| |
Fungus
Registered: Sep 2002
Posts: 617 |
Reliable REU detection
I've been struggling with this for days. I can detect any REU size except 256K.
From what I understand, I should be able to write data to bank 4, and read back and get crap instead, but I'm getting valid data back.
I don't know if this is a problem with the real thing or a vice issue.
Doesn't anyone have a SIMPLE way to check this, most routines I've found are overly complex for such things. Either the memory is mirrored, or its open bus. It can't be both. |
|
| |
tlr
Registered: Sep 2003
Posts: 1714 |
Quoting FungusDoesn't anyone have a SIMPLE way to check this, most routines I've found are overly complex for such things. Either the memory is mirrored, or its open bus. It can't be both.
It has to be either of those for a given REU implementation yes. You don't know which though. For a stock 256KiB REU it is supposedly open bus for the upper 256KiB of an 512KiB address space.
If confused, I would recommend scanning whatever REU implementation you want to target to see which areas are mirrors and which are unmapped. This way you can decide on how to proceed. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
And... there is no SIMPLE way :) That's why so many detection routines "out there" are broken (even the one on codebase is problematic, unfortunately)
Also, when testing, make sure to really test on actual REUs. Not some reimplementation like that on 1541U - those are all broken one way or another. (recent VICE should be somewhat correct, except for some very obscure timing quirks that don't matter for this) |
| |
Fungus
Registered: Sep 2002
Posts: 617 |
Yeah, I've decided it's just too much bother to support something that is effectively broken.
Good ole commodore engineers. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
Its not "broken" at all
edit, for the records - MacBacon provided a working one (and i replaced the buggy one on codebase) https://codebase64.org/doku.php?id=base:reu_detect |
| |
Trash
Registered: Jan 2002
Posts: 122 |
If all banks are mapped but not there, write a single byte (starting with 0) to address 0 in each bank, count that value up by one and then read back address 0 from the first bank, it should contain 0 on a 16mb reu, 128 on a 8mb reu and so on. |
| |
Mr. SID
Registered: Jan 2003
Posts: 421 |
Quote: Its not "broken" at all
edit, for the records - MacBacon provided a working one (and i replaced the buggy one on codebase) https://codebase64.org/doku.php?id=base:reu_detect
Yes, this seems very similar to the method I use in Sonic, which I've tested on everything and the kitchen sink (except TC64). |
| |
Fungus
Registered: Sep 2002
Posts: 617 |
Thanks I've had a look at it. Although as I said if a bank has a floating bus, I should be able to detect that as is. I'm not sure what the error is, but I'll try again later.
Trash: That's what doesn't work, that was the first thing I did. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11116 |
The nice thing with this method is that it works right also on broken implementations (which mirror bank 4-7 in 256k REU) |
| |
Frantic
Registered: Mar 2003
Posts: 1627 |
Quote: Its not "broken" at all
edit, for the records - MacBacon provided a working one (and i replaced the buggy one on codebase) https://codebase64.org/doku.php?id=base:reu_detect
Thanks! |
