Blackthorne (U) doesn't start in 3-29-06 WIP

denzilla · Post by **denzilla** » Sat May 06, 2006 5:09 am

Doesn't start in latest WIP. Music appears to begin playing but gets stuck and screen stays black.

NSRT v3.3 - Nach's SNES ROM Tools

---------------------Internal ROM Info----------------------
File: Blackthorne (U).SFC
Name: BLACKTHORNE Company: Interplay
Header: None Bank: LoROM
Interleaved: No SRAM: 0 Kb
Type: Normal ROM: 8 Mb
Country: USA Video: NTSC
ROM Speed: 120ns (FastROM) Revision: 1.0
Checksum: Good 0x3AC9 CRC32: 856BEAB1
MD5: EE33097D436B4901A71F7E4C9A5CB409
--------------------------Database--------------------------
Name: Blackthorne
Country: USA Revision: 1.0
Port 1: Gamepad Port 2: Gamepad
Genre 1: Platform Adventure Genre 2: Shooter

Post by **Agozer** » Sat May 06, 2006 11:51 am

Confirmed.

denzilla · Post by **denzilla** » Sun May 14, 2006 10:10 pm

Confirmed still broken in newest 5-14-06 WIP as well.

Deathlike2 · Post by **Deathlike2** » Sun May 14, 2006 10:18 pm

This probably won't be fixed until the new core is completed... I'm not sure if this game uses a chip (probably not), but it probably has intricate timing that ZSNES's current core cannot do just yet.

denzilla · Post by **denzilla** » Sun May 14, 2006 10:38 pm

But it worked in older versions, I think. I wasn't compaining, just updating the current status of the bug.

snkcube · Post by **snkcube** » Mon May 15, 2006 1:19 am

denzilla wrote:But it worked in older versions, I think. I wasn't compaining, just updating the current status of the bug.

Things break over time and it'll eventually be fixed.

Jonas Quinn · Post by **Jonas Quinn** » Mon May 15, 2006 5:34 pm

Disable Stereo Sound to fix it.

Edit: All games that broke with the latest DSP changes seem to work with Stereo Sound disabled.

This commit message broke it:

Code: Select all

2006-02-27 18:32  grinvader

	* src/: initc.c, cpu/dma.asm, cpu/dspproc.asm: Silly DSP code
	removed, should fix a lot of games (including EWJ2) [pagefault].
	Removed now useless EWJ2 hacks.

Post by **grinvader** » Mon May 15, 2006 7:55 pm

Jonas Quinn wrote:This commit message broke it:

Code: Select all

2006-02-27 18:32  grinvader

	* src/: initc.c, cpu/dma.asm, cpu/dspproc.asm: Silly DSP code
	removed, should fix a lot of games (including EWJ2) [pagefault].
	Removed now useless EWJ2 hacks.

Hmm, the only change in that with any real impact on games is in dspproc.asm.
Maybe too much was deleted ?

Post by **pagefault** » Mon May 15, 2006 9:48 pm

Fixed and in SVN.

http://tinyurl.com/orgkf

Jonas Quinn · Post by **Jonas Quinn** » Mon May 15, 2006 10:13 pm

Awesome. This fixed a lot of issues like FEOEZ freezing.

denzilla · Post by **denzilla** » Mon May 15, 2006 10:54 pm

Thanks pagefault

ICEknight · Post by **ICEknight** » Tue May 16, 2006 3:51 pm

Jonas Quinn wrote:
Code: Select all
Removed now useless EWJ2 hacks.

So does the sound in EWJ2 play correctly now?

Post by **pagefault** » Wed May 17, 2006 9:28 am

ICEknight wrote:
Jonas Quinn wrote:
Code: Select all
Removed now useless EWJ2 hacks.
So does the sound in EWJ2 play correctly now?

The fix was for the music, not the sound effects.

byuu · Post by **byuu** » Wed May 17, 2006 10:04 am

The sound effects in this game are really, really time sensitive. For example, when I rewrote my APU core I was accidentally adding one extra cycle per opcode (called add_cycle() immediately after opcode fetch, despite the fact that opfetch memory read added a cycle anyway), and this totally broke EWJ2's sound effects.

Until ZSNES implements true APU opcode cycle counting, I do not believe EWJ2's sound effects can be fixed without an absolutely brilliant hack.

FitzRoy · Post by **FitzRoy** » Wed May 17, 2006 7:33 pm

Have you any guess as to why the (e) region still exhibits sound problems despite this fix? Now that's wierd to me. Would the code for the game be that different between regions, or is bsnes doing something specific for PAL that is allowing the bug to still happen?

byuu · Post by **byuu** » Wed May 17, 2006 7:35 pm

I don't know. I don't have a PAL unit so I am guessing the clock speed of the PAL processor, as well as things like cycles/scanline, when events occur, etc.
It's mostly a clone of the NTSC timing with a slightly adjusted CPU clock rate and with an extra 50 lines per field.

I'm more interested in why Mortal Kombat (E) works. The US one is doing something extremely screwy. I can't even tell what's happening because the screen is so chaotic. I'll try EWJ2(E) with my improved APU core and sync code and see how things go tonight.

Post by **Nach** » Wed May 17, 2006 7:43 pm

byuu wrote: I'm more interested in why Mortal Kombat (E) works. The US one is doing something extremely screwy. I can't even tell what's happening because the screen is so chaotic. I'll try EWJ2(E) with my improved APU core and sync code and see how things go tonight.

Take MK U and run it in PAL mode, and it'll probably work too.
At least back in the old ZSNES days that worked.

It's actually because of Mortal Kombat I allowed the old versions of NSRT to hack the video mode of a ROM.

Post by **pagefault** » Wed May 17, 2006 11:45 pm

Nach wrote:
byuu wrote: I'm more interested in why Mortal Kombat (E) works. The US one is doing something extremely screwy. I can't even tell what's happening because the screen is so chaotic. I'll try EWJ2(E) with my improved APU core and sync code and see how things go tonight.
Take MK U and run it in PAL mode, and it'll probably work too.
At least back in the old ZSNES days that worked.

It's actually because of Mortal Kombat I allowed the old versions of NSRT to hack the video mode of a ROM.

The only difference in PAL mode is that rigth now it is slightly bugged to allow more executing tha NTSC, so if you adjusted the execution percentage in the config file you should be able to reproduce the same results as running the game in PAL mode.

byuu · Post by **byuu** » Thu May 18, 2006 12:58 am

Well, the thing is my NTSC timing is (probably) 99.97% accurate or so, and it's the NTSC version of MK that is bugged :/

I can only guess it's something like a bug in the NMI / IRQ handler...