A2091/A590 scsi.device 14 MHz patch

[SpeedGeek]

Hello my Great Amiga Coding Wizard friends!

I'm looking for ideas and assistance to patch the A2091/A590 scsi.device to provide better support for synchronous mode with the 14 MHz mod.

http://eab.abime.net/showthread.php?t=58433

I've written a small tool to check the 33C93A Config but this does not address a possible problem with the dynamic (software) calculated sync period or the patching of scsi.device itself.

I know that most of you guys are light years beyond my feeble coding skills. I will upload a complete 16KB image of the C= V7 ROMs to the Zone. Thanks for any and all of your help!

[SpeedGeek]

More Technical info:

- The A2091 is an Auto-Config device so the board can be located at any 64KB block of Zorro2 I/O space. (The Board is Configed by expansion.library)

- The first 8KB of the 64KB block is where the Auto-config, DMAC and 33C93 registers live.

- The ROMs are offset by $2000 and currently use 16KB of address space with mirror images appearing for the remaining 40KB of address space. (So $2000 is good place to start dis-assembling the ROM binary).

- The ROM's contain the RomBoot device code to install scsi.device bootable volumes which differs from the resident type code of the IDE scsi.devices (and of course they also contain the low level code for SCSI I/O)

[breech]

This guy seems to have given it a go (and posted the rom code) - at least I think that's what Im looking at..
I cant help you with the code, but I will be doing the A590 14mhz mod will be happy to test this if you make any progress

[SpeedGeek]

** NEWS UPDATE **

I patched 4 bytes of code in the ODD ROM (U12)* and now I can boot with 33C93A registers set for 14MHz! The 33C93A_14MHz Util is NO longer needed for C= V7.0 ROMs (but a new 33C93A_Config tool is available for testing patched ROMs).

So far working good with old (flaky) Quantum 52 & 105 + 550 MB Compaq HDs. You have to set JP5-4 on + Synch enable jumper on some HDs.

Dynamic calculation of Synch period still not patched but I don't think that's really necessary now since the A2091 scsi.device was ALREADY using the same default value as the A3000 scsi.device.

A very special thanks to Don Adan for his valuable resourced disassembly of this real scsi.device!

*NOTE: The even numbered ROM actually refers to the odd byte!

[waltermixxx]

I Tried it and it seems to work like a charm.
thank you

[Don_Adan]

Originally Posted by SpeedGeek

** NEWS UPDATE **

I patched 4 bytes of code in the ODD ROM (U12)* and now I can boot with 33C93A registers set for 14MHz! The 33C93A_14MHz Util is NO longer needed for C= V7.0 ROMs (but a new 33C93A_Config tool is available for testing patched ROMs).

So far working good with old (flaky) Quantum 52 & 105 + 550 MB Compaq HDs. You have to set JP5-4 on + Synch enable jumper on some HDs.

Dynamic calculation of Synch period still not patched but I don't think that's really necessary now since the A2091 scsi.device was ALREADY using the same default value as the A3000 scsi.device.

A very special thanks to Don Adan for his valuable resourced disassembly of this real scsi.device!

*NOTE: The even numbered ROM actually refers to the odd byte!

I have latest (?) A2091 and A590 scsi.device v40+, if you need these versions for your works, send to me mail.

[SpeedGeek]

Originally Posted by Don_Adan

I have latest (?) A2091 and A590 scsi.device v40+, if you need these versions for your works, send to me mail.

Hi Don,

OK, I'll send you an email.

P.S. I never did get a resourced version of scsi.device to assemble with DevPac. There are just to many errors. Switching resource to old syntax reduced the errors but doesn't eliminate them completely.

[SpeedGeek]

OK, here's some good and bad news:

I received the V40+ versions from Don Adan. Thanks!

I tested the two A2091 BindDrivers versions (there are also two separate versions for the A590 but for no practical reason I can think of) and they seem to limit the Max. transfer rate to 1.4MB/sec.

I even patched the latest V40.4 (from the V2 directory) for 14MHz operation and tested it with Synch transfer mode enabled but I still get 1.4MB/sec Max. So, as far as I can tell they're a big downgrade from V37.64 (V7.0) which can easily manage a 3MB/sec transfer rate.

[alexh]

I wonder if it was compiled with debug or some rubbish like that?

[JanciB]

The v40+ version for A590 supports hdd bigger than 8GB?

[Dale423]

Originally Posted by SpeedGeek

OK, here's some good and bad news:

I received the V40+ versions from Don Adan. Thanks!

