Is THIS the reason why Harley has severe geometry problems?

[GambitRogue777]

I remember when running Harley in MAME (version 120), and messed around in test mode, and I accidentally slipped into "MEMORY TEST". I noticed there are some bad ROMs.

Right now, I'm going through memory test, and trying to see what the bad ROMs are.

Bad in "CPU ROM TEST":
(CROM3 EPROM) IC.17, 074F
(CROM2 EPROM) IC.18, 868A

All of the video ROMs are bad in "VIDEO BOARD ROM TEST".

The culling and polygon memory, polygon FIFO tests, and the depth buffer fail.

Also, the texture memories are bad.

Should this game get a re-dump? I feel it should. If you see videos on YouTube of people testing this game, you'll see the severe geometry issues. Some people have said the problem is because the game handles so much stuff on screen.

[Bart]

No, the ROMs are fine. All the Model 3 ROM dumps are correct. Many games appear to fail the ROM tests because I patch them in a few places to bypass some obscure hardware tests. The video memory (VROM, polygon FIFO, etc.) tests fail for every game because the CPU is not actually accessing them at all; it is performing indirect tests (JTAG boundary scans, which are a hardware debugging feature built into ASICs by engineers for testing and verification, and otherwise are not part of normal functionality).

Harley (and Ocean Hunter, ECA, and the reversed tracks in Daytona) is broken because the scene graphs (the tree and list structures that the Real3D parses to build a frame) it uploads seem to work differently than other games. Scene graph traversal is not that well understood despite being at the core of the rendering process. My current interpretation either works great (playable games) or fails miserably.

[GambitRogue777]

No, the ROMs are fine. All the Model 3 ROM dumps are correct. Many games appear to fail the ROM tests because I patch them in a few places to bypass some obscure hardware tests. The video memory (VROM, polygon FIFO, etc.) tests fail for every game because the CPU is not actually accessing them at all; it is performing indirect tests (JTAG boundary scans, which are a hardware debugging feature built into ASICs by engineers for testing and verification, and otherwise are not part of normal functionality).

Harley (and Ocean Hunter, ECA, and the reversed tracks in Daytona) is broken because the scene graphs (the tree and list structures that the Real3D parses to build a frame) it uploads seem to work differently than other games. Scene graph traversal is not that well understood despite being at the core of the rendering process. My current interpretation either works great (playable games) or fails miserably.

I see. Do you think giving up the Real3D's secrets (which is what you said on MAMEWorld) could solve these problems? You DID say that fixing this problem will require only some "outside-the-box" thinking.

[Bart]

I see. Do you think giving up the Real3D's secrets (which is what you said on MAMEWorld) could solve these problems?

I said I need to get the Real3D to 'give up its secrets'. That's just a colorful way of saying it will take more reverse engineering (which is most of what constitutes the hard work in emulator development). There's never any guarantees that any more of it will become clear any time soon -- the low hanging fruits have been picked already.

You DID say that fixing this problem will require only some "outside-the-box" thinking.

Only some 'outside-the-box' thinking? Good one

[SupremeTextures]

It is pretty fast that game, close in speed to burnout i think, not quiet as fast. In real life Harleys are very slow compared to other bikes.

[GambitRogue777]

It is pretty fast that game, close in speed to burnout i think, not quiet as fast. In real life Harleys are very slow compared to other bikes.

There was another Harley-Davidson game called "Race to the Rally", and it's considered to be a rip-off of the Burnout series. The unique thing about L.A. Riders is that when you sit on the machine and put your hands on the handlebars, the handlebars have a very good smell.

[kuran]

It is pretty fast that game, close in speed to burnout i think, not quiet as fast. In real life Harleys are very slow compared to other bikes.

The unique thing about L.A. Riders is that when you sit on the machine and put your hands on the handlebars, the handlebars have a very good smell.

Wow.

[GambitRogue777]

It is pretty fast that game, close in speed to burnout i think, not quiet as fast. In real life Harleys are very slow compared to other bikes.

The unique thing about L.A. Riders is that when you sit on the machine and put your hands on the handlebars, the handlebars have a very good smell.

Wow.

Kuran, are you being sarcastic?

[Bart]

The unique thing about L.A. Riders is that when you sit on the machine and put your hands on the handlebars, the handlebars have a very good smell.

You'll be happy to hear that one of the top priorities for Supermodel 0.3 is OpenSMELL API support.

[SupremeTextures]

It is pretty fast that game, close in speed to burnout i think, not quiet as fast. In real life Harleys are very slow compared to other bikes.

There was another Harley-Davidson game called "Race to the Rally", and it's considered to be a rip-off of the Burnout series. The unique thing about L.A. Riders is that when you sit on the machine and put your hands on the handlebars, the handlebars have a very good smell.

You have obviously never played it! 100 seconds to 120, he is picking up speeds just as fast as burnout without the turbo boast.

http://www.youtube.com/watch?v=1GAxlXPo3uY

Are these even real people Bart or Spammers!