts of “device plugins” will be easier
screw. True, the emulsion will turn out to be slow... but that’s okay, the vintel
there's enough stupidity. ;)
Hello, boo_boo
Yes, you're right, this emulation is not entirely correct: v2_tong2:
Hello Lethargeek
This is of course good on a clockwise basis, and there’s enough nonsense (and with a good emulator architecture
maybe it will be better), but information on intimate matters of hardware within teams is needed
I'll have to look further...
Hello, boo_boo
in which ones is it - many? I have in the process of executing the command H ON EVERY CYCLE
a video beam is being built, for example. And I'm sure the same is done in everyone
emulators that work with ULA (and not with loose powder). Otherwise it’s no longer multicolor,
and partial multicolor.
There is a sea of information. Go to worldofspectrum, everything is there, and even updated.
Hello Lethargeek
Let> It’s high time to fully emulate not only percents “from the inside” yes
Let> spinning top, but also the bus (clockwise state). And all sorts of “device plugins”
Let> will be easier to screw. True, the emulsion will be slow... but
Let> it's okay, Vintel has enough stupidity. ;)
I'll try it and see how much dope the vintel has :rolleyes;
Nov>; Barwise this is of course good, and there’s enough stupidity (and with good
Nov> emulator architecture may be better), but here is information on intimate matters
Nov> you will still have to look for hardware inside the commands...
for what “matters within teams”?
Vla> in which of these - many? I am in the process of executing the command HA EACH
Vla> A video beam is built in a CYCLE, for example. And I'm sure the same is done in
Vla> all emulators that work with ULA (and not with loose powder). Otherwise it's already
Vla> is not multicolor, but partial multicolor.
Well, I didn’t look at yours, the sources are closed :rolleyes;
V; zemu, in the glitch it works up to the ULA opcode. In the US everything is kosher.
How did you implement this? ULA jerks hard at every beat, or as in
US -- after recording to video memory/port?
Hello, boo_boo
Intimacy is, for example, LDIR, conditional transitions and generally something more or less
"complex". The order in which registers are changed by these instructions varies across clones.
Z80, but on different Z80 manufacturing technologies. For accurate emulation, you need
it would be to choose a specific percentage.
Hello NovaStorm
Nov> Intimacy is, for example, LDIR, conditional jumps and something else
Nov> more or less “complex”. The order in which these instructions modify registers
Nov> varies on Z80 clones, and on Z80 manufacturing technology that differs.
Unproven, very unproven... :v2_wink2:
Hello Titus
Tit> Quote:
Tit> Message from NovaStorm
Tit> Intimacy is, for example, LDIR, conditional jumps and something else
Tit> more or less "complex". The order in which these instructions modify registers
Tit> varies on Z80 clones, and on Z80 manufacturing technology that differs.
Tit>
Tit>
Tit> Unproven, very unproven...
I ran zexall on several z80 emulators - the result was depressing. And he
Not the most complete testing is being carried out yet...
Hello NovaStorm
Nov> Intimacy is, for example, LDIR, conditional jumps and something else
Nov> more or less “complex”. The order in which these instructions modify registers
Nov> varies on Z80 clones, and on Z80 manufacturing technology that differs.
Nov> For accurate emulation, it would be necessary to select a specific processor.
this only concerns the notorious MEMPTR, but with him everything is already clear
(http://zx.pk.ru/showpost.php?p=44058&postcount=150). about different technologies
and strange clones are rumored, but specific specimens have not yet been identified.
if you have it, give it here :v2_smile;
Err>; I ran zexall on several z80 emulators - the result
Err> depressing. And he is not yet conducting the most complete testing...
of course. but, fortunately, with the hard Z80 everything is much more pleasant :) there is a small
difference (see link above) but in general it’s one hell of a deal :) so it’s quick
an indicator of the unfinished nature of some emulators