I tested the two A2091 BindDrivers versions (there are also two separate versions for the A590 but for no practical reason I can think of) and they seem to limit the Max. transfer rate to 1.4MB/sec.

I even patched the latest V40.4 (from the V2 directory) for 14MHz operation and tested it with Synch transfer mode enabled but I still get 1.4MB/sec Max. So, as far as I can tell they're a big downgrade from V37.64 (V7.0) which can easily manage a 3MB/sec transfer rate.

I have done the 14mhz mod on the A590. I have stock 7.0 ROMS. I am little confused as what I need to do to get past this 1.4mb barrier? I am using 3.1.4 SCSI.DEVICE. Do I need to burn a modified version of the 7.0 Roms?

[SpeedGeek]

Originally Posted by Dale423

I have done the 14mhz mod on the A590. I have stock 7.0 ROMS. I am little confused as what I need to do to get past this 1.4mb barrier? I am using 3.1.4 SCSI.DEVICE. Do I need to burn a modified version of the 7.0 Roms?

IIRC your question was answered on the Amibay A2091 14MHz thread last year. The same answer applies here. The only reasons you would use this C= v7.0 ROM patch are:

1) To enable synchronous mode, which doesn't always provide higher transfer rates than over-clocked asychronous mode.
2) To use multiple SCSI target devices with the proper SCSI timeout value.

BTW, please don't confuse the psuedo (IDE) scsi.device versions with the real scsi.device versions. Thanks.

[Dale423]

Originally Posted by SpeedGeek

IIRC your question was answered on the Amibay A2091 14MHz thread last year. The same answer applies here. The only reasons you would use this C= v7.0 ROM patch are:

1) To enable synchronous mode, which doesn't always provide higher transfer rates than over-clocked asychronous mode.
2) To use multiple SCSI target devices with the proper SCSI timeout value.

BTW, please don't confuse the psuedo (IDE) scsi.device versions with the real scsi.device versions. Thanks.

Yes realize that now in relation to the scsi.device. I do see the one in ROM being used is the V37.64. My issue is that I am only seeing 1.4mb/sec with the 590 and 14mhz mod on a stock 500 with 3.1.4 and 3.2 kickstart. The 590 does have 2mb of RAM. I did burn the modified U12 7.0 ROM but saw no change. Just wondering if I am missing something.

[levo-ace]

I have just upgraded my A590 to 20Mhz ( https://www.levosretrocomputerprojec...a590-20mhz-mod ) and would like to know if you still have the details of what you changed in the rom!

I have done a comparison between the two and only 4 locations have changed.
00157 , 01EF8 and 01F2E from 07 to 47
0016E from 17 to 2E

I assume these are values written to the own id and the synchronus transfer register .
Is this right , and do all three 07 to 47 changes write to the own id register?

I would like to change some values and correct the timing for the 20Mhz speed , although it seems stable now , unless i use an external drive.

You can use an eeprom instead of an eprom . I am using a winbond 27e257 but you need to remember that A14 ist tied high and A13 does change so you need to load double copies of the eprom data in the 2 top locations (one at 4000 and one at 6000 ). and it is pin compatible , just pop it in the socket!

[SpeedGeek]

Originally Posted by levo-ace

I have just upgraded my A590 to 20Mhz ( https://www.levosretrocomputerprojec...a590-20mhz-mod ) and would like to know if you still have the details of what you changed in the rom!

I have done a comparison between the two and only 4 locations have changed.
00157 , 01EF8 and 01F2E from 07 to 47
0016E from 17 to 2E

I assume these are values written to the own id and the synchronus transfer register .
Is this right , and do all three 07 to 47 changes write to the own id register?

I would like to change some values and correct the timing for the 20Mhz speed , although it seems stable now , unless i use an external drive.

You can use an eeprom instead of an eprom . I am using a winbond 27e257 but you need to remember that A14 ist tied high and A13 does change so you need to load double copies of the eprom data in the 2 top locations (one at 4000 and one at 6000 ). and it is pin compatible , just pop it in the socket!

First, you should read the 33C93A datasheet which explains that the input CLK is subject to an internal CLK divisor with the final result being approx. 5 MHz. So 10/2 or 15/3 or 20/4 = 5 MHz.

The datasheet also explains the SCSI timeout and synchronous offset calculations. So you have nothing to gain with the 20 MHz clock speed except a headache trying to figure out how to modify the SCSI driver code for these new timing parameters. I must apologize for the sarcastic reply (but it helps in clarifying the point).