Accelerator board ROM

[GianDO]

I have a SuperBigBang in my A2000 and would like to emulate it with WinUAE. On the WinUAE website I saw that in the "WANTED BOOT ROOMS" menu it says "Contact me if you have accelerator or HD controller listed below." and in the list is my accelerator. How can I get the ROM file to use with WinUAE?

Thanks
G

[Toni Wilen]

Uncommon board, nice :)

Boot ROMs usually can be dumped using software only (obviously ROM reader is better option)

First step: either run WB 3.0+ tools/showconfig with debug parameter ("sys:tools/showconfig debug >ram:output.txt") or if you only have WB 1.x or something, use this: https://download.abime.net/winuae/files/b/expdump.zip and include output.

btw, I am very interested in seeing if SCSI controller is Hardital designed or if it is a clone of some other board + copied boot ROM with visible strings edited like Hardital TQM or Hardital Dotto..

EDIT: run showconfig or expdump with and without autoboot jumper. (=need floppy if you only have SCSI drive connected to this controller).

[GianDO]

Thanks for reply.

The SCSI controller on this board is a Synthesis (syndisk.device) integrated board. In attachment can you see output files for the command as you ask me in the previous message.

Thanks
G

[DamienD]

Originally Posted by GianDO

In attachment can you see output files for the command as you ask me in the previous message.

Did you forget to attach something GianDO?

[GianDO]

Originally Posted by DamienD

Did you forget to attach something GianDO?

done

[Toni Wilen]

Thanks. Looks good so far. ROM disable jumper apparently disables the ROM chip because autoconfig data is identical. (Some hardware changes autoconfig data, some simply disables ROM chip select)

ROM is at board + 0x8000 (and IO is most likely at the lower part of board). Very common hardware setup.

So dumping from address $E98000 to $E9FFFF should create full ROM image. What markings does the (EP)ROM chip have? 27Cxxx most likely?

One way to do it: Download http://aminet.net/dev/moni/mon165.lha (or any other program that can be used to save memory ranges, this is my favorite)
Run it, type "] e98000 8000 ram:file.bin" (without quotes)

Save to RAM disk because there is chance of instant crash/hang if part of dumped region is not ROM space. (Unlikely but can happen)

[cpiac64]

nice info Toni

[GianDO]

Originally Posted by Toni Wilen

Thanks. Looks good so far. ROM disable jumper apparently disables the ROM chip because autoconfig data is identical. (Some hardware changes autoconfig data, some simply disables ROM chip select)

ROM is at board + 0x8000 (and IO is most likely at the lower part of board). Very common hardware setup.

So dumping from address $E98000 to $E9FFFF should create full ROM image. What markings does the (EP)ROM chip have? 27Cxxx most likely?

One way to do it: Download http://aminet.net/dev/moni/mon165.lha (or any other program that can be used to save memory ranges, this is my favorite)
Run it, type "] e98000 8000 ram:file.bin" (without quotes)

Save to RAM disk because there is chance of instant crash/hang if part of dumped region is not ROM space. (Unlikely but can happen)

Hi,
I followed your instructions. I have attached the ROM (27C256-12) dump file. I hope I did well. Let me know ..

Thank you
G

[Toni Wilen]

Dump looks valid. It is normal that every other byte is FF, ROM is 8-bit and usually mapped to odd or even bytes only. Different looking part at the beginning is also normal (and expected), it has autoconfig nybbles and nybble based "loader" code (because KS 1.3 had a bug in expansion.library which broke byte wide mode)

Did your board have any version number labels on top of ROM chip? ROM has version string "syndisk.device v3.5r1"

[Toni Wilen]

It is a clone of ICD AdSCSI and works with AdSCSI emulation. (Hardital Dotto is a clone of ICD AdIDE)

00C01B40 FFFF FFFF FFFF 4943 4420 426F 6F74 526F  ......ICD BootRo
00C01B50 6D20 2843 2931 3939 3200 4943 4442 6F6F  m (C)1992.ICDBoo
00C01B60 742E 6465 7669 6365 0000 646F 732E 6C69  t.device..dos.li
00C01B70 6272 6172 7900 4661 7374 204D 656D 6F72  brary.Fast Memor
00C01B80 7900 4368 6970 204D 656D 6F72 7900 48E7  y.Chip Memory.H.

Again they missed (or failed to understand how boot roms are structured..) ASCII in nybble based data.

Oddly enough amiga.resource.cx says manufacturer id is 2080 but your tools/showconfig output says it is 2071 = ICD..

Even more interestingly ROM structure does not match any existing ICD ROM dumps so most likely there is later than v2.2 ROM that is not yet dumped. I don't think they (Hardital) didn't do anything else than some ASCII replacement (ICD -> Syn)..

Because ROM is 32k (27C256) and only every other byte is used, ROM should be also mapped to low 32k of board space just like AdSCSI. Could you try also dumping $e90000 to $e97fff? ("] e90000 8000 ram:file.bin"). There is chance this hangs the system but I think it did work with AdSCSI. Unfortunately all the important address spaces are overlayed by autoconfig so dump isn't that useful but it other locations contain only FF, it can be assumed to be identical to ICD version.

