News: 11 March 2016 - Forum Rules, Mobile Version
Current Moderators - DarkSol, KingMike, MathOnNapkins, Azkadellia

Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - sluffy

Pages: [1] 2 3 4 5 6 ... 11
Just noticed that block randomly happens in snes9x too (down-left walk). So 10A2 can blow up the rom. Unsafe. And going over 2MB isn't working yet. Guess I'll have to try diff'ing the roms to know exact changes. I'll think of something.

Thanks for reporting. A few more posts and you'll be out of moderation. ;)

And since this might've got buried some posts back:
Preset ID: 109C (Super Mario RPG)
Block: 0-1FFC

Preset ID: 109F (Kirby Super Star) [needs confirmation]
Block: 1F00-1FFE

Preset ID: 123D (Yoshi's Island)
Block: 7C00-7E7B

Preset ID: 1245 (Star Fox 2) [needs confirmation]
Block: 3912-3A45

Seems that SNES games that have special chips generate their SHA-1 for saves differently from the standard method.

This info is needed by BigLongJohnson for doing Canoe sram transfers.

Probably need video of 1 EA sports hang. Later ones are all similar engines.

So one idea is screwed until someone figures out how to hack homebrew SA-1 games for Canoe. Other methods aren't getting through.

test1-3 = 0000 pid boots okay. 3-4 MB LoROM. Transparency problem.
*** We know that LoROM mapping is okay.

test1-3 = 10A2 pid dies. Unknown reason.
test4 = Kirby3 disguise. SA-1 dies regardless. Emus okay. Unknown reason.

JP = 0000 boots okay. 2MB LoROM. Transparency problem.
JP = 10A2 works. Maybe needs nolow-latency. Grant artifacts.

This will replace raptors, T-Rex with Grant. 10A2 pid. rev1. clean. 2MB LoROM. Purely for info finding.
== If we get same artifacts, it's blending layering algorithm
== If we get no artifacts, 10A2 is corrupting rom
== If we get different artifacts, could be even more 10A2 rom corruption
== If we get no boot, Canoe is being an *#

Moving on to others:
- Hoping Uniracers joypad test works. That would be critical for (dis)proving OAM theory. And get 2-3 more games out the door.

- EA Sports. Got lots of requests. So. These hang right before speech sample is playing? Take NHL - referee triggers hdma audio streaming for blowing whistle. Portrait shows and hangs? Or mutes mid-sample?

- Top Gear. May want orange highlighter on docs. Many will overlook this. Glitch #2? And video of #3 for frame stepping.

- Space Football: One on One. Docs says: no boot. Other says hang after title?

- Dark Law. I should take a look at fixing the glitch hud.

Saw hidden post now.

game boots with 0000 for all the patches, glitchness around Grant persists

Now I'm confused. :)

Grant glitched pixels:
vanilla 0000 = okay? bad?
vanilla 10a2 = bad?
test 0000 = same glitch?
test 10a2 = black

0000 pid = sram okay. Should match.
10a2 pid = sram program not run. says NINTENDO.
10a4 pid = sram program not run. says NINTENDO.

Inconclusive. Basically idea failed since can't run test correctly yet. Thanks for trying.

vanilla = FF42B93F32C7921A88EF46FDF79A8445
JP test4 = 6FAF1B48D6B260C92416FC56EFCFE8D2

===1 Fun code
Code: [Select]
NHL 96 -- boot crash for snes9x 1.51
$87/F120 4C 20 43    JMP $4320  [$87:4320]   A:0000 X:AF93 Y:5001 D:4300 DB:7E S:03FC P:eNvmxdIzc HC:0212 VC:147 FC:26 I:00

run code from dma registers ...
874320 mvn $c1,$7e            A:0000 X:af93 Y:5001 S:03fc D:4300 DB:7e NvmxdIzc V:140 H: 644 F:24
874323 jmp $f0f5     [87f0f5] A:ffff X:af94 Y:5002 S:03fc D:4300 DB:7e NvmxdIzc V:140 H: 690 F:24

Uniracers -- debug testers (0000 pid, USA, clean rom)

Start a game. Don't move. Wait for sticky cpu. Now press select. Keep doing this until both racers disappear. Count how many clicks it takes.

Now believe that Uniracers, Nosferatu, Eye of the Beholder all share same oam address bug. Not uploading to correct spot.

For Street Racer, it was changing sprite size mid-frame. Mohawk title. Outlander title. Have to downsize sprites manually somehow.

For JP, maybe Canoe 10A2 didn't expect a 3MB LoROM game. May try SA-1 converting this (Kirby3 = 4MB LoROM). But want to get Uniracers another look.

