I've forked a copy of JGH's NFS 3.40/3.50/3.60/3.65 source code and been working on an Electron version. So far I've got 3.65 going with Roland's board (and possibly the earlier versions: they assemble, but I've not tried them) and that seems to work OK. I've also used the technique from the IntegraB IBOS code that Ken suggested for relocation and sorted out the Tube host code so versions 1.00 and 1.10 of that work OK (or at least seem to!). The code is here:
https://github.com/mincebert/AcornNFS/tree/elk
I still have one major issue, though, which is the noise causes by the NMI handler in MODEs 0-3 (everything still works in those MODEs but you just get a load of screen noise). IanB said:
There are also some references in the code to the 6522 VIA but I'm not clear what should happen with those on the Electron or if the address that was in the build file is nonsense. Things seem to be working fine but I'm not clear how they're used (perhaps related to above!).
Coupled with a patch to allow slower (>8µs mark) clocks on the PiEconetBridge, with this version things seem to be much more reliable than the 3.35/3.42 version, so it would be good to get it working!
Thanks for any help!
https://github.com/mincebert/AcornNFS/tree/elk
I still have one major issue, though, which is the noise causes by the NMI handler in MODEs 0-3 (everything still works in those MODEs but you just get a load of screen noise). IanB said:
I'm not entirely clear what's needed here (I can make guesses but I'd really be fishing around because I'm not clear how this all works). Does anyone (Ian?) know more specifically what needs to be done as I'm happy to have a go, but I'd need a bit of a clue where to go next for this.It is a version of BBC NFS 3.35e patched for the Electron Econet addresses and also has some additional * commands to cope with screen noise:
When an NMI is triggered (used to service the 68B54 ADLC) the Electron prioritises the CPU over the screen so that the NMI can be serviced quickly. This results in on screen noise in modes 0 to 3 whenever there are NMI interrupts. With the disc filing system this is hidden by blanking the screen during disc access and this screen blanking can also be used in a similar way with Econet but there is an additional problem because NMIs are also generated by traffic for other machines on the network so you get screen noise whenever there is any network traffic even if it is not related to the Electron.
As mentioned in an earlier post above, to work around this the following * commands are available: ...
There are also some references in the code to the 6522 VIA but I'm not clear what should happen with those on the Electron or if the address that was in the build file is nonsense. Things seem to be working fine but I'm not clear how they're used (perhaps related to above!).
Coupled with a patch to allow slower (>8µs mark) clocks on the PiEconetBridge, with this version things seem to be much more reliable than the 3.35/3.42 version, so it would be good to get it working!
Thanks for any help!
Statistics: Posted by Mince — Thu Apr 11, 2024 11:55 pm