[GianDO]

Originally Posted by Toni Wilen

Dump looks valid. It is normal that every other byte is FF, ROM is 8-bit and usually mapped to odd or even bytes only. Different looking part at the beginning is also normal (and expected), it has autoconfig nybbles and nybble based "loader" code (because KS 1.3 had a bug in expansion.library which broke byte wide mode)

Did your board have any version number labels on top of ROM chip? ROM has version string "syndisk.device v3.5r1"

The label on the rom chip says Rev 3.0

[GianDO]

Originally Posted by Toni Wilen

It is a clone of ICD AdSCSI and works with AdSCSI emulation. (Hardital Dotto is a clone of ICD AdIDE)

00C01B40 FFFF FFFF FFFF 4943 4420 426F 6F74 526F  ......ICD BootRo
00C01B50 6D20 2843 2931 3939 3200 4943 4442 6F6F  m (C)1992.ICDBoo
00C01B60 742E 6465 7669 6365 0000 646F 732E 6C69  t.device..dos.li
00C01B70 6272 6172 7900 4661 7374 204D 656D 6F72  brary.Fast Memor
00C01B80 7900 4368 6970 204D 656D 6F72 7900 48E7  y.Chip Memory.H.

Again they missed (or failed to understand how boot roms are structured..) ASCII in nybble based data.

Oddly enough amiga.resource.cx says manufacturer id is 2080 but your tools/showconfig output says it is 2071 = ICD..

Even more interestingly ROM structure does not match any existing ICD ROM dumps so most likely there is later than v2.2 ROM that is not yet dumped. I don't think they (Hardital) didn't do anything else than some ASCII replacement (ICD -> Syn)..

Because ROM is 32k (27C256) and only every other byte is used, ROM should be also mapped to low 32k of board space just like AdSCSI. Could you try also dumping $e90000 to $e97fff? ("] e90000 8000 ram:file.bin"). There is chance this hangs the system but I think it did work with AdSCSI. Unfortunately all the important address spaces are overlayed by autoconfig so dump isn't that useful but it other locations contain only FF, it can be assumed to be identical to ICD version.

This time the mon command hangs the system.

[Toni Wilen]

Change start address to e90050, if it still hangs, try e90060 (e90040 is "pseudo"-dma io port)

[GianDO]

Originally Posted by Toni Wilen

Change start address to e90050, if it still hangs, try e90060 (e90040 is "pseudo"-dma io port)

I got to e90100 (excluding e90040) and the system hangs. do i have to continue until i find a valid address?

[Toni Wilen]

I guess whole lower part of ROM is unmapped after board has been autoconfigured (it isn't needed anymore), making it impossible to read it without ROM reader (at least without custom KS ROM but that isn't worth the trouble).

There probably isn't anything interesting anyway.

[GianDO]

Originally Posted by Toni Wilen

I guess whole lower part of ROM is unmapped after board has been autoconfigured (it isn't needed anymore), making it impossible to read it without ROM reader (at least without custom KS ROM but that isn't worth the trouble).

There probably isn't anything interesting anyway.

The dump of the last few days I did it both with autoconfig and without autoconfig, while that of tonight I did it without autoconfig. This board has no jumpers for autoconfig, but in the rear facing outwards, in addition to the db25 connector, it has 3 microswitches, the first to disable the 68030 and switch to 68000 mode (fallback switch), the second to enable / disable RAM and the third to enable / disable SCSI. the last dump you told me to do I did it with the SCSI disabled. Could the hang be due to the fact that SCSI was disabled?

[GianDO]

I also tried with SCSI enabled, but but the system still hangs.

[Toni Wilen]

Most likely "disable" only disables upper part of the ROM, leaving all the other parts of hardware enabled. This is quite normal behavior because it needs very little hardware support (=fits in existing PAL chips)

There is still one tricky method remaining that could work but it needs short custom dump software. In next few days maybe..

[GianDO]

Ok. I look forward to hearing from you ..

[Toni Wilen]

Nasty hack attached.

Requirements (and it still might not work):
- CPU must be 68000
- HD controller must be first autoconfig board (=Z2 RAM must be disabled). Tools/showconfig must have "Board + ROM (HD?)" listed as first board.
- Boot from floppy. Program takes over the system and needs manual reset. (SCSI can be autoboot disabled)

Usage:

Boot from floppy with attached program and mon included.
run resdump
Red and green stripes should be visible on screen. If something else happens: it didn't work.
Reset manually. Do not power cycle.
Save memory from $60000 to disk ("] 60000 10000 df0:file.bin")
Hopefully whole ROM data is included..

Technical info:

Program executes CPU reset (uses same trick as original Double Dragon 2 which allows CPU execution even if it was running in chip ram)
Reset resets (unconfigures) all autoconfig board. First board will appear at $e80000. There is chance whole ROM is visible when in unconfigured state because SCSI chip IO registers use same address space as autoconfig nybbles and nybbles must be visible for expansion.library to see the board.
Program then copies whole $e80000 to $e8ffff region to $60000.

EDIT: attachment updated. Now uses $e0000 to save $e80000 contents and added delay before copy.
->
"] e0000 10000 df0:file.bin"