Jurassic Park (rev1, USA, 10A2 pid, clean rom)

SA-1 map. Unknown results.

Patch good to fix flicker line. Needs 1068 to fix Volcano transparency regardless of patch.

Long day at work. Street Racer.

Street Racer (USA, 0000 pid, clean rom) --- broken menu racers

Battle Pinball (Japan, 0000 pid, clean rom) --- underwater table crash

NMI within NMI = very bad.

I think it depends on whether it's cpu bottleneck (streaming music sheet like Lost Vikings II) or purely apu side. They tend to be rather difficult to optimize and a lot of work can get you just a few vlines (= nothing in my experience with several games). PAL ==> NTSC timing problems can be monstrously hard on cpu alone (those extra vblank lines are definitely helpful).

But I do know that Canoe is running cycles like 9x 1.51 engine == between slow and slower. You feel slowdowns more often in that emu. I'll keep it in mind if I find some easy trick. Lost Vikings II was pretty sloppy at times, and surprisingly efficient elsewhere. Squaresoft .. never looked that deeply before. But they're like Quintet -- structured everything.

Or! You could cheat and add basic sa-1 coprocessor to do stupid things like oam filling. But I forget if SA-1 is all LoROM only. That'd be hellishly easier than breaking down the rom. Since it's Canoe and not copiers. Of course, I thought about that much later after LV2,Cacoma was sped up decently. Would've saved waterfalls of tears.

Right. Back to next game.

Most experience I've had is cheat hacking. Same for other systems. Emus, dosbox, adventure games I'm more familiar with. Snes allows rom "expansion" or jumping around so it's easier to play with. Can't do this so easily with say PSX. NES,GB/C so-so. Cheat Engine trainers yeah.

And some games have slow-down. Which bothered me enough to try LV2 for experiment. Plus it's harder to fix emu than actual game sometimes.

Thanks for saves. Working on #3 first. And they add more complexity - updated dynamic scrolling crud. Un-fun.

Ark: If C7 didn't crash, 0000 pid would still draw black. Just like Little Magic. Have to fix draw engine to use small 8-bpp tiles.

Mystic Ark (0000 pid, Japan, clean rom) --- fix cutscenes 1,2,3

Jurassic Park (10A2 pid, rev1, USA, clean rom) --- combine both testers
===> Just check if sprites are corrupt

Might be Front Mission like problem with test2 failing.

With 0000 pid, Mystic Ark patch fixes C7 and 1st cutscene background. When I get other 2 saves, have to patch those bg1+2 cutscene tilemap data into rom. Those 2 scenes will look wrong in all emus atm. It'll be very obvious.

tl;dr -- cutscene 1 fixed. Waiting on #2,3 save files to fix those. Then done.

Thanks all for testing.

Great Battle 4 (Japan, 0000 pid, clean rom) --- gui flicker (wait h-blank)

Great Battle 3 (Japan, 0000 pid, clean rom) --- gui flicker (wait h-blank)
===> This one can't see so taking guess from #4

Has all Mystic Ark fixes. Have to tweak each cutscene - it'll look mini sized

Tin Star (USA, 0000 pid, clean rom) -- cutscenes

JP1 test1 = please dump sram after game boots. Compare 10A2 vs 10A4. If first two bytes is not 0000, upload file.

Not understanding why test2(a) won't boot. test1 = 3MB LoROM also.

Try this for mystic ark slowdown  (0000 pid, japan, yada)

Mystic Ark (Japan, 0000 pid, whatever rom) --- cutscene 1, waking giant

Short answer: hook game carefully and send my own tilemap. Lucky vram fits just enough to add 2 more tilemaps.

JP Rev1:
test1 = dump sram to see if pid 10A2/10A4 changed anything
test2 = 3MB LoROM with pid 10A2/10A4 to test graphics fix

test2a = 4MB LoROM with pid 10A2/10A4 to test graphics fix (modify header)

There's no room to refit graphics into 2MB so must expand.

Thanks as always. I'd say keep aio debug1, since closest to original. Something about cleaning echo buffer is crashing Canoe audio. Don't see what trips on; maybe sometime later since could be useful for remaining C8 games.

Jurassic Park. Could be ... huh. Maybe not hikachi detecting rom correctly? It's 3MB LoROM so unusual sized. #2 should've clean booted.

Top Racer (Japan, 0000 pid, clean rom) -- flickering horizon

Lost Vikings 2 (USA, 0000 pid, clean rom) --- aio debuggers (runs, c8, black)

Pages: [1] 2 3 4 5 6 ... 11