The Sinclair QL Forum

Hi Derek,

The boot sequence I see is similar to what you describe but it doesn’t “copy” the ROM to RAM in C0000 it “pages” in a Rom which when called at startup simply jumps to the real ROM at 10000

As the Rom is paged (this has to be the case with 512+128 as there is no RAM at C0000) rather than copied this is why I don’t understand why a reset breaks the MT3 under specific circumstances

Thanks
Tony

Actually I've just realised that with a total of 640K (512K+128K) in the QL, setting ram top to 640K (so no memory size change) also knocks out the detection of the MT3 card, whereas after pressing reset button the MT3 is detected

So whatever the problem I'm having is (with JS) must be related to the difference between a "reset 640" and just pressing the reset button

I think I need to learn more about the physical reset and TC PAL to see if a workaround is possible

Have also now realised that by using RESET command without a size the MT3 IS detected (even though it's still 640K)

Will move my testing to the RESET code to see if the problem (or at least a clue to the cause) lies in there!

Mind you, given "res_128" also knocks out detection of the MT3 that probably rules out me finding a solution in the RESET code

(res_128 issue doesn't relate to the change made in v2.18 as I'm using older versions to avoid that change)

I've got the memory of a Goldfish!

Of course the reset code has a code section that attempts to copy the TC ROM from 10000 to C0000 before the reset occurs

I think my only chance now is to get my hands on an original Trump 512K card and hope the PAL is readable

Or learn how to edit the logic in the existing PAL

I'm thinking I actually need a Trump Card 0K PAL1. maybe

Are there any guides on how I could recompile the code in this thread viewtopic.php?t=3880&sid=aaf4db8a3d799d ... 8b4dad31c3 into a JED file (on MacOS or Windows) I could burn please?

Many thanks
Tony

Thanks to RalfR helping me get MAGEPROM running I thought I'd try running the TC ROM (after a reset 128/256/384/512) using it and it works!

I've no idea yet if the extra memory it uses will prevent any games that need reset nnn (128 in particular) from running, but may be able to make the ROM image much smaller as it probably only needs the first bit of code that jumps to the real rom at $10000

Of the games that need flp AND the memory cut there is only 4 now

A fair bit of testing to do but it's something to work with I think

Turns out there's no need to load the TC ROM into memory, just "call 65626" is sufficient to re-enable the MT3

So a simple program (to run after memory reset to 128/256) is

CALL 65626:FLP_USE FLP:LRUN FLP1_BOOT

This works for 3 of the 4 problem games BUT ONLY if QIMSI SD card is removed during startup so it doesn't take any memory

That means I can't automate the process from QIMSI games menu. I'll probably try having the code in a boot file on a vDrive image so on F1/F2 it boots the code from vDrive MDV1. That's still to be tested!

The only game that won't run is Steve Davis Snooker, and that one is really weird.

Even though DIR FLP1_ works fine, LRUN (or LOAD) FLP1_BOOT says NOT FOUND!

No idea what's going on on here. Will try creating a new FLP image (but it works on all other physical memory/card combinations)

A new floppy image for Snooker has fixed that weird issue

You have to be careful sometimes. It's quite possible that one machine code program is reloading another, and in that case, the absolute file path is already included in the code, e.g., "mdv1_something_cde." Unfortunately, a new image won't help in that case.