IRQ timing bug in SMB3 (J)?

Discuss emulation of the Nintendo Entertainment System and Famicom.

Moderator: Moderators

Post Reply
User avatar
Zepper
Formerly Fx3
Posts: 3262
Joined: Fri Nov 12, 2004 4:59 pm
Location: Brazil
Contact:

IRQ timing bug in SMB3 (J)?

Post by Zepper »

It occurs even in Nintendulator. Pic from RockNES.
Super Mario Bros. 3 (J) 000.png
The only emulator that displays it perfectly is VirtuaNES (see below)
Super Mario Bros. 3 (J) 201702092119578.png
Super Mario Bros. 3 (J) 201702092119578.png (8.93 KiB) Viewed 5961 times
Last edited by Zepper on Thu Feb 09, 2017 4:22 pm, edited 1 time in total.
tepples
Posts: 22708
Joined: Sun Sep 19, 2004 11:12 pm
Location: NE Indiana, USA (NTSC)
Contact:

Re: IRQ timing bug in SMB3 (J)?

Post by tepples »

Splits in Super Mario Bros. 3, Mega Man 3, and possibly other early MMC3 games are glitchy on hardware. Have you compared this to the behavior of an authentic Super Mario Bros. 3 cassette in an authentic Family Computer?
User avatar
Zepper
Formerly Fx3
Posts: 3262
Joined: Fri Nov 12, 2004 4:59 pm
Location: Brazil
Contact:

Re: IRQ timing bug in SMB3 (J)?

Post by Zepper »

tepples wrote:Splits in Super Mario Bros. 3, Mega Man 3, and possibly other early MMC3 games are glitchy on hardware. Have you compared this to the behavior of an authentic Super Mario Bros. 3 cassette in an authentic Family Computer?
I don't own a Famicom.

EDIT: looks like it's not a bug after all.
https://www.youtube.com/watch?v=lXSJeKgbBeQ
User avatar
zeroone
Posts: 939
Joined: Mon Dec 29, 2014 1:46 pm
Location: New York, NY
Contact:

Re: IRQ timing bug in SMB3 (J)?

Post by zeroone »

It is difficult to tell exactly what you are referring to based on your description and pics. Are you referring to one of these:

https://wiki.nesdev.com/w/index.php/Game_bugs
User avatar
Zepper
Formerly Fx3
Posts: 3262
Joined: Fri Nov 12, 2004 4:59 pm
Location: Brazil
Contact:

Re: IRQ timing bug in SMB3 (J)?

Post by Zepper »

zeroone wrote:It is difficult to tell exactly what you are referring to based on your description and pics. Are you referring to one of these:

https://wiki.nesdev.com/w/index.php/Game_bugs
Difficult!? :shock: Look again.
glitch1.png
glitch1.png (7.76 KiB) Viewed 5938 times
tepples
Posts: 22708
Joined: Sun Sep 19, 2004 11:12 pm
Location: NE Indiana, USA (NTSC)
Contact:

Re: IRQ timing bug in SMB3 (J)?

Post by tepples »

In case anyone missed it, Zepper refers to a white line between the copyright notice and the checkerboard.
User avatar
zeroone
Posts: 939
Joined: Mon Dec 29, 2014 1:46 pm
Location: New York, NY
Contact:

Re: IRQ timing bug in SMB3 (J)?

Post by zeroone »

tepples wrote:In case anyone missed it, Zepper refers to a white line between the copyright notice and the checkerboard.
That actually look me a minute to notice.
Zepper wrote:EDIT: looks like it's not a bug after all.
As Zepper pointed out, there are plenty of videos of SMB3 (U) on YouTube recorded from real hardware and they do indeed have that white horizontal scanline on the title screen.
zxbdragon
Posts: 498
Joined: Mon Dec 12, 2011 8:15 pm

Re: IRQ timing bug in SMB3 (J)?

Post by zxbdragon »

smb3 j
PRG0,is have this line
PRG1,is not have this line.

smb3 u
PRG0,PRG1 have this line.
User avatar
Zepper
Formerly Fx3
Posts: 3262
Joined: Fri Nov 12, 2004 4:59 pm
Location: Brazil
Contact:

Re: IRQ timing bug in SMB3 (J)?

