FPGAmstrad on MiST-board

[freemac]

First day of THSF festival, whatever age of Amstrad tester, one common return : "have you got a 2 player games ?"
So we played :
- Double Dragon II (several times in fact)
- Arkanoid (alternated)
But failed in running Ikari (player lefted), and 1943 (joy+keyboard only ?)



Damn, why didn't I launched Psycho pigs ?


Have you got a list of 2 player games ? with a preference using two joysticks, I do remember a moto game...
EDIT : http://www.cpcwiki.eu/forum/games/amstrad-multi-player-games/

EDIT2 :[/size] - Grand Prix 500cc 

[freemac]

Third day of THSF : three hours of Gauntlet II game, cooperative. Death stupidly at level 35 against sort of blood sucker or Darth Vader.

[freemac]

r005.8.14.3 : robustness (quality of code)

[freemac]

r005.8.14.4e : Green OSD while screen is in green mode. And an experimental 800x600@71.18Hz

[attachimg=1]

After some experiments, I succeed having a resolution that can be shown in a SAMSUNG 16/9 (poor low cost SyncMaster B2430), if I do press the "auto" button at bottom of screen :
[attachimg=2]
[attachimg=3]

Tested ok on 4/3 screen, seems running fine on a stupid 5/4 screen.

r005.8.15 is still in progress (arnoldemu testbench : 12% of begining tests are running ok...)

[ralferoo]

I think @ralferoo had already written FPGA code for tape reading for his FPGA CPC. Maybe you can borrow some code from him?

CPC FPGA had a physical tape input port. I never bothered implementing CDT playback mostly because I have hardly any CDT files but lots of real tapes.


Oddly, I rarely used CPC FPGA for demo development as I had a real CPC on the desk next to the PC anyway.

[freemac]

r005.8.15 : arnoldemu's old FDCTEST testbench applied. Failing results are listed and explained here :http://www.cpcwiki.eu/index.php/FPGAmstrad#arnoldemu.27s_testbench_results

Batman Forever does now run ok using the single disk version.

Somes "Nigel Mansell's Grand Prix" tracks unlocked.


https://www.youtube.com/watch?v=WvtE50yzr_Q

[Gryzor]

I love the green OSD

[freemac]

r005.8.16c3 : begin of Z80 instruction timing effort, 32 instructions timing fixed in MEM_wr:slow mode.

So MEM_wr:slow OSD parameter became serious (it was a theorical parameter, you can turn it now by default in practice ^^')

Tests done :
* Still Rising demo : less garbage pixels, can be run completly using MEM_wr=slow
* Logon's run - 3D meets the aging bits : better, can be run completly using MEM_wr=slow

[freemac]

r005.8.16c4 : Z80 instruction timing effort, all main instructions timing fixed in MEM_wr:slow mode except one (for me 2A does take    4 NOPs...)
WAIT_n generator updated : for instruction using 5T during M1 cycle, I insert two WAIT_n (experimental)

Tests done :
* Still Rising : no more pixel garbage in plasma using MEM_wr=slow
* 30YMD : main menu (with cursor) running fine using MEM_wr=slow

It's still a candidate version (c4 equals 4th backup of wip). I could only give two days of full work on this realise (c3+c4)... I do also a lot of other activities theses holidays. Have fun !


https://www.youtube.com/watch?v=ovAuDRD2wj4

Tests done:
* @00:20 Wunder.bar 2017 OK
* @02:30 Logon's run - 3D meets the aging bits 2017 OK : does not freeze
* @06:30 Still Rising (Vanity) 2013 OK : plasma unlocked
* @12:40 Batman 2015 KO : chip flying ground full of pixel garbage (CRTC offset ?)
* @24:00 Ecole buissonière 2000 OK
* @34:00 30YMD 2016 menu OK, demo GPA KO : a lot of garbage lines (CRTC offset ?), same problem in r005.8.16c3 <= oups, 30YMD's GPA runs better using CRTC0 ^^'

[freemac]

r005.8.15.1: VGA counters fix... How can't I see that before ? (found while merging ZX-Uno version...)
Taken into account also in experimental r005.8.16c30 (experimental != validated)

[DanyPPC]

Thanks freemac, what kind of demo/game may we test ?

[Gryzor]

I toyed with this core over Xmas. I didn't test it extensively, I must say, but the progress is admirable, so thanks for updating mate!

