Home Page Forums MAME MAME 8bit Apples - Apple I, II, /// and the 16 bit GS

Thanks, that's working better :-). Now to see if i can track down my software issue..

So I thought I'd give it another go to try the bgfx ntsc monochrome decoding under ubuntu and it just makes the screen go black.

mono screen before enabling ntsc


getting ready to enable NTSC:


and screen going to black:


Am I doing it wrong?

Not a big deal, just gotta try all the features!

BTW it's cool now the //e has a "zipchip 4mhz" machine configuration option 8-) You can really tell when the zipchip is active because the cursor really blinks fast.

Your GPU or drivers may have compatibility issues. Different -bgfx_backend APIs may help, you should continue to see the B&W screen when you initially turn on NTSC.

Once that's resolved, you also have to set NTSC A and B to 0.0 and Scanline Duration to 39.1.

The accelerator options are primarily for Total Replay, which enables them when it's decompressing things and then turns back to original speed to play the games. But they're useful for other things too.

I was afraid that it may be my GPU or drivers.

I'm using integrated graphics on a laptop, and aside from the NTSC thing, it all works remarkably well with mame, even when using a usb DisplayLink dock to get 5 total monitors (2 on the dock and 3 off the laptop (counting the laptop screen). I wonder if the latest Ubuntu has better drivers, I'm afraid to upgrade since everything's been running so well.

00:02.0 VGA compatible controller: Intel Corporation 4th Gen Core Processor Integrated Graphics Controller (rev 06)
Model name: Intel(R) Core(TM) i7-4700MQ CPU @ 2.40GHz


After seeing the 4 mhz zipchip I thought, maybe that could speed up Instant Pascal, as I remember that being slow back in the day. (Still slow, even at 4mhz 8-) Such a pretty GUI interface though.

I wonder if Instant Pascal uses the same "MGTK" (Mouse Graphics Tool Kit) Apple DHR GUI toolkit that Apple II DeskTop and a few other things did. It's got a similar look.

Yeah, no, it really shouldn't be a GPU driver issue. I'll see what I can do.

I can confirm that -bgfx_backend opengl reproduces this issue on Windows as well.

Another thing for the Apple III emulation that I have encountered a bit lately is the fonts don't get loaded well / get corrupted when testing with my problock3 driver. On the real machine i don't have any issue.

I had a bit of look at the schematic trying to understand how they load. It looks like one of the enables for the font loading is when the screen is in the VBL period, in the Apple II/III this is 1/4 of the total screen scan time (I love that book Understanding the Apple II by Jim Sather). I think what really should happen is the char rams get loaded progressively as the vbl scan period scans the screens holes following the same quirky Apple II/III memory scheme. I hope that makes some sense.

The current trigger for font loading is when VBL starts, it then loads them all. I did a quick mod to invert the call to load the fonts (added .invert()) so it does the character load at the end of the VBL period, and it works much better. I did get a font error once with BOS, but much better visually with this quick fix.

/Rob

Thanks Rob. Does the entire font get loaded in one VBL? This is something that would be interesting to test on hardware.

Ahh, my description using the word 'all' was a little bit misleading. It looks like it is able to load 8 characters x 8 rows each VBL period. This lines up with the number of screen holes available. I relooked at an example of font loading code that i recently disassembled (makes more sense now), and they preload the 8 characters font data, enable the chrwrt, and then wait for pos VBL, then neg VLB, then disable chrwrt. This lines up with the thinking on how the video scanning loads them. See below for code snip of this.

/Rob

ldx #$90 ;msb of address for font data
lda #$00 ;lsb of address for font data
jsr L0CAB ;go load font
jmp L0E00

; load character set
L0CAB: sta SRCPTR
stx SRCPTR+1
lda #$00
sta L1401
sta STARTBLK
lda #$18 ;disable VBL interupts
sta E_IER
lda E_PCR
and #$0f
ora #$30 ;set CB1 positive edge, CB2 negative edge
sta E_PCR
@L0CC5: lda #$07
sta ENDBLK
@L0CC9: jsr @L0CFE
inc STARTBLK
lda SRCPTR
clc
adc #$08
sta SRCPTR
bcc @L0CD9
inc $01
@L0CD9: dec ENDBLK
bpl @L0CC9
bit SETENCWRT ;enable font loading
ldx #$18 ;clear both CB1 & CB2 flags
stx E_IFR
@L0CE5: lda E_IFR
and #$10 ;look for CB1 flag, pos edge start VBL
beq @L0CE5
stx E_IFR
@L0CEF: lda E_IFR
and #$18 ;look for CB1 & CB2 flag, CB2 = neg edge end VBL
beq @L0CEF
bit CLRENCWRT ;disable font loading
lda STARTBLK
bpl @L0CC5
rts