SMS core

[slingshot]

The same bug with the left column is there with SMSFLY_MIST_20190427.rbf.
I can say it's buggy when the left column is not masked. X==0 is the bogus column, but it really should be the first one. I noticed you've changed hblank_end from 511 to 0, but it was there because if you set hblank at 0, it'll be effective at the next clock only. Not sure if it's the problem however.

[theflynn49]

Just found out that column_mask and pal are not connected correctly in the MiST toplevel. But correcting them won't fix the issues. Will try from your repo.[/quote]


good catch

[theflynn49]

The same bug with the left column is there with SMSFLY_MIST_20190427.rbf.
I can say it's buggy when the left column is not masked. X==0 is the bogus column, but it really should be the first one. I noticed you've changed hblank_end from 511 to 0, but it was there because if you set hblank at 0, it'll be effective at the next clock only. Not sure if it's the problem however.

Ok let me explain.
I had to move all sprites 1 pixel left. I prefered to move the background 1 pixel to the right so I don't duplicate logic 64 times.
While I was doing this I found another bug : sprites were clocked by ce_cpu and not ce_pix, so they were off by half a pixel;
Once I fixed that bug, it was off 1 pixel again, and I finally chose to move the sprites left 1 pixel (vdp_sprites.vhd line 165)

note that the sprites are right 1 pix because the logic load them in the register but they get out to "color" only the next clock in time.
So I might have left the background in a bizarre state here.

[slingshot]

Yepp, as I see, your changes moved the background from the range of 0-255 (inclusive) to 1-256. But then x==0 should be the border color then.
Like this in vdp_main.vhd

Code: Select all

-               if ((x>48 and x<=208) or (gg='0' and x<=256)) and
+               if ((x>48 and x<=208) or (gg='0' and x>=1 and x<=256)) and

PS: that worked. Now only the USE_SP64'less mode should be fixed.

[theflynn49]

@slingshot,
please try to change vdp_main line 130

IS : if ((x>48 and x<=208) or (gg='0' and x<=256)) and
s/b : if ((x>48 and x<=208) or (gg='0' and x<=256 and x>0)) and

tell me if that change something.

[theflynn49]

now this is funny.

very good catch, thank you slingshot

[Sorgelig]

did i release it too early?

[slingshot]

did i release it too early?

If you enable the border, you'll get a garbage column at the left if it's not masked.

[slingshot]

now this is funny.

very good catch, thank you slingshot

Found the same solution, so it must be correct
But I'll leave the USE_SP64 thing for you!

[theflynn49]

@sorgelig
Well, that's an ugly bug if lots of people have old CRT screens. I think my screen was more accurate on HBlank so I didn't see this.
I wish I could fix mist/SMS.sv too, plus I should change the parameter name "Disable mapper", I'll do that tonight.

@slingshot Ok I found a bug 3 weeks ago about the number of sprites, so it should be clear in my mind. Notice that when you are at 64 sprites, you use a diffrent clock for computing the sprite array, that is a nice hideout for nasty bugs

[slingshot]

I can update the MiST version before a release, it's not urgent to fix it.

[theflynn49]

SMSFLY-HOTFIX1_20190427.rbf committed if someone wants to play with it.

[theflynn49]

@slingshot : I just committed the fix for the USE_SP64 flag.

Easy, I am the one who broke it.

[slingshot]

@theflynn49 Everything's fine now, thanks!

[albconde]

Some more test with SMSFLY-HOTFIX1_20190427.rbf in gamegear, only found a new bug:

earthworm jim : in game you don´t see the lives or the energy at the bottom of the screen , the game is playable

there are some garbage in jungle king , but maybe game related (bad sms euro port ) ,the same problem in emulators.

[theflynn49]

Some more test with SMSFLY-HOTFIX1_20190427.rbf in gamegear, only found a new bug:

earthworm jim : in game you don´t see the lives or the energy at the bottom of the screen , the game is playable

that was a tough one:)
The specs do not define clearly what should be done for sprites when they are tall and wide at the same time. With that game I've been able to find out exactly Fix is in SMSFLY_20190429.rbf in my repo

there are some garbage in jungle king , but maybe game related (bad sms euro port ) ,the same problem in emulators.

Well, isn't the answer in the question ? lol ...

[theflynn49]

New SMSFLY_20190430.rbf plays Ernie Els Golf.

Enjoy

[albconde]

excelent dizzy collection work too in MSFLY_20190430.rbf

[theflynn49]

excelent dizzy collection work too in MSFLY_20190430.rbf

I am afraid only the third game, "Dizzy Panic" works, and in PAL unless you have graphic glitches at the top.

[albconde]

I am afraid only the third game, "Dizzy Panic" works, and in PAL unless you have graphic glitches at the top.

In version 29 only work the third , in version 30 the three games work

[theflynn49]

Mine doesn't work. Could you put here the CRC32 of the game's rom so I can be sure we are using the same version ?

Also, OutRun.sms (not Europa) does work with Mist but not with Mister with my boards ... weird

[albconde]

Excellent Dizzy Collection, The (Europe).sms
Size: 524,288
CRC-32: AA140C9C
MD-5: 613376b7b53e43ba17ae1b62da3c9251


sms Outrun work with your last version , with sound problems

[theflynn49]

Ok : not the same

262144 juin 28 2017 'Excellent Dizzy Collection, The (USA, Europe) (En,Fr,De,Es,It) (Proto).sms'

From the size I'd think I miss a good part ot it

edit: And I just found that most of my OutRun roms begin with $200 bytes of garbage, reason why they don't work.

[slingshot]

edit: And I just found that most of my OutRun roms begin with $200 bytes of garbage, reason why they don't work.

I think the skipping of an 512 byte header is still there in SMS.sv (or should be).

[theflynn49]

@slingshot : ok I just read that part. I don't quite get why ioctl_addr[9] does mean there is a 512-bytes header ... the rom is > 128k ?