Post by Zepper »

zeroone wrote:As Zepper pointed out, there are plenty of videos of SMB3 (U) on YouTube recorded from real hardware and they do indeed have that white horizontal scanline on the title screen.
I mean the japanese version! In other words, "there are a few videos of SMB3 (J) on YouTube recorded from real hardware and they have NO white horizontal scanline on the title screen".
zxbdragon
Posts: 498
Joined: Mon Dec 12, 2011 8:15 pm

Re: IRQ timing bug in SMB3 (J)?

Post by zxbdragon »

Zepper wrote:
zeroone wrote:As Zepper pointed out, there are plenty of videos of SMB3 (U) on YouTube recorded from real hardware and they do indeed have that white horizontal scanline on the title screen.
I mean the japanese version! In other words, "there are a few videos of SMB3 (J) on YouTube recorded from real hardware and they have NO white horizontal scanline on the title screen".
SMB3 J
PRG0 version is have this white horizontal scanline.
I have PRG1 version cart.
Attachments
IMG_20170212_202218.jpg
zxbdragon
Posts: 498
Joined: Mon Dec 12, 2011 8:15 pm

Re: IRQ timing bug in SMB3 (J)?

Post by zxbdragon »

this is USA version
Attachments
IMG_20170212_202218.jpg
ap9
Posts: 43
Joined: Sat Jun 01, 2013 11:55 am
Location: Maine, U.S.A.
Contact:

Re: IRQ timing bug in SMB3 (J)?

Post by ap9 »

What makes it clear that there is an IRQ bug in the earlier MMC3-A Japanese ROM(s) is the glitch line produced in the castle levels where the ceiling drops. It's revealed in the YouTube video mentioned by Zepper in this thread, beginning at 6:08.

It appears that only one of the official releases for Super Mario Bros. 3 drops most of the white line on the title screen — the more common Japanese PRG1, so it's safe to assume that the black area was unintentional. (And if it was intentional, it's still not in the modern form expressed in all future releases, U.S.-on.)

There's more extensive research into the differences between the ROMs on The Cutting Room floor website, where the black area can be seen:

https://tcrf.net/Super_Mario_Bros._3/Ve ... ifferences
Attachments
With puNES, which emulates the MMC3-A board very closely to the real thing
With puNES, which emulates the MMC3-A board very closely to the real thing
YouTube frame of actual Famicom
YouTube frame of actual Famicom
kazblox
Posts: 35
Joined: Thu Feb 23, 2017 2:27 pm

Re: IRQ timing bug in SMB3 (J)?

Post by kazblox »

Comparison of IRQ line timings to the Japanese prototypes:
Attachments
Famimaga Sept. 1988
Famimaga Sept. 1988
GTV Jul. 1998
GTV Jul. 1998
GTV Jul. 1988
GTV Jul. 1988
ap9
Posts: 43
Joined: Sat Jun 01, 2013 11:55 am
Location: Maine, U.S.A.
Contact:

Re: IRQ timing bug in SMB3 (J)?

Post by ap9 »

kazblox wrote:Comparison of IRQ line timings to the Japanese prototypes:
The early prototypes make it obvious that Mario is slightly intersecting the checkerboard pattern. Technically, he's always intersecting the first split line, but aesthetically, it may have been better to have the characters meet a black line first (as the sprites have black outlines). The white line is very much present in the prototype snapshots.

It was probably necessary to move the second part of the split a little lower due to DMC CPU read delays during DMC sound effects (the effects of the delays still show up in overscan-clipped areas). It would be interesting to know whether or not DMC sound effects were even used until later, considering the early test videos had no game play sound. The controller read / DMC DMA conflict bug only made things more complicated; however, we all know Super Mario Bros. 2 (U) used DMC in its 1988 release.

There were a number of bugs to work out, and some still present including occasional glitches during "water float" sequences while several actions are slowing down the game, so two extra black scanlines added considerable IRQ split leeway in the end. The top of the scoreboard split is one pixel lower in a level than it is on a world map.
Post Reply