[freemac]

Thanks freemac, what kind of demo/game may we test ?

Not new demos unlocked for r005.8.15.1 (same as r005.8.15, but just more robustness)

[DanyPPC]

Ok, many thanks for you work.


I have a MiST since 2016 and I bought it mainly for the availability of an Amstrad CPC core. It's one of the better cores available on this platform.

[Gryzor]

That's true. I've spent countless hours going through all the cores and games (all available cracks!) for the ST, and compared to that, the CPC is just goodness.

[freemac]

r005.8.15.2: merge of OSD menu with r005.8.16c34.
Now scanlines is mandatory (one year I didn't touch this option on OSD :p)
The "MEM_wr:quick|low" became "WAIT_n:quick|low" but has same implementation than previous version here.

r005.8.16c34: merge of OSD menu with r005.8.15.2.
Now scanlines is mandatory
The "WAIT_n:quick" is in theory the same as in r005.8.15.2 "WAIT_n:quick", it is to say that Z80 instruction are just modulo 4, but not slowed more using a "CPC Interruption Timing" array. In fact most of games are running fine with "WAIT_n:quick" so I prefer to let this option here
The "WAIT_n:low" is a mix of theory and human calibration succeeding in launch fine the "Logon's run - 3D meets the aging bits" (coolest with option VGA:dbscan50Hz => does show the 1943 plane in background)

Both has now the VGA:dbscan50Hz|vram72Hz option (pure VGA 72Hz VRAM, or VGA via scandoubler component)

Got currently a serious problem on r005.8.16c34 with winape's plustest testbench, I can add whatever WAIT_n per whatever Z80 instruction, but if plustest "Instruction Timing" testbench does pass, "Logon's run - 3D meets the aging bits" does run worst than if I add theoricals "Instruction Timing" delays (ones from plustest ran directly on winape versus Z80 doc), so plustest testbench is not enough to calibrate my "CPC Z80 Instruction Timing", so as plustest does say "does run fine only if NOP and interrupt are well implemented" I think that "NOP or else my interrupt has problem" (so I tryed to delay my NOP, then it is really worse, so finally I think that my Z80 interrupt implementation has a problem...)

I can try a third fork hacking the WAIT_n entry of Z80 by slowing down the Z80 clock entry directly as on r005.8.15.2 ignoring this way the Z80 "WAIT_n" wire implementation, using also a "CPC Instruction Timing" array approach.

[Gryzor]

Thanks for keeping working on it

[freemac]

http://www.youtube.com/watch?v=FhiPkY34yJ4

r005.8.16: WAIT_n:slow is using WAIT_n generator and does pass plustest.dsk
"Logon's run - 3D meets the aging bits" does still pass using VGA:scandb50Hz
"phX" does pass (missing some parts effect but sound is good, and the rotating square also) using VGA:vram72Hz

[freemac]

r005.8.16.1: Pinball Dreams is back

[DanyPPC]

Yes, now it works 


Why no more scanlines option in 72Hz mode ?
I'd like to activate/deactivate them.


Is it possible to fix the graphic bug scroll in Commando ?


Many thanks for you work !

[freemac]

Why no more scanlines option in 72Hz mode ?
I'd like to activate/deactivate them.

scanlines are always activated in this version. But I have seen that Sorgelig does add more options but still using the OSD 6 menu lines, perhaps I can switch between 4 options in color menu : color/green/color+scanlines/green+scanlines.

Is it possible to fix the graphic bug scroll in Commando ?

Commando bugs are in the known bug list (scroll is bad in level 1, moto isn't on bridge in level 2) : http://www.cpcwiki.eu/index.php/FPGAmstrad#On_MiST-board_CoreAmstrad_version
That's a mistery, perhaps it is using CPIR/CPDR Z80 instructions that are not implemented in T80 (does same as CPI/CPD in T80)

[freemac]

r005.8.16.2: + OSD options : VGA:vram72Hz|scanlines72Hz|scandb50Hz

[DanyPPC]

Well Done !

[freemac]

r005.8.16.3: commando unlocked by Sorgelig (I did insert a regression in Z80, my bad)

r005.8.16.4: plustest.dsk tb 5 full OK (CPIR/CPDR timing ok)

[DanyPPC]

WoW !