Resolved: [Bug Report] Valkyrie Profile 2 [NTSC]

[Krakatos]

ok this is an example for everyone who wants to report a bug.


# Pcsx2 version: official playground release 1.0.0395
# Cpu eerec on, vurec0 and vurec1 on, MTGS on
# Plugins used: Gsdx 0.1.12 or zerogs 0.9.6, spu2ghz 1.9.0 or zerospu 0.46
# Description: Game will hang before booting. No setting or plugin is known to fix it. No version of Playground nor official pcsx2 is known to solve this problem. the game is simply unplayable.
No need to attach screenshots or memcards obviously since the gae does not boot.

And just so you know the problem lies in the encryption/obfuscation of the game data. Pcsx2 can't emulate with sufficient precision the numbers now, and a solution is really hard to find. Do not expect this to be fixed soon

[refraction]

If any of the devs want insight in to whats happening, there is a "hack" for it to make it work which was discussed and rooted out by members on IRC. I have rough knowledge of what to do (its been a while since ive done it ) but maybe someone can find a solution.

[rama]

refraction:
I'm very interested to know how thats done.
I guess you do the usual opcodes on fpu and vu, but patch the registers with the correct values?

[ameppe]

If any of the devs want insight in to whats happening, there is a "hack" for it to make it work which was discussed and rooted out by members on IRC. I have rough knowledge of what to do (its been a while since ive done it ) but maybe someone can find a solution.

This is the same case in Star Ocean 3.I remember someone said the hack worked with both games.

[tiku]

I guess it will be a common issue at all Tri-Ace games.

A total shame, 'cause SO3, Radiata Stories, VP2, etc may look really impressive with higher resolutions and filterings

[refraction]

refraction:
I'm very interested to know how thats done.
I guess you do the usual opcodes on fpu and vu, but patch the registers with the correct values?

IIRC its actually to do with the decryption method run on VU0, using one of the VU's "Add" instructions, if you put +1 on the Z vector of one of them (cant remember which) it works. Its all due to a miss calculation

[rama]

Very interesting.
If you look at the svn, you'll see that FPU just got a big fix in that direction.
I'm sure it would work on the VUs as well

[cottonvibes]

IIRC its actually to do with the decryption method run on VU0, using one of the VU's "Add" instructions, if you put +1 on the Z vector of one of them (cant remember which) it works. Its all due to a miss calculation

do you mean just setting the Z vector of the result to +1 whenever it does an Add() instruction?
did you save the old code where you got it to work? that'll be very helpful

but at least now i know its a VU0 problem and something with the Add opcodes, that narrows it down.
although it might not really be the Add opcodes fault, but instead bad-data going into the Add opcodes :o

maybe we should move this discussion into the dev section to see if we can find a solution.

[Krakatos]

No please no moving this thread If you want, start a new one in dev forum, this is my example thread for the usage guidelines of this forum XD

[icemann]

Very interesting.
If you look at the svn, you'll see that FPU just got a big fix in that direction.
I'm sure it would work on the VUs as well

Cept all of that got commented out in 426 judging from the